Help

What is Belka Food

Belka Food is a recipe builder and social network for anyone who cooks professionally or simply cares about precision in the kitchen. The site recalculates a recipe for any number of servings or the weight of any ingredient, automatically works out calories/protein/fat/carbs and cost, tracks allergens, and keeps recipes in three languages — Russian, Norwegian and English. It suits chefs and catering businesses who need accurate cost breakdowns and production sheets; food bloggers who publish recipes and grow an audience; and everyday users who just want to cook from reliable recipes that automatically rescale to the number of servings they need. You can browse recipes without registering. An account is only needed to publish your own recipes, save favorites, follow authors and use the recipe builder.

Open as a separate page →

Free vs Pro — what is the difference

The core features are free forever: creating and publishing recipes, automatic serving-size recalculation, nutrition facts, basic cost calculation, the meal planner, following authors, favorites. Pro unlocks professional tools: gross/net weight and waste % on every ingredient line, recipe components (reusable sub-recipes shared across multiple recipes), food cost % and sale price, exporting a production sheet to PDF and Excel, your own purchase prices for ingredients, event planning (banquets/catering), importing a recipe from a link on another website, one-click translation of recipes into the site's other languages, and access to the builder's AI assistant (build/edit a recipe by voice or text, recognize one from a photo — see the "AI assistant" section of this help center). Details and sign-up are on the Pricing page.

Open as a separate page →

Feed, search and filters

The home page is the recipe feed. At the top are category tiles (click to show only that category) and the search bar: it searches recipe titles and descriptions, ingredients and the author or channel name. Filters above the feed: cuisine, cooking time, calories per serving, diets (vegan, gluten-free, lactose-free, healthy), “Without ingredient” (e.g. no onion) and “Without allergens” — hides recipes with the chosen allergens, and also recipes with no allergen data yet (to be safe). Filters can be combined; the “N recipes found” counter shows the result. Sorting: newest, popular, quick, cheap, top rated. The site language (RU/NO/EN) is switched in the header — each language has its own recipe feed.

Open as a separate page →

Site assistant and support

At the bottom right of every page is the 💬 button. It opens the AI assistant: ask in any language how to do something on the site (create a recipe, work out the cost, get Pro, withdraw money) — it answers from this help and links to the articles. If the assistant does not know the answer or the question needs our team (a bug, payment, account), a “Write to support” button appears. Write your message — we receive your conversation with the assistant along with it, so we do not have to ask again. The answer comes by email. You can write to support without registering by giving your name and email, and also from the “Support” page. The assistant is free, with a daily limit on questions. We analyze questions to the assistant to improve the site and the help.

Open as a separate page →

Content Factory — videos for YouTube and TikTok

The Content Factory is a separate Belka Food service that makes videos automatically: it writes a script, voices it, picks footage or has the AI draw it, edits with subtitles and music and publishes to YouTube, TikTok and Telegram. It has studios with a presenter, AI bloggers, cartoons and series with consistent characters, and a scheduled autopilot. The service is in beta: first submit an application on the access page; after approval, get site Pro (opens the free Content Factory tier) or any Content Factory tier. Details are in the site menu (“Content Factory”) and in the help inside the panel.

Open as a separate page →

How to create a recipe

Click "+ Recipe" in the site header. Then fill in the title, description, category and cuisine, cooking time, number of servings (or pieces — cookies and patties are easier to count individually), and a photo. Ingredients are added one at a time: start typing a name and autocomplete kicks in, searching the shared catalog and your own personal ingredients. The recipe is saved as a draft automatically the first time you add a photo or save — only you can see it until you click "Publish". Nutrition facts and cost are calculated live as you enter the ingredients, no separate "calculate" step needed.

Open as a separate page →

Components — why use them and how

A component is simply one of your own recipes, embedded inside another recipe as a single "ingredient". A typical example: you have "Garlic Sauce" as a separate recipe, and you use 60 g of that sauce inside "Shawarma". Instead of retyping the sauce's ingredients into every recipe that uses it, you cost the sauce once and then just reference it by weight. Why this matters: — Cost cascades automatically: if the price of garlic in the sauce changes, both the sauce and every recipe using it recalculate — no manual edits needed anywhere. — No need to duplicate a long ingredient list (dough, broth, marinade) in every recipe that uses it. — On the recipe page, the component can be expanded — a visitor sees exactly what it's made of without leaving the page. How to add one: in the builder, click "+ Recipe component" (visible when Pro mode is on). Start typing a name — the site searches only YOUR OWN recipes (you can't turn someone else's recipe into a component — it's your personal reusable sub-recipe). If no matching recipe exists yet, a "+ Add ingredient" option appears right there, letting you build the component's ingredient list on the spot without leaving the page; on save it becomes a separate draft in "My Recipes" (the "Components" tab). Components inside components (two levels of nesting) aren't supported yet — only one level.

Open as a separate page →

Cost calculation and your own ingredient prices

A recipe's cost is calculated from ingredient prices: the sum of (each ingredient's weight × price per kg/l/piece) plus the cost of any components. There are two price sources: Base price — shared across all users, set for ingredients in the main catalog (not every ingredient has one — if there's no price, that ingredient simply doesn't contribute to the total, and the cost is honestly marked as incomplete). Your own price — you set it yourself in the ingredient catalog ("Ingredients" in the profile menu), or right in the builder via the "price?" button next to a priceless line. Your own price overrides the base price and instantly applies to ALL your recipes that use this ingredient — enter your supplier's price once, and every recipe's cost recalculates. Currency is personal per user (set in your profile; the available currencies depend on the site language you use). You can enter a price in whatever unit is convenient — kr/kg, kr/l or kr/piece — the site converts it automatically for calculations. Cost is visible to every visitor on the public recipe page — that part is open. Food cost % (cost as a share of sale price) and the sale price itself are private, visible only to you in the builder.

Open as a separate page →

Gross weight, net weight and waste %

Net weight is the clean amount of an ingredient that actually goes into the dish and is used for nutrition calculations. Gross weight is the amount before peeling/trimming/cooking losses (e.g. potatoes before peeling). Cost is calculated from the gross weight — that's what you actually buy, not what's left after prep. On an ingredient line (in Pro mode) there's a "waste %" toggle next to the gross weight field — they're linked: enter a waste percentage and the gross weight recalculates automatically, or enter the gross weight directly and the percentage works itself out. The net weight used for nutrition facts doesn't change from this field — only the cost does.

Open as a separate page →

"To taste" — ingredients without a quantity

Not every ingredient has an exact amount — salt, spices, garnish herbs are often added "to taste". For such lines there's a toggle next to the unit field: switch on "to taste?" and the quantity field disappears. This ingredient doesn't count toward the recipe's weight, nutrition facts or cost, but it still shows up in the allergen list (if it's an allergen) and in the shopping list — just without a specified amount.

Open as a separate page →

Servings and pieces

A recipe can be measured two ways. "Servings" is the usual mode (soup for 4, cake for 8). "Pieces" is for dishes that are naturally counted individually: cookies, patties, dumplings. In this mode you specify the number of pieces and the weight of one piece — on the public recipe page a visitor can rescale the recipe to the number of pieces they need, the same way servings are usually rescaled. You can switch modes in the "Portioning" block when creating a recipe.

Open as a separate page →

Ingredient catalog and “My products”

The “Ingredients” section is a reference of over 9,000 products with nutrition per 100 g, split into categories, searchable in three languages. An account is required. Your own purchase price (Pro): click the price of any product and enter yours — per kg, liter or piece, in your currency. It immediately applies to all your recipes and replaces the base price. You can also set a price right in the builder (“price?” on a row) or on the recipe page. The “My products” tab shows products with your prices and personal ingredients (created when a product was not in the reference). The “Recipes” button on a product shows which of your recipes use it; a personal ingredient can only be deleted when it is not used anywhere. There is also a “price per 1 kg” calculator: enter the package weight and price, and with the “Cooked yield” button — gross and net weight, to get the price per kilogram of the finished product.

Open as a separate page →

Recipe photos and video, step photos

You can add up to 6 photos to a recipe: click the large photo or “+” in the builder. The first photo is the cover; to make another one the cover, drag its thumbnail onto the large photo. If the recipe is not saved yet, it is saved as a draft automatically — it only needs a title. Recipe video — one, up to 60 seconds, MP4 or WEBM, up to 40 MB. It is shown instead of the photo on the recipe page and in the feed card. Each cooking step can have its own photo — the camera button next to the step. Step photos are visible on the recipe page and in cooking mode.

Open as a separate page →

AI assistant — build and edit a recipe by conversation

The recipe builder has an "AI assistant" button — it opens a chat panel next to the recipe form. You can type or speak (a microphone button sits next to the input field): dictate the dish's name, the ingredients with quantities (or say "to taste" when there's no exact amount), and the cooking steps — the assistant works out what you said and fills in the recipe form right in front of you: title, category, cuisine, cooking time, ingredient list, steps. You can dictate in parts, over several messages — the assistant remembers what you've already told it in this conversation. If you open the assistant on a recipe you've already saved, it sees its full content and can give you professional advice — what to add for a better flavor balance, how to change the cooking technique, what to substitute for an ingredient. You can attach a photo — for example, take a picture of something you have on hand and ask whether it can replace something in the recipe; the assistant looks at the photo and answers directly, updating the ingredient list itself if needed. Once the recipe is complete (it has a title, ingredients and at least one step), the assistant asks — once per conversation — whether you'd like it translated into the site's other languages; say yes, and it gets translated and published automatically (see "Translating a recipe into other languages" in this same section). Available on the Pro plan. Every message spends from your AI balance — see "AI balance: topping up and what it's spent on" for details.

Open as a separate page →

AI balance: topping up and what it's spent on

The builder's AI assistant (chat, photo recognition) is a paid feature on top of the Pro subscription — it has its own prepaid balance in kroner, separate from the subscription payment. Top it up on the "AI balance" page (linked from the profile dropdown menu): there are a few preset amounts or you can enter your own (from 20 kr), paid by regular card via Stripe, credited instantly. It's charged based on actual usage — the real cost of the tokens the AI model processed plus a small site markup, not a flat "per attempt" price. You're charged for any response the AI actually returns — including when a photo turns out not to contain a recipe, or the assistant couldn't extract a single usable ingredient: the AI still did real work and spent real tokens on your request. Only genuine technical failures are free — when no response comes back at all (network error, service unavailable). A voice message is first transcribed to text for free, and from there it's billed the same as a regular typed message — speaking instead of typing doesn't add extra cost. A single voice recording is capped at three minutes. Translating a recipe into other languages (via the button or through the assistant), on the other hand, never touches the AI balance at all — that's a separate free mechanism. To protect it from abuse, voice messages and translations are both rate-limited to a few dozen per hour — normal builder use never comes close to that limit. The "AI balance" page shows a full transaction history (top-ups and charges, each with a description of what was paid for), with pagination and PDF/Excel export.

Open as a separate page →

Recognizing a recipe from a photo

The "Recognize from photo" button in the builder is visible before the recipe is saved for the first time. Upload 1 to 3 photos — a screenshot of a recipe from another site, a photo of a page from a cookbook, a photo of a handwritten recipe — and the AI extracts the title, ingredients with quantities, cooking steps and category, creating a draft that just needs a quick check before publishing. The AI balance is charged for the act of processing the photo itself — regardless of the outcome. If the photo turns out not to contain a recipe, or the AI couldn't make out a single usable ingredient, you're still charged: the request genuinely went to the AI model and spent tokens, the result just wasn't useful. Only a genuine technical failure — when no response comes back from the AI at all — is free. Available on the Pro plan, charges the AI balance based on actual usage (see "AI balance: topping up and what it's spent on" for details).

Open as a separate page →

Translating a recipe into other languages

Belka Food runs in three languages — Russian, English and Norwegian. A recipe is always created in one language, but it can be translated and published in the other two — in two ways. A dedicated "Translate" button — on the page of a recipe you've already saved in the builder, next to "AI assistant". It opens a checklist of the two missing languages: tick one or both and click "Translate" — the recipe gets translated and published in the chosen languages right away, and the original itself gets published too if it was still a draft. Languages already translated are marked in the list and link to the finished version. Via the AI assistant — once you've dictated or typed a recipe in the chat and it's fully complete, the assistant asks whether to translate it into the site's other languages; say yes and you get the same result as with the button. Translating doesn't touch your AI balance — it's a free part of the Pro plan (limited to 10 translations per hour to protect the mechanism from abuse — normal use never comes close). After translating, if you switch the site's language version (RU/EN/NO), the matching translated copy of your recipe shows up in "My Recipes" on that version.

Open as a separate page →

How nutrition facts are calculated

Nutrition facts are calculated from each ingredient's net weight (using catalog values per 100 g) and shown per serving and per 100 g of the finished dish. Components are handled recursively — the sauce's nutrition inside a shawarma is calculated from the sauce's own ingredients, not entered separately. On the public recipe page there's a "reference" feature — click any ingredient's amount and enter how much of it you actually have, and the whole recipe (including nutrition and cost) rescales proportionally.

Open as a separate page →

Allergens

The 14 official EU allergens (the same list applies in Norway) — gluten, crustaceans, eggs, fish, peanuts, soy, milk, nuts, celery, mustard, sesame, sulphites, lupin, molluscs. They're detected automatically from the recipe's ingredients and shown as a banner on the recipe page. If even one ingredient in the recipe hasn't had its allergens classified yet, the banner honestly warns that the list may be incomplete — we don't label a recipe "safe" when we're genuinely not sure.

Open as a separate page →

Eating for a goal: calorie/macro target and a weekly plan

The “Eating for a goal” section (Pro) calculates your daily target for calories, protein, fat and carbs and helps you eat for your goal: lose, gain or maintain weight. 1. “Profile and goal”: sex, date of birth, height, current and target weight, activity, pace and number of meals. You can take a diet into account and exclude allergens and specific ingredients (for example, peanuts or shrimp). 2. The target uses the Mifflin-St Jeor formula multiplied by an activity factor; to lose weight a safe deficit is subtracted (no more than 1 kg a week), to gain — a surplus is added. The target never goes below 1200 kcal for women and 1500 kcal for men. Details are in the “How is the target calculated?” block on the page. 3. Then two ways: “Choose dishes myself” — in the “Meal plan” each day shows “actual / target”, and a portion size is suggested when you add a dish; or “Build a plan for the week” — the site picks dishes from the catalog for the empty days of the week to fit your target. It never touches what you planned by hand. 4. Log your weight — the chart shows the trend and how much you have lost or gained. This is not medical advice: if you have a condition, are pregnant or follow a special diet, consult a doctor.

Open as a separate page →

What's included in Pro

Pro unlocks: gross/net weight and waste % on ingredient lines, recipe components, food cost % and sale price, exporting a production sheet to PDF/Excel, your own purchase prices for ingredients, event planning (banquets and catering) with a combined shopping list sized to the guest count, importing a recipe from a link on another website, and one-click translation of recipes into the site's other languages. Pro also unlocks the builder's AI assistant — you can dictate a recipe by voice or text, recognize one from a photo, and get advice on a recipe you've already built (see the "AI assistant" section of this help center for details). The assistant itself and photo recognition spend from a separate prepaid AI balance on top of the subscription; translating recipes into other languages, like every other Pro tool above, is free as part of the subscription. Everything else — the builder, publishing recipes, basic cost calculation, the meal planner, nutrition facts, following authors — is free forever.

Open as a separate page →

How to subscribe and cancel

On the Pricing page, click "Subscribe" and complete a regular secure card payment via Stripe. The subscription renews automatically every month until you cancel it. You can cancel on the same Pricing page — the "Manage subscription" button opens the Stripe portal, where you can change your card or cancel renewal. Cancelling keeps your Pro access until the end of the period you've already paid for — you're not cut off instantly.

Open as a separate page →

Author channel — your own post feed

With Pro, you get your own channel — a feed of short posts, separate from recipes. A post can have text plus either up to 6 photos or one video up to 60 seconds long (not both at once). Readers see the channel on your profile, can react and leave comments under posts, and your followers get notified about new posts. Go to your profile → "Channel" → "Create channel", then use the "Write a post" button. You can set a channel name and description in settings (the gear icon above the feed) — otherwise a default name is used. The channel is tied to an active Pro subscription: if it lapses, the channel stops being visible to readers, but all posts are kept and become available again as soon as you renew Pro.

Open as a separate page →

How the referral program works

If you have enough followers and published recipes (or you've been granted access manually), a personal referral link appears in the "Finance" section. Someone who registers through it pays you a commission every month for as long as they themselves pay for Pro — but not permanently: it's their current choice of "favorite author", not a one-time binding. Any user can change their favorite author at any time — on their own Pricing page, or on the profile page of any author in the program. The link only sets the STARTING choice at registration: if the person later switches to a different author, commission goes to the new one from that point on. Anything already accrued before the switch is not clawed back. The "Finance" page shows the whole funnel: link clicks, sign-ups, how many people currently actually support you (this can be lower than the sign-up count, and that's normal), your current balance and the full earnings history.

Open as a separate page →

Getting paid — payouts via Stripe Connect

Accrued commission isn't transferred to your card automatically — you need to connect payouts once first. In the "Finance" section, choose your country and click "Connect payouts" — a secure Stripe form opens where you confirm your identity (a document, sometimes a selfie for matching) and provide a bank account to receive money. This identity check is handled entirely by Stripe itself — the Platform never sees or stores your documents or account number, only a "connected / not connected" flag. Stripe Connect doesn't work with every country — if yours isn't in the list when connecting, please contact support. Once payouts are connected, accrued commission is transferred automatically, with no further action on your part — usually within an hour of being earned. The minimum amount for an actual transfer to your bank account is set by Stripe itself (around 20 kr for Norwegian accounts) — if the amount is smaller, it simply waits for the next payment to add up.

Open as a separate page →

Meal planner

The "Plan" section is a weekly grid with 4 meals a day (breakfast/lunch/dinner/snack). You can add a recipe to any cell (your own, a favorite, or found via search) and set the number of servings. The site calculates nutrition and cost per day and per week, and builds one combined shopping list from all the chosen recipes — merging matching ingredients and grouping them by category.

Open as a separate page →

Events

The "Events" section (Pro) is for banquets and catering. Create an event, set the guest count and price per guest, add dishes from your recipes — the site calculates the total cost, food cost % and a single combined shopping list for the whole event, exportable to PDF/Excel for the kitchen. There's an "Add to event" button on every recipe page — you can add dishes while browsing the feed, without opening the section itself.

Open as a separate page →

Shopping cart

The "Shopping cart" section (Pro) is a shopping list you build yourself before heading to the store, independent of recipes or the meal plan. Add items from the shared ingredient catalog (with autocomplete, an expected price and nutrition for each) or type your own — with any name and, if needed, a manually entered price. You can keep several carts at once — for different stores or occasions, for example. Inside a cart, tick off items as you buy them; the total sum and nutrition update live. There's a "Copy list" button (copies only what's still unbought — handy to paste into a messaging app) plus PDF/Excel export for printing.

Open as a separate page →

My purchases — history and calendar

Once you're done shopping, tap "Complete purchase" in the cart — any items you didn't tick off are deleted from the list for good at that point (this is deliberate: the history keeps only what was actually bought, not the whole draft list). The cart itself then turns into a receipt and moves to the "My purchases" section. There you'll find a calendar: days with purchases are marked with the amount spent. Clicking a day expands every receipt from that day — with its items, sum and nutrition. Past receipts can't be edited — they're a fixed record of your spending, not another draft.

Open as a separate page →

Ways to sign in

You can register with email and password, via Google, or via Telegram (a login widget button). If you already registered by email and later sign in with Google using the same email address, the site links Google to that same account instead of creating a second one.

Open as a separate page →

Profile and currency

In your profile you can set your name, photo, a short bio, and social media links (YouTube, TikTok, Instagram, Facebook, Snapchat) — they appear as icons on your public author page. The same page lets you choose the currency used for ingredient prices and cost (the available currencies depend on the site language you're using).

Open as a separate page →

Ratings, comments, “Add to my recipes” and reports

On someone else's recipe page: - “Your rating” — 1 to 5 stars; clicking again changes it. You cannot rate your own recipe. The average rating is shown in the recipe header and in the feed, and the “Top rated” sort lifts the highest-rated ones. - Comments — you can reply to other comments (replies are nested), like or dislike, and delete your own. - “Add to my recipes” — copies the recipe to your drafts with all ingredients and photos; change it however you like, the original stays as is. - The “Report” flag — spam, a recipe error, a bad photo or other, with a comment. A moderator sees the report. - “Share” — a link to WhatsApp, Telegram, Facebook, X, by email or copy the link.

Open as a separate page →

Following, favorites, notifications and social links in your profile

Following. On an author's page click “Follow” — their new recipes arrive in your notifications (the bell in the header). New followers and replies to your comments are notified there too. Favorites — the heart on a recipe card or page. All saved recipes are in the “Favorites” section of the profile menu. Social links. In “My profile” you can add links to YouTube, TikTok, Instagram, Facebook and Snapchat — your public page shows icons only for the ones you fill in. “Share to TikTok”. Connect your TikTok in your profile — on a recipe page you can then post the recipe's video or photo to your TikTok: the caption is editable before posting, you choose visibility, and for promoting your own or someone else's brand you mark commercial content.

Open as a separate page →

Installing the site as an app

Belka Food is a progressive web app (PWA) — you can install it on your phone like a regular app, without the App Store or Google Play. Open the site in your phone's browser → browser menu → "Add to Home screen" (Android) or "Add to Home Screen" (iPhone, via the Share button in Safari). An icon appears on your home screen, and the site opens in full-screen mode without the browser's address bar.

Open as a separate page →

Shopping list and cooking mode

On the recipe page, the "Shopping list" button shows the ingredients already rescaled to the current serving size; checked-off items are saved in your browser and survive a page reload. "Cooking mode" opens a large, full-screen step-by-step view — handy in the kitchen, and the screen stays on while cooking (where the browser supports it).

Open as a separate page →

Your first AI video in a couple of minutes: the beginner wizard

Make your first video with no camera and no editing: a topic in, a finished video out. In...

A tour of every section of Content Factory

A quick tour of every section of Content Factory, in the order of the menu. In this lesso...

One person, ten channels: 3 ways to earn with AI video

One person can run ten channels — no camera, no editing, no team. Three ways to earn with...

How to choose a YouTube niche and launch a channel

Find a profitable niche and get a ready channel kit in a minute. In this lesson: • «YouTu...

Video examples: repeat any format in one click

How to use «Video examples» in Content Factory. In this lesson: • finished videos in diff...

Trend ideas: fresh long-video ideas from global trends

How «Trend ideas» works in Content Factory. In this lesson: • popular videos from several...

What is the Content Factory

The Content Factory is a video factory: you give it a topic or a finished script, and the system voices the text, picks matching footage, edits a video with subtitles and music and, if you wish, publishes it to YouTube, TikTok and Telegram. How it works, step by step: 1. Script — the AI writes it from a topic or you paste your own text. 2. Voiceover — an ElevenLabs voice reads the script. 3. Footage — for each sentence a clip is picked from the free Pexels stock (or an AI image, or a talking avatar — depending on the project mode). 4. Editing — stitching, subtitles, background music, and ad inserts if needed. 5. Publishing — right away or on a schedule, to your connected channels. The sections on the left are grouped by purpose. Create: New video (a one-off clip), Quick clip (a short video with one character from a photo), From my videos (editing your own footage), AI blogger. My work: My projects, My videos, My films, Series, My music. Publishing: Social networks, Auto-publishing, Ads (a library of ad campaigns). Ideas and analytics: YouTube niches, Trend ideas, Video examples, Statistics. Account: My finances, My limits. Help: this Help and Support. The 💬 button at the bottom right opens the AI assistant for this help and the chat with support.

Open as a separate page →

Quick start: your first video in 5 steps

1. Open “New video” (a one-off clip without a project) — the fastest way. 2. Pick a method: “Topic” (the AI writes the script) or “Own script” (paste text, end every sentence with a period). 3. Choose a voice and language. You can preview voices with ▶ and mark favorites with ★. 4. Optionally choose music and adjust subtitles and voice speed. 5. Click “Create video” → check the script → “Generate”. The finished clip appears on the video page, where you can download or publish it. If you plan many clips on one theme, create a Project: voice, language, orientation and style are remembered, and every new video is created in one click.

Open as a separate page →

The “Video examples” section

The “Video examples” section collects ready videos of all formats the Content Factory can make: stock videos, avatar, avatar on top or bottom, shorts and others. Each example has a description — what the video is, where and how it is best used — and a list of parameters (video type, orientation, language, subtitles, voice speed). The “Repeat” button opens the “New video” form with the same settings — you only enter your own topic or script. The example author's personal data (their cloned voice, personal avatar, ad media) is not passed on — choose your own voice and avatar. Only an administrator can add, edit and delete examples.

Open as a separate page →

Assistant and support chat

In the bottom-right corner of every panel page there is a 💬 button. It opens a help window with two tabs. 🤖 Assistant — an AI helper for the Content Factory. Ask in your own words (“how do I record my voice”, “why is there no sound”, “how do I connect YouTube”) and it answers step by step based on this help and gives links to the relevant articles. It is free, but has a daily question limit. If it is unsure or the question needs our team, it will suggest writing to support. 💬 Support — a live chat with our team. Write your question: the message reaches us and the reply appears in the same window (if you talked to the assistant first, that dialogue is attached automatically so you don't have to explain everything again). A red counter appears on the “Support” menu item and on the 💬 button when there is an unread reply; if Telegram is linked, a copy of the reply is sent there too. Tip: describe what you did, what you expected and what happened — we will help faster.

Open as a separate page →

YouTube niches: what to make and where the money is

“YouTube niches” is a paid Content Factory option (connected separately or included in the top plan). It shows which directions are growing right now and turns a chosen niche into a ready project. What is inside: • YouTube niches — a ranking of directions (category + language) scored 0–100: demand, money, opportunity, competition, stability, feasibility for us. The “little data” label means the score is based on few channels — treat it as a guide. Revenue and RPM are model estimates from public data, not a promise. • Niche card — what the score is made of, top channels and videos (links lead to YouTube), score dynamics, “What works in this niche” (title formulas, structure, tone — in English, it is an instruction for the AI). The “Start a project in this niche” button creates a project with the language, the direction description and the script length; nothing is rendered until you start it. • Analyze a channel by link — paste a YouTube channel link to get subscribers, views, upload frequency and recent videos. If the channel is in Nexlev data you also see monetization and revenue; otherwise a rough own estimate is shown. Channels can be added to favorites ☆. • Ideas from trends — fresh popular videos from different countries are turned into trends and 10 original ideas in the chosen language. “Create a video from the idea” opens the video form with the topic filled in; “More ideas” does not repeat earlier ones. Important: we never copy other channels or videos — we take the direction, format and techniques, and scripts are written from scratch.

Open as a separate page →

How to get access: application, approval, payment

The Content Factory is in beta, so access opens in three steps. 1. Application — free. On the “Access to the Content Factory” page, tell us what you want to do, your niche, how many videos you plan and whether you run channels. Click “Request access”. 2. Approval. An administrator reviews the application and sends you a notification. While it is under review, the panel is unavailable. 3. Payment. After approval, connect one of the options: - Belka Food site Pro — it opens the free Content Factory tier (Free: stock mode, watermark, no autopilot); - or any Content Factory tier (Starter, Pro, Business) — then site Pro is not needed. If your subscription ends, projects and videos are kept, and access returns as soon as you pay. Tiers and prices are on the “Content Factory pricing” page, linked from the site menu and the access page. The site Pro trial (15 days after registration) does not open the Content Factory — a paid subscription is required.

Open as a separate page →

Your first video in a couple of minutes: the beginner wizard

When you open the Content Factory for the first time and have no projects or videos yet, you get the first-video wizard instead of an empty dashboard. What to do: 1. Write what the video is about — or click one of the ready-made topics, it takes one click. 2. Choose the language, a voice (male or female) and the format: horizontal for regular YouTube or vertical for Shorts and TikTok. 3. Click “Write script”. The AI writes a text of about a minute — you can read and edit it. 4. Start the video. The factory does the rest: it voices the script, picks stock footage for every sentence and adds subtitles and music. In a few minutes the finished video appears on the video page — you can download it or publish it to your connected channels. The wizard is only for your first try. To set everything up in detail — your voice, style, length, avatar — click “Skip to projects” and create videos via “New video” or projects. The wizard also has a link to the video lessons.

Open as a separate page →

Projects, folders and «All works»: keep every channel organised

How to organise your work in Content Factory. In this lesson: • what a project is: one vi...

A news show on autopilot: hosts, news, weather and sponsors

How to run a news episode with permanent hosts in Content Factory. In this lesson: • the...

My projects: what they are and why

A project is a “channel” or “column” with permanent settings. You set once: name, language, voice, orientation (landscape 16:9 for YouTube or vertical 9:16 for Shorts/TikTok), music volume, voice speed, subtitles, script style (a description of the niche and tone) — and all videos of the project inherit them. Inside a project you see all its videos, their costs and the autopilot settings. One-off videos (the “New video” button) live in a hidden service project and appear in the “Videos” section. A project can be disabled (the “Project enabled” checkbox) or deleted — together with its videos and files in storage (irreversible; it cannot be deleted while a render is running).

Open as a separate page →

Project modes: stock, avatar, split screen and more

“Stock video only” — for every sentence a clip is picked from Pexels. The cheapest mode. “Avatar + inserts” — a talking head (HeyGen) alternates with stock inserts; the AI decides where to show the presenter. “Avatar only” — the presenter for the whole clip. The avatar is billed separately (~$1–2 per minute); pick a look shot for the right orientation. Split screen — the frame is divided in two: the presenter (or an ad) in one part, stock inserts in the other. “Stock+ad” — 70% of the frame is stock, 30% is your silent ad media. “Trigger card” — a short card (photo/video + a question on top) without voiceover, for catchy previews. “Full AI render” — instead of stock, AI images (Ideogram, Flux etc.), optionally animated. More expensive, but the visuals are entirely yours. “Black bars” (for vertical videos) — a landscape frame fits entirely instead of being cropped at the edges.

Open as a separate page →

The “Show” mode: a daily news bulletin

The “Show” mode is for projects with a similar bulletin every day featuring the same hosts: news, a digest, a weather forecast. Turn it on with the “Show mode” checkbox in the project settings; until it is ticked, the whole block stays collapsed and out of the way. Hosts. Set up to 3 hosts: a name (shown as an on-screen caption while they speak), a short title, an appearance description (for image generation) and an ElevenLabs voice. First, on the “Hosts and pictures” tab, create a base picture for each and approve it, then generate 7–10 angles (“8 angles for …”) — wide and medium shots; the system deliberately avoids close-ups so imprecise lip movement is not noticeable. A favourite angle can be marked with the “Starting” star — every episode opens with it. Animate the angles (“Animate” button): if a host has animated clips, only those are used in the show, still photos are skipped. The order of hosts is hard-coded, not left to the AI: the first host delivers the greeting, the day's topics, the news and the sponsor mention, the second — only the weather block and the sponsor recap/sign-off that follows it. The same host's frames never repeat back to back. Weather. The “Weather” field accepts three formats: a single country (e.g. Norway) — the AI picks several major cities itself, the count is set by the “How many cities” field (1–6); a comma-separated list (Oslo, Bergen, Trondheim) — exactly these cities are used; a single city — as before. Each city gets one short sentence so the block does not bloat the show. Length and number of news items. “Max seconds” (30–178) is a hard ceiling for YouTube Shorts/TikTok: if the script comes out longer, the last news items are dropped first, then the voiceover is sped up slightly. “Number of news items” (1–6) sets how many topics the agent includes. Script sample. The optional “Script sample for the AI” field — paste your own example of how the hosts should sound (greeting, transitions, jokes). The AI copies the delivery style, not the facts, names or dates from the sample. Sponsors. A separate “Show sponsors” card (the “Mention sponsors” checkbox enables it): name, tagline, description or a website link (the AI reads it itself), a photo or video. With media the sponsor appears as an on-screen insert for a few seconds; without media it is a text banner with the tagline. At most two enabled sponsors make it into a show. Trial run and autopilot. The “Create a show now (trial)” button on the “Sponsors and show” tab assembles one episode right away — handy for checking the settings before turning autopilot on. Autopilot for a “Show” works the same way as for regular projects (schedule in the autopilot settings), it just assembles a fresh news bulletin every time instead of a regular topic-based video.

Open as a separate page →

Create a blogger: your own AI blogger with weekly looks

The “Create a blogger” section is a wizard that builds a channel with a permanent host in a few minutes. The blogger talks straight into the camera: the lips move to their voice, angles change, and topic inserts appear between phrases. Available from the Content Factory Pro tier. How to fill it in: 1. Name and “What they look like” — ONLY appearance (age, hair, face, glasses, build). Do not describe clothes or the place here — they are set in the weekly looks. 2. “Your photo” (optional) — the blogger will have this face, while clothes and style follow the description. You can upload yourself to get a digital twin. 3. Picture style, channel topic, character and manner of speech — the AI writes first-person scripts from them. 4. Language, orientation (9:16 for Shorts/TikTok or 16:9 for YouTube) and voice. 5. “Looks for the week” — clothes and place for each day (English works best). Clear a day and the blogger appears in the base look that day. What happens next: a project in “Studio with a presenter” mode is created, and in the background a base look plus 3 angles for each day are drawn — the same person in different clothes. Each picture is billed like a normal drawing (about 0.45 kr); the cost is shown in the form in advance. The looks appear in the project settings under “Presenters and pictures”: delete the ones you do not like, and if your balance ran out or you deleted failed ones, click “Draw the missing looks”. A video uses the looks of the weekday on which it is rendered. Everything else works like any project: topics or your own scripts, autopilot, publishing to social networks, project ads.

Open as a separate page →

Motion videos: what they are and when to use them

A motion video is a 30–60 second ad or explainer where text and graphics are the main visual rather than footage: kinetic phrases, counters, charts, buttons, calendar, timer, 3D text and a live fluid background. It is built by code to the words of the voice-over, so every element appears exactly on the voice. Where: Content Factory → Motion videos → New. After launch: the AI director writes the voice-over text and storyboard from your brief, a voice narrates it, the engine renders frames (on a GPU if premium graphics is on), then everything is mixed with music and sound effects. Usually 5–10 minutes. Good for: product, feature or plan ads; launches and announcements; facts and numbers (the director turns them into counters and gauges); step-by-step processes; before/after comparisons. Not for stories with real people, real UI or food demos — use the regular stock video or AI blogger for that. A finished video has “Repeat with the same settings” (new text and scene on the same brief) and “Redo” with a style change — only the frames are rebuilt, with no new AI or voice cost.

Open as a separate page →

What to put in a motion video brief

The brief is plain text with 3–8 points, not a script. The director writes the voice-over and picks the graphics itself, so facts matter more than wording. Must have: (1) what it is and for whom — one sentence; (2) 2–4 concrete benefits or features, ideally with numbers (“1224 channels”, “in 10 minutes”) — numbers become counters and gauges and are spoken as words; (3) one pain the product removes — it becomes the hook in the first seconds; (4) what to do and where — site, button, section — this goes into the final title. You do not need to describe format, length or style — those are set on the page. For a specific scene, add a hint right into the brief: “end with a manual vs autopilot comparison”, “show a 10-minute timer”. Bad example: “tell how great our service is” — without facts you get generic text with no graphics. Speech length: 30 s ≈ 90–105 words, 45 s ≈ 115–140, 60 s ≈ 150–175; split longer topics into several videos.

Open as a separate page →

Motion video style: every setting and the preview

The “Video style” block sets the look, not the content: the director gets it as constraints and the style is applied at render time. That is why you can change the style of a finished video via “Redo” — only the frames are rebuilt. Presets: Gold · premium, Neon · techno, Acid · street, Magazine · editorial, Alert · red, Violet · dream. A preset fills every field below; any field can be changed. Accent color and background — swatches or a custom color. Headline font and numbers font — from the palette (Unbounded, Oswald, Playfair Display, Russo One, Rubik Mono One, JetBrains Mono, Caveat); the director also varies fonts between scenes. Caps: auto / always / never. Pace: fast (chopped phrases) or calm. Energy — strength of hits, glitch and camera shake. Text placement: center, top (graphics below) or bottom like subtitles. Background graphics: auto, 3D wireframes, particles, fluid (GPU only), clean. Fluid palette: warm, cold, two colors, single color. Film look: grain and vignette. Duration: 30/45/60 s — the target speech length for the director. Checkboxes: italic serif on key words (font contrast inside a phrase) and brand label in the corner. The style preview is a live mock-up: it shows colors, fonts, text placement, grain and vignette together. There is no animation, 3D or fluid in it — it is a color proof before launch. Premium graphics (GPU) adds 3D text with lighting and a live fluid; it renders on a rented GPU and takes longer. Pick music in the “Background music” block — without it the video has only voice and effects.

Open as a separate page →

How to make a YouTube video with no camera and no editing in 5 minutes

A full YouTube video in a few clicks: voiceover, footage, subtitles and music — no camera,...

Turn your own footage into a finished video

How to make a video from your own footage in Content Factory. In this lesson: • upload yo...

A digital AI host for your videos — no camera

A host speaks your text on screen — no camera, no studio, no editing. In this lesson: • «...

A one-minute video that collects comments: the trigger card

The fastest and cheapest Content Factory format: 10–15 seconds, a picture and a catchy que...

Shorts from one long video: 3 ways

Three ways to make Shorts for YouTube Shorts, TikTok and Reels. In this lesson: • vertica...

One video in three languages — automatic translation

How to publish one video in several languages with Content Factory. In this lesson: • a s...

Your own music for videos: generate tracks and protect monetisation

How to use your own music in Content Factory videos. In this lesson: • why someone else's...

AI motion graphics and 3D explainer videos

Animated ads and 3D explainers — no stock footage and no host. In this lesson: • a motion...

Put your ad into every AI video automatically

Create an ad once — and it goes into every video of your project. In this lesson: • the «...

A TikTok and Telegram carousel in a minute: an AI image post

How to make a carousel — a post of several images — from a single topic. In this lesson:...

An agent runs your Telegram channel: AI text posts for almost nothing

The agent finds fresh topics, writes posts of the length you want and publishes them to a...

AI timelapse video: construction, restoration, a growing garden — from one spot

A sped-up transformation: building a house, restoring a car, clearing a plot or a garden g...

Chronicle from archive: real NASA and Wikimedia footage narrated by AI

Chronicle from archive is a video built on real footage from open archives: the AI watches...

Topic or own script, languages, limits

“Topic” — you write a topic and the AI creates a script in the chosen language. You can tick several languages at once: each gets its own video with its own script. Extra context (tone, what to mention) goes into the single AI script call. “Own script” — paste ready text. The language is detected from the text. End every sentence with a period, question or exclamation mark: footage is picked per sentence. If there are almost no periods, the system neatly adds punctuation before rendering (the words do not change). Script limit — up to 7000 words (~60 minutes of voiceover); rule of thumb: 150 words ≈ 1 minute. If the text is longer, shorten it with the “Shorten with AI” button. You can set your own word limit and target length range in the project settings. You can also upload your own video files — the AI analyzes them, suggests a script and edits a clip from your footage.

Open as a separate page →

Voice, voice speed, music and subtitles

Voice — choose from the ElevenLabs list (▶ plays a sample, ★ adds to favorites). Your own cloned voice does not appear in the shared list — enter its ID and a name in the project settings. Voice speed — from 0.7 to 1.2 (empty = normal). For news and fast formats 1.1–1.2 works well. Subtitles stay in sync. Music — the “Background music” block: no music by default; tick “BelkaFood music” (the service library, automatic or a specific track) or “My music” (your tracks in order) — details in the “My music” article. Default volume is 0.12 (0–1); it can be changed in the project. Subtitles — the “Subtitles on video” checkbox: unticked means no subtitles on the clip. Size (30–100) and position (top/middle/bottom) are set in the project or in a one-off video. Subtitles are “karaoke” style: words light up as they are spoken.

Open as a separate page →

Ads in a video: two modes and the marker

In the video creation form you can add up to 3 ad inserts. For each one choose how to show the ad: • “Muted — the AI weaves the ad text into the script”: you enter the ad text and upload a photo or video. The AI retells the ad in its own words inside the script (with a heads-up and a thank-you to viewers) while your media plays silently on screen. Works in “Topic” mode only. • “With its own sound — whole clip, narration waits”: you upload a ready ad clip (video). It plays in full with its own sound; the narrator's voice and the music go silent meanwhile and continue afterwards. No subtitles are shown on the ad. Works in “Topic” and “Own script” modes. Where the clip with its own sound goes — the “When to show” field: • “By marker in the text” — the [AD_1] button becomes active: click to copy and paste the marker into the script after any sentence. In “Topic” mode the AI places the marker itself, and you can move it later in the script editor. • “At the start / In the middle / At the end” — the ad goes there automatically (no marker needed). If there is no marker but “By marker” is chosen, the ad goes to the end. The product link and ad description go to the START of the video description, above the clip's own description. For now the own-sound ad is supported only in the regular stock mode (not in avatar and split-screen modes).

Open as a separate page →

My music: your own tracks, playlists and copyright

The “My music” section (under “My work”) is your personal background music library. Only you see and use it. Tracks you generated with AI (ElevenLabs Music) appear here automatically, marked “generated”; tracks you uploaded are marked “uploaded”. Adding a track: “My music” → “Add track”. Upload a file (MP3, WAV, M4A, FLAC, OGG, up to 50 MB — the server compresses it to MP3 192 kbps) and fill in: title, artist/author, source (e.g. YouTube Audio Library, Pixabay, your own recording), license, link to the license or track page, the “License requires crediting the author” checkbox and the attribution text for the video description. You can add a personal comment, e.g. “for calm videos”. You can upload up to 50 tracks. A track can be edited or deleted (videos already using it don't change). Copyright: if the license requires credit, fill in the attribution text — the track won't save without it. Copy exactly what the source requires: in YouTube Studio → Audio Library, tracks marked “Attribution required” have a “Copy” button. This text is added automatically to the end of the description of every video using the track (when publishing to YouTube and Telegram), as a line with 🎵. If a video uses several tracks, all credits are added without duplicates. Tracks that don't require credit are not added. The “Background music” block is the same in every creation form (New video, project video, From my videos, Quick clip, films, series) and in project settings. It has two checkboxes, both off by default — then the video has NO background music. “BelkaFood music” is the service's ready-made library: choose “Pick automatically” (different tracks in turn) or a specific track, each with ▶ preview and a personal comment. “My music” is your own tracks. Only one of the two can be on. The top right of the block shows what is selected. Choosing your tracks: tick “My music” — the “Choose tracks” button becomes active. Click tracks in the order you want: each gets a number 1, 2, 3… (clicking again removes it, numbers shift). You can also play a track ▶ and edit its comment there. The order and total length are shown under the button. How the playlist plays: tracks play one after another in the chosen order and loop until the video ends. Example: a 10-minute video with two 3-minute tracks → 1, 2, 1, 2 and 1 again (cut at the end of the video). A single track shorter than the video simply repeats. Track loudness is levelled automatically, there is a smooth 3-second crossfade between tracks, and the music fades out at the end. Overall music volume is a project setting. Music for the whole project: project settings → “Music” tab — the same block; click “Save project music”. All project videos, including autopilot, use it; in the new-video form of the project the block is pre-filled from the project and can be changed for a specific video. For a project, BelkaFood music is always picked automatically. A new project is created without music. Picking one specific track in the video editor replaces that video's choice. Generating a track with AI (ElevenLabs Music, commercial license included): “My music” → “Generate your own music” — describe the mood, choose the length, the track appears in your list in about a minute. Generation is only available in this section.

Open as a separate page →

Quick clip: a photo comes alive in 5–15 seconds

“Quick clip” (Create menu) is a short video up to 15 seconds with one character from a photo: a person, cartoon character, pet or toy. For longer stories with scenes and several characters use “My films”. How it works: your photo is the first frame — the character, place and light come from it. The clip is built from 5-second scenes (5 sec = 1 scene, 10 = 2, 15 = 3), each continuing the previous one from its last frame. Each scene is animated by the LTX-2.5 model on our GPU server (LTX-2.3 as backup), then scenes are joined and voice, subtitles and music are added. Script — two modes, the “AI director” checkbox: • On (default): one field “What happens in the clip”. The AI (Gemini) splits the idea into 5-second scenes, invents the movements and, if the character speaks, short lines, and translates everything into English for the model. Handy when you just have an idea. • Off: a field appears for each scene. Your text goes to the model as written — no text AI is involved and the photo isn't described by AI either. The model understands English best — write simply and concretely: “the character waves and smiles at the camera”. If the character speaks, each scene has a line field voiced word for word (up to ~10 words per scene). Scene parameters (“Parameters” button): camera movement (static, push-in, pull-out, pan, orbit, follow, handheld), shot (wide, medium, close-up, detail), character movement intensity and emotion. They become precise wording for the model. Always on: same character, always fully in frame, same place, no other people and no scene changes — so the clip doesn't drift. Voice: “The character speaks” — choose a voice (ElevenLabs or Gemini), lips are synced with speech; “No speech” — movement only, no voice or subtitles. Background music — the “Background music” block (no music by default). Tips: a clearly visible character in the photo and simple actions give the most predictable result. The model follows your wishes but doesn't guarantee every detail — if a scene fails, simplify the text or change the parameters.

Open as a separate page →

Video thumbnail: caption, update and AI thumbnail

The finished video page has a thumbnail block. What you can do there: • “Thumbnail hook caption” — a catchy text on the image. A long text is wrapped automatically into 2–4 lines so it does not run off the edges. • “Update thumbnail” — the same frame with the new hook text (even if you have not pressed “Save” yet). • “Another frame from the video” — picks a different random frame. • “Generate thumbnail with AI” — choose a model (for example Ideogram or Nano Banana) and the AI draws a juicy, clickable thumbnail from the title, description and hook. You do not write a prompt. This is a paid operation: it is charged from the AI balance like a regular AI image; if the model fails, nothing is charged. • “Upload your own thumbnail” and “Download thumbnail”. Note: YouTube does not let our channel upload thumbnails via the API, so YouTube uses an auto-cover from a video frame by default. You can download the thumbnail here and upload it to YouTube manually. The “Regenerate description/hashtags/hook” button recreates only the texts — the footage does not change. To get a different video with the same settings, use “Repeat” in the “Parameters” panel.

Open as a separate page →

Videos without a voiceover

If you do not need a voice, untick “Voiceover” in the project settings or in the “New video” form. ElevenLabs is then not used (the voiceover is free), and the clip is assembled like this: the script is split into sentences, a stock video is picked for each, the footage changes at reading pace, with background music and subtitles (readable) and, if needed, text captions on top. Reading pace is set in words per minute (160 by default; the lower, the longer each shot stays). The minimum length of one sentence is 2.2 seconds so the text can be read. Without a voice the music is the only sound, so its default volume is higher (0.5); you can change it in the project settings. The mode works with stock and AI-image modes. It is not available with an avatar (the avatar speaks with a voice). For short question cards there is a separate “Trigger card” mode. Your own voice instead of a voiceover: on the video page of such a project there is a “Record my voice” button — you read the text in the browser and the clip is built to your pace (see the article “Recording your own voice: by phrase and teleprompter”).

Open as a separate page →

Recording your own voice: by phrase and teleprompter

For videos from a “no voiceover” project you can record your own voice right in the browser instead of ElevenLabs — no downloading or external editor needed. The voiceover cost is zero. How to start: on the video page click “Record my voice”, allow microphone access (a standard browser prompt) and choose a mode. • “By phrase” — one phrase on screen. Press “Start” or Space, read it, press Space again — the phrase is saved and the next one starts immediately. Any phrase can be played back (▶) and re-recorded. The most precise option: every phrase is recorded separately. • “Teleprompter” — the whole text scrolls smoothly, the current phrase is highlighted. The slider sets the speed (words per minute) and can be changed while reading — find a comfortable pace. “Pause” — to breathe or cough (recording pauses too), “Previous phrase” — if you slipped: the failed attempt will not get into the clip, “Next (Space)” — to move the highlight faster. Read along with the highlight. At the end you can listen to the whole recording and read again. After “Save and build video” the clip length and the shots adapt to your actual reading pace, and the music plays quieter than the voice. If you change the script later, the recording is reset (the text changed — record again). “Remove recording” returns the video to silent. Tips: record in a quiet place, keep the microphone at a constant distance; there is no noise suppression.

Open as a separate page →

Video editor: replace shots and on-screen text

A finished clip can be tweaked without re-voicing it and without paying for the voiceover again. On the finished video page click “Editor: replace shots and text”. For every phrase of the script you can: • Change the keywords for finding the shot (in English) and the type: video or photo. • “Find a shot” — 8 options from Pexels with previews; click an option to pin that shot to the phrase, “Reset choice” returns automatic selection. • Set on-screen text: a title, a list or a quote. “Save and rebuild video” queues the clip: the voiceover and timings are taken from the existing ones, only the picture changes. “Save only” stores the edits without rebuilding. In avatar modes only stock inserts are replaced, the presenter's phrases stay as they are. The editor is available after the first video generation. For now you cannot reorder phrases or upload your own shots — only replace stock shots and edit captions.

Open as a separate page →

A version of a video in another language

Any video with a script can quickly be repeated in another language. On the video page choose a language (RU, EN, NO, SV, DA, DE, FR, IT, ES, PL) next to the “Version in another language” button and click it. What happens: the script is translated by the AI (ad markers [AD_1] are kept), a new draft is created with the same project, music and voice, the title gets a suffix like “(EN)”. You land on the edit page — check the text. Important: the voice is copied as is. For the new language pick a suitable voice (ElevenLabs multilingual voices also read other languages, but an accent may be noticeable). Then start generation — the voiceover of the new clip is billed as usual.

Open as a separate page →

Video languages: which languages you can make clips in

The Content Factory makes clips in 10 languages: Russian, English, Norwegian, Swedish, Danish, German, French, Italian, Spanish, Polish. The language is chosen in the video creation form and in project settings (“Default language”). What depends on the language: the script (when the AI writes it from a topic), punctuation, footage tagging, and the title, description and hashtags for publishing. The voiceover reads the script in the language it is written in: the ElevenLabs voice detects the language from the text itself. So if you paste your own script in German and pick “Deutsch”, the clip comes out in German. Choose a voice that sounds good in the target language and make a short test clip before a series — pronunciation and accent differ between voices. The panel interface can also be switched to any of these 10 languages: use the language switcher (language code) in the top-right corner of the panel. The public site stays in Russian, English and Norwegian; help articles in the other languages are shown in English.

Open as a separate page →

My films: a cartoon of scenes with consistent characters

The “My films” section creates sequential films: the AI writes a story with dialogue, draws the characters, animates the scenes, voices the characters with different voices and assembles the finished film. The characters stay the same from scene to scene. How to create: “My films” → “Create film”. Describe the idea, choose a style (3D cartoon, anime, watercolor, clay, cinematic, photorealistic), the format (landscape 16:9 or portrait 9:16 — frames are drawn in the chosen proportion), the duration (30 seconds to 10 minutes), the narrator language and voice; you can turn on subtitles and add advertising (see below). Step 1 — story (free). The AI writes the characters and scenes: mostly dialogue between the characters, lines of 8–24 words, varied shots, a beginning and an ending. The film stops with the status “Waiting for story review”. You see a cost estimate and can edit character descriptions, lines, frame and motion descriptions, pick a separate voice for every character (▶ plays a sample) and the speaker of each scene. Don't like it — “Write the story again (free)”. Step 2 — creation (paid). Press “Approve and create film” and confirm the price. Reference portraits, scene frames (based on the references, so the characters do not drift) are drawn, frames come alive as 5 or 10 second clips (depending on the line length), lines are voiced and the lips of speaking characters are synchronised with the voice. Background music is added. If a model refuses to draw a frame, the system tries other wordings, softens the description, uses a backup model and as a last resort substitutes a neighbouring frame — the film does not stop. If something still stops (for example the balance runs out), press “Continue from where it stopped”: finished work is not paid for again. Any scene can be regenerated (paid). Your own script: when creating a film choose “I have my own script”, paste the text (a play “NAME: line”, a story, book chapters — up to 60,000 characters) or upload a .txt. The AI finds the characters and lays out the lines almost unchanged, you check the characters and voices, draw the characters, voice the script (pennies) and press “Create film” — the AI invents frames for the finished voice-over, you review them and confirm the exact price. Such a film is at most 30 minutes long; it appears in “My films” in the “Films from your script” block. Finished film: download it or press “Publish / schedule” — it becomes a regular video with publishing to channels and the Planner. Cost: about 4–5 kr per 5-second scene (a 10-second clip costs twice as much) plus about 2.5 kr for lip sync per character line; the exact estimate is shown at the review step. Charged from the AI balance as generation goes.

Open as a separate page →

Studio with a presenter: the same host in every video

The “Studio with a presenter” mode is for channels where every video has the same host (news, reviews, stories) without paying for a “talking head” in each video. Setup: in the project settings choose “Studio with a presenter”, describe the host (for example: “a confident man of 35 in a dark jacket, a news studio”), pick the picture style and save the project. A “Studio host” block appears below. Host pictures: press “Create base picture” (about 0.45 kr). Then create more angles (“looks aside”, “holding a tablet”) — they are built from the first picture, so the character stays the same. Press “Approve” on the pictures you like and delete the rest (up to 6). Until at least one picture is approved, videos in this mode will not start. Animating: press “Animate” on an approved picture — the AI makes a short 5-second loop (the host blinks and moves the head and hands). It takes 1–3 minutes and costs about 3.8 kr. If the project has animated loops, videos use only them (in rotation), otherwise still pictures with a gentle camera move. To remove a picture from videos, remove its approval or delete it. How the video is built: the host's pictures or loops are placed on the host's lines, and between them (if “Alternate the host with stock footage” is on) — stock footage on the topic. Voice-over, subtitles and music work as in regular projects; autopilot and your own scripts work the same way. Why it is cheap: you pay for the host's pictures and loops once, when you create them — after that they are reused in all the project's videos for free, like your own stock footage. In each video you pay only for the voice-over. Limitation: the 5-second loop repeats and the host's mouth does not move exactly with the words — enough for news and review videos.

Open as a separate page →

Series: consistent characters, episodes and your own script

The “Series” section is a shared world with consistent characters, and 1–10 minute episodes continue the story. Characters are drawn once and look the same in every episode. Two ways to create a series (chosen at creation): • “AI invents from a topic” — you describe the world and mood, the AI invents 2–5 characters (you can edit them). Draw each one (“Draw”, about 0.45 kr) and pick a voice — by default the AI casts different voices itself. Then “New episode”: describe the plot or leave it empty — the AI continues the story using summaries of previous episodes (written automatically after every finished episode) and leaves a hook for the next one. • “I have my own script” — paste a play (“NAME: line”), a story or book chapters (up to 60,000 characters) or upload a .txt. The AI finds the characters and lays out the lines, hardly changing your words. Check the characters and voices, draw the characters and press “Voice the script” (pennies): the exact length becomes known. The script is split into episodes by time without breaking scenes; change the number or length of episodes with “Recalculate split” — every episode shows its duration and price. “Create episode” builds the episode for the finished voice-over: the AI invents frames and motion, you review and confirm the price (the lines themselves can no longer be edited in this mode). An episode opens as a regular film (review, generation, publishing). Dialogue, different voices and lip sync work as in “My films”. Advertising in series: in the “Advertising in series” card add a product (upload a photo — a hero uses it in the frame, packaging and logo are preserved) or a service (heroes mention it in dialogue). You can set scenes per episode and episode numbers. Scenes with advertising automatically carry an “Advertising” label. In own-script mode product advertising works; spoken service lines are not added there because the voice-over is already recorded.

Open as a separate page →

The “Ads” section: a library of ad campaigns

The “Ads” section is a library of your ad campaigns. Create an ad once and pick it from the “From the ad library” list in any ad block: in a video and a single video, in project ads, for show sponsors, in films and series. What to fill in for a campaign (“New ad”): - Name and brand; - What you advertise — a product or a service; - “What to say and show” — facts and details. In videos the AI weaves it into the script in its own words; in films and series the characters mention it or show the product; - Call to action / short description — for example “20% off with a promo code”; - Link — goes into the video description; - Product photo — a film character or the blogger will hold the product, keeping the packaging and logo; - A ready picture or clip — shown full screen during the ad insert and in “Stock + ad” mode. If the clip has its own sound, tick “show in full”. When you pick a campaign, the ad block fields do not need to be filled in — everything is filled in automatically. If you delete a campaign, the ad stays in videos already made: they keep their own copy of the data. The section itself is free; you only pay for generating the videos the ad is placed in.

Open as a separate page →

Editing films and series: “Dynamic”, “Classic” and verified presets

When you create a film or an episode there is an “Editing” block — how the video is assembled: speech speed, pauses, how often the picture changes, subtitles and music. “Dynamic” is selected by default — a brisk pace without pauses or frozen frames. “Classic” is calmer, as before. Presets marked “verified” are settings of successful videos saved by an administrator. “Detailed settings — what they do and cost”: - Speech speed — speeds up the voiceover without changing the pitch (×1.12 — brisk but clear). Free. - Pause after a line — how long the frame is held after the last word. Free. - Trim silence at the start and end of lines. Free. - Max shot length — a long line is split at a pause into short shots, the second one tighter (a “push-in”). No new pictures are drawn, but each shot is a separate animated clip (paid per shot). 0 — do not split. - Animation model (lips to sound) — runs on our GPU; the clip price is shown next to it. - ElevenLabs voice stability — higher: the character's voice is the same in every line; lower: more emotion. Gemini voices are kept consistent by one instruction per character and are free within the quota. - Word-by-word highlighted subtitles — the spoken word lights up (if on-screen subtitles are on). Free. - Final call-to-action card — a banner over the last scene, e.g. “Comment WANT”. Free. - Your own music (ElevenLabs Music) — a unique track for the length of the video, saved to your library. Paid per minute; empty — a track from the library or rotation. Series settings apply to all of its episodes.

Open as a separate page →

Your own photo for a character or presenter

You can use your own photo as the face reference in three places. Film and series characters. The character card has “Your own character photo (optional)” with two modes: - “Photo as is — becomes the reference” — free, the photo is used directly for all frames with this character; - “From my photo + description” — the AI draws the character with this face, but in the right look and film style (for example, cartoon). Paid, like drawing a character. Studio presenter. In the project settings, “Presenters and pictures” block: “Your photo” when creating the base picture — the presenter gets this face, while the look and studio come from the presenter description. Other angles are built from the base picture. You can also upload your own ready presenter frame or your own short presenter video (up to 12 seconds) — free and more reliable than any generation. Blogger. In the “Create a blogger” wizard — the “Your photo” field. Tips: a front-facing photo, good lighting, one face in the frame. Glasses and jewelry from the photo are kept. Upload only your own photo or photos of people who have given consent.

Open as a separate page →

A video from your own footage

The “From my videos” section assembles one video from your own recordings — for example, from ten short clips of one repair job. 1. Upload your videos: several files, each up to 2 GB, MP4, MOV (from a phone, as is) or WEBM. Large files upload in parts; if an upload breaks off, delete it and add it again. Previously uploaded videos are in “My sources” — you can add them again, name them and star ★ them. 2. Choose what video to make: - “Keep my voice” — you speak on or off camera; the AI transcribes the speech, removes pauses, repetitions and digressions, and adds subtitles and quiet music; - “Voice-over story” — the AI watches the video, picks the best moments and writes text read by the chosen voice; - “No words, to music” — the best moments edited to music with short step captions. 3. “What happens in the video” — optional but helps the AI a lot (for example: “replacing a wheel bearing, Mercedes 190”). 4. Video length (30 seconds – 10 minutes) and orientation. If there is less useful footage, the video will be shorter. Before assembly you will see the cut and can adjust it. Deleting a source permanently also removes the file from storage — finished videos stay, but they can no longer be rebuilt from that file.

Open as a separate page →

Music: your own library and AI-generated tracks

Every creation form (video, single video, film, episode, project) has a “Background music” block. - “Automatic” — a track from the shared pool in rotation (the one not played for longest goes first). - A specific track — from the shared pool or your personal library. You can preview a track ▶, star it ★ and leave a note only you can see. - “Generate your own music (AI)” — describe the mood (for example “calm background, soft synths”) and click “Generate”. It usually takes about a minute; you can keep filling in the form — the track appears by itself. A long track suits long videos. The track is saved to your library and available in all future videos, a commercial license is included. If generation fails, nothing is charged. A project can keep its own music pool (the music tab in the project settings) — project tracks alternate between its videos.

Open as a separate page →

Insert design, subtitle style and the AI image model

Insert design — how titles, lists and quotes look over the video (cards, backing panels, darkening, different fonts, entrance animation). “Auto” picks its own design for each new video. Subtitle style — color, outline, highlight and how many words per line. Films and series also have “Word-by-word highlighted subtitles” in the “Editing” block. AI image model — in modes where the AI draws the frames (full AI render, films, studio). Models differ in price and style: Flux Schnell — the cheapest, Flux Pro — higher quality, Recraft V3 — illustration and vector, Ideogram and Gemini — good at text and scenes, Qwen-Image and SDXL-Turbo — on our own server (cheaper). Next to a model you can leave your own note — only you can see it (handy for noting what worked). The animation model (how the picture moves) is chosen where it is needed: in the project or in a film's “Editing” block.

Open as a separate page →

Animated clips, scene regeneration and stopping generation

While generation runs, the video, film or series page shows a live stage indicator — the page refreshes by itself; you can close it and come back later. While you wait, you can pick a dinner recipe — the banner leads to Belka Food and can be closed with the cross. Animated clips. In films and series, each scene card plays the animated clip itself, and the “⬇ animated clip” link downloads it separately. In regular videos with AI frames, links to each frame's picture and clip are in the video editor. Regenerating a scene. In the scene card, tick what to redo: the frame, the voiceover or the clip (motion), and click “Regenerate”. Only what is ticked is charged. A new frame or voiceover of a speaking character automatically remakes the clip too, so the lips match the sound. Nothing ticked — everything is regenerated. Stopping. The “Stop generation” button — everything already done is kept, and you can resume from where it stopped. Finished frames and clips are not charged again. If generation fails because of a server problem (for example, a GPU did not start), you are not charged for the failed part.

Open as a separate page →

Voices: ElevenLabs and Gemini

The voice picker has two kinds of voices. ElevenLabs — the most natural voices, billed by text length. In films you can set “Voice stability” in the “Editing” block: higher — the voice is the same in every line, lower — more emotion. Gemini (voices marked gemini) — free within the quota; each character has a permanent voice instruction (manner, pace), so they sound the same from line to line. You can preview any voice ▶ and star it ★ — favorites are shown first. In films and series, voices are assigned to characters automatically (different male and female voices); you can change them in the character card. Your own cloned voice is set in the project settings voice field and is available only to you.

Open as a separate page →

Carousel: a multi-image post for TikTok and Telegram

A carousel is a post of several images that people swipe through. On TikTok and Instagram such posts often do better than videos: people save and share them. How to make one: “Create” → “Image posts” → “Carousel”. Write the topic, choose the language and the number of slides (4 to 10). If the topic contains a number of items — for example “5 mistakes when brewing coffee” — there will be enough slides to cover every item: an opening slide, the items themselves and a call to action at the end. Images — two options: • Stock photos — free. • Generate with AI (Nano Banana) — beautiful images made for each slide. You choose how many slides to generate: generated images go on the first slides, the rest come from stock. One image costs about 0.5 kr and is charged to your AI balance only if it succeeds. If the AI cannot generate an image right now, the slide gets a stock photo, nothing is charged and the page tells you about it. The AI writes a large headline and a short text for every slide, with a catchy question or promise on the first one. The post caption and hashtags are filled in automatically. Publishing: on TikTok as an image post, on Telegram as an album. YouTube does not accept image posts. You can download all slides as one archive together with the caption — for example to post on Instagram by hand. One finished carousel uses 1 minute of your plan's monthly limit.

Open as a separate page →

Text posts: the agent runs your Telegram channel or group

The “Text posts” mode is for people who run a Telegram channel, group or page: news, laws, tips, expert notes. No video or voiceover — the agent writes the post, picks an image and publishes on schedule. How to set it up: 1. “Create” → “Posts” → “Text posts”. 2. Write the task for the agent: what the channel is about and what posts to write. For example: “Fresh changes in Norwegian law for immigrants. Plain language, with a “What it means for you” block at the end”. The more precise the task, the better the posts. 3. Choose the post length in characters: 500 — a short news item, 1000 — a regular post, 2000–4000 — an article. 4. Image: stock photo (free), AI-generated (about 0.5 kr) or no image. 5. Where the text goes: in the image caption (up to 1024 characters — Telegram's limit) or as a separate message after the image (up to 4096). 6. Optional: the post title on the image and its font, emoji, source links at the end. 7. On the “Autopilot” tab set the publishing times and channels. Where it publishes: Telegram — to a channel, a group or a specific topic of a group (connect the topic by its link in “Social networks”; several topics are possible, each works as a separate channel). TikTok and Pinterest — as a photo post with the text in the description (they need an image). YouTube does not accept text posts via its API. What the agent does each time: finds a fresh topic for your task, checks facts on the web (it does not invent numbers or dates), writes a post of the chosen length in your style and structure, picks an image and publishes. You can also make a post by hand: the “Create post now” block in the project settings. The finished post opens on its own page — you can edit the text before publishing. “Your own text” mode — when you write the post yourself. The “Create post now” block has a “Your own text” field: write your thoughts as they are, with typos and no structure, or press “Dictate” and say it out loud (up to 5 minutes, the text appears in the field and can be edited). The agent does not search for a topic and invents nothing: it keeps your meaning, tone and promises, fixes mistakes, removes repetition, splits the text into paragraphs, adds a catchy first line and a short call to action. Leave the field empty and the agent finds a fresh topic from the project task as usual. With an image or text only: next to “Create post now” choose the image for this particular post — as in the project settings, stock photo, AI image or “No image — text only”. Text only is published to Telegram as a regular message; TikTok and Pinterest need an image. Price: about 0.2 kr per post from the AI balance (topic search and text), an AI image adds about 0.5 kr, stock photos are free. 10 posts use 1 minute of the plan's limit — the free 30 minutes give 300 posts a month. It is the most affordable Content Factory format.

Open as a separate page →

3D videos: a 3D world for your topic

A 3D video is a short clip (30–60 seconds) in a real three-dimensional world: a city with cars, a kitchen, a room, a forest camp. A host character walks around the scene, objects appear and move, coloured zones show “right / dangerous”, 3D captions and numbers hover above the scene. The camera automatically shows the place where the action happens. How to make one: “Create video” → “3D video” (or the “3D videos” section in the menu). Choose a world, write a topic — for example “who goes first at an uncontrolled intersection” or “how to put out a campfire safely” — the language, duration (≈30, 45, 60 or 90 seconds — the AI writes a script of matching length), format (vertical 9:16 or horizontal 16:9), voice and music. The rest is automatic: the AI director writes a voice-over for the chosen duration and stages the scene using what exists in the world; a voice narrates it; a GPU is rented for your video and Blender renders the 3D world; then sound effects (engine, brakes, a whoosh on captions), music and subtitles are added. Usually 15–20 minutes. Cost: GPU rendering is charged to the AI balance at actual cost — about 1–3 kr per video, plus the voice-over. A balance of at least 5 kr is required. A finished video can be downloaded, “Repeated with the same settings” (new text and scene on the same topic) or “Redone” with another topic. With the same topic only the video is rebuilt, and the voice-over is not charged again.

Open as a separate page →

3D video worlds: what you can do in each

The world is chosen in the 3D video form and in the project autopilot. There are four now; new worlds appear in the same list. 🚗 City intersection — traffic rules, driving schools, “who goes where”. Cars of different colours, taxis, police, ambulance, fire trucks and lorries arrive from any side, go straight or turn left and right. The intersection can be uncontrolled, have priority road signs (“Give way”, “Stop”) or traffic lights that change as the story goes. A green zone under a car means it goes, red means it gives way; numbers above the cars show the order. Example: https://img.belka.no/belkafood/content-factory/scene3d/2/2/video-pA7zyR.mp4 🍳 Kitchen — recipes, food storage, kitchen safety. The chef host walks to the fridge, sink, cutting board, stove and table. About a hundred foods and kitchen items: eggs, bread, cheese, meat, fish, vegetables, fruit, frying pan, pot, plates, cutlery, ready dishes. Food appears on the counter and flies to the stove or the plate; a red zone means hot or dangerous, green means right; labels show time and amount (“3 MIN”, “2”). Example: https://img.belka.no/belkafood/content-factory/scene3d/2/3/video-C25HLY.mp4 🏠 Room and workplace — work, study, home, gadgets, habits. The host walks to the desk with a computer, the TV, the sofa, the coffee table and the bookcase. Items: laptop, monitor, books, mug, boxes, plants, lamp, speaker, snacks. Example: https://img.belka.no/belkafood/content-factory/scene3d/2/4/video-j6DyIr.mp4 ⛺ Forest and camp — hiking, survival, outdoor safety, ecology. A clearing by a river: a tent, a campfire with log benches, a big rock, a trail with a signpost, forest. Gear: bucket, axe, shovel, box, sleeping bag, fishing rod and fish, firewood, a pot. Example: https://img.belka.no/belkafood/content-factory/scene3d/2/5/video-72ukRt.mp4 All worlds have 3D captions (up to three short lines), a big “?” for questions and an automatic camera. Tip: write the topic so it can be shown with the world's objects — the AI stages the scene literally from the text.

Open as a separate page →

Timelapse: a sped-up film of a transformation

A timelapse is a video where a house is built, a car is restored, an overgrown plot is cleared or a garden grows before your eyes. It looks as if it was filmed from one fixed spot over weeks of work, and is shown in 15–60 seconds. How to make one: «Create video» → «Timelapse» (or «My works» → «Timelapses»). Choose a scenario (construction, restoration, cleaning and landscaping, room renovation, growth and seasons, or your own story) and describe what changes — for example «a wooden sauna on the shore of a mountain lake». Settings: • Video type — cinematic, photorealistic, 3D cartoon, anime, watercolour, claymation, miniature. • Light and weather — sunset, overcast, night, winter. • Format — horizontal 16:9 (YouTube) or vertical 9:16 (Shorts, Reels, TikTok). • Length — 15, 30, 45 or 60 seconds; the number of frames is calculated automatically. • Pace — «fast» (speed-up, stronger timelapse effect) or «calm» (no speed-up, fewer frames, cheaper). • Shots — one view, or with a change to a medium and close-up shot, as in film editing. • Final reference — upload a photo of the result you want; it becomes the last frame and the stages are drawn backwards to the start. • Project — the video appears there for publishing; the project's ads are added there too. Storyboard: the AI plans the stages and draws the frames in order — each continues the previous one, so the place and the view stay the same. At this step you only pay for the images. Any frame can be edited or redrawn — the following frames adapt. Animation: after «Approve and animate» a graphics card is rented, smoothly animates the transitions between frames and is switched off right after the work. Then the video is assembled: speed-up, stage captions, music. Usually 10–20 minutes. The finished video can be downloaded, published to your channels and added to the planner. Autopilot: in the project settings → «Autopilot» choose the source «Timelapses», set the series topic and settings — the AI picks a new subject every time and releases a video on schedule without manual approval. Cost: a storyboard frame costs one image (shown on the create page), animation is charged by the actual graphics card time, usually a few kroner per video. Everything is taken from the AI balance.

Open as a separate page →

Chronicle from archives: real footage from open archives

"Chronicle from archive" is a format where the video is built on real footage instead of stock clips laid under a script: a volcano erupting, a rocket launch, historical film, nature. The AI watches the footage you pick and tells what is happening in it. Where to find it: "+ Create video" → group "Real footage from archives" → "Chronicle from archive". How it works: 1. Choose the video type (event, space and science, historical chronicle, nature, technology) and write the topic in any language. The agent writes English search queries itself and picks clips. 2. Sources: NASA, Wikimedia Commons and vetted Internet Archive collections. Only material that may be used in commercial videos is taken: public domain, CC0 and CC BY. The source, author and licence are shown on every card. 3. Press ▶ on the preview to watch the clip in the card. "⬇ Download" saves the video as is, without processing. "Use this video" sends it to assembly. 4. After "Use this video" the "Video from your own materials" page opens in narration mode: the AI picks short moments and writes text for each, without pauses. Choose the voice (or a blogger — see below), subtitles, music, length and format. 5. The source and licence are added to the end of the finished video's description automatically — publish it with them unchanged. What it costs: • The agent's search — 0.20 kr from the AI balance per search (an empty result is not charged; factory admins do not pay). • Downloading the video as is is free. • Assembly on our machines: voice and processing are charged from the AI balance, video minutes come from your plan limit. Blogger instead of a voice: the assembly page has a "Who tells the story" block. "Blogger at the bottom" or "Blogger at the top" — a HeyGen presenter says your story in the first person and takes half of the screen while the video plays in the other half. The blogger is charged from the AI balance by the minute of the finished video and costs noticeably more than a voice-over. Ads are not woven in in this mode. Good to know: • Open archives have little fresh news or disaster footage; nature, space, science and historical film are found best. • Footage is often horizontal: in a vertical project it is cropped to the centre. • If an archive does not respond, wait a minute and search again.

Open as a separate page →

Favourites and folders in Chronicle: collect videos for one topic

In "Chronicle from archive" you can save videos you like to favourites and sort them into folders, to gather several videos for one topic. How to save: press ☆ in the corner of a preview. A choice appears: "No folder", any of your folders or "New folder" (type a name and press "+"). If the video is already in favourites, the star is yellow and the window has "Remove from favourites". Saving the same video again simply moves it to another folder. The "★ Favourites" tab: at the top there are filters "All", "No folder" and your folders with counters. Here you can create, rename and delete a folder (videos of a deleted folder stay in favourites without a folder). A video in favourites can be played, downloaded and used in a video just like in search. Build one video from a folder: open the folder and press "Build a video from the folder". All its videos (up to 6) are downloaded and "Video from your own materials" opens: the AI watches them and writes one coherent story about the whole topic. The sources and licences of all videos are added to the description. Limits: up to 500 videos in favourites, up to 30 folders, up to 6 videos in one build. Favourites themselves are free.

Open as a separate page →

Stock photos and videos in “Chronicle” and the “From Chronicle and stock” button

“Chronicle” has a “Stock photos and videos” tab: search across Pexels and Pixabay — free services with an open license: commercial use is allowed and no credit is required. Search is free. How to search: open “Chronicle” → “Stock photos and videos” tab, type a query (any language), tick “Photo” and/or “Video” and the services (Pexels, Pixabay), press “Search”. What you can do with results: • ☆ — save to favorites and sort into folders, like archive videos (photos carry a “Photo” badge). • ⬇ — download as is. • Videos can be taken into a clip (“Use this video”), published as is, or assembled from a folder. Photos never go into video assemblies: they are for images (storyboard, sample, reference). The “From Chronicle and stock” button: next to every image upload field in the Content Factory (films, series, timelapse, bloggers, etc.) there is a “📁 From Chronicle and stock” button. The window has two tabs: “My folders” (your saved photos) and “Search stock” (Pexels and Pixabay search right in the window). Tick one or more photos and press “Add” — they are put into the field like a normal upload and then work like your own file. If the page was open for a long time and search says “CSRF token mismatch”, reload the page. Good to know: • The stock license allows commercial use but gives no right to show identifiable people in advertising. Do not animate identifiable people or use them in ad videos. • Mark AI-generated content on platforms according to their rules. • Video search needs a working service key; if a service is temporarily unavailable, the search tells which one.

Open as a separate page →

AI films, cartoons and series from one sentence

Make a film, cartoon or series from a one-sentence idea. In this lesson: • styles: anime,...

Make AI cartoon series and films with the same characters

How series and films are made in Content Factory. In this lesson: • a series: world, hero...

The Laboratory: full manual control over AI films

The Laboratory is the manual control desk for long AI films in Content Factory. In this l...

The Lab: who can use it and how it is paid

The Lab is a manual console for shooting a film scene by scene: characters with a photo and a description of their looks, a start frame for each scene, the action from start to end, a line of dialogue, film settings (shot size, lens, camera, light, colour) and the settings of the video model itself. Every scene can be shot in several takes so you pick the best one, and then the whole film is assembled. Access: from the Starter plan. The “Lab” item appears in the Content Factory menu automatically. Payment: generation runs on GPUs and paid AI models, so every action — a character portrait, a start frame, shooting a scene, assembly — is charged to your AI balance at the real cost plus a small markup. To start a generation you need at least 5 kr on the balance; if there is less, the Lab asks you to top up and starts nothing. A charge never takes the balance below zero. Saving the film and changing scenes and settings is free — money is spent only on generation.

Open as a separate page →

Connect YouTube, TikTok, Pinterest and Telegram channels

How to connect your channels to Content Factory for automatic publishing. In this lesson:...

YouTube autopilot: a channel that posts by itself

Set it up once — and your channel fills itself. In this lesson: • connecting YouTube, Tik...

The AI agent that picks video topics that work

How the autopilot's AI agent chooses topics for your channel in Content Factory. In this...

The publishing calendar: schedule every video and post

How the «Auto-publishing» calendar works in Content Factory. In this lesson: • the monthl...

Long YouTube videos and your own thumbnail: what to know

How to publish long videos and custom thumbnails on YouTube with Content Factory. In this...

AI replies to your YouTube comments in every language

How Content Factory answers your viewers in the comments. In this lesson: • the comment q...

Statistics: what your video factory produces and what works

How to read the «Statistics» section of Content Factory. In this lesson: • production: vi...

Grow a new channel: auto-Shorts from every long video

How Content Factory cuts Shorts from every long video to bring it its first viewers. In t...

Social networks: connecting YouTube, TikTok and Telegram

The “Social networks” section is where publishing platforms are connected. YouTube — the “Connect” button, sign in to your Google account and allow video uploads. You can connect several channels. TikTok — sign in via TikTok. Telegram — add our bot as an administrator of your channel and enter the channel's @name or ID. Every channel has a “Comment” field (up to 300 characters) — a note for yourself, e.g. “main, news” or “test”. You can disconnect a channel at any time. Once connected, a channel can be chosen when publishing a video, in “Auto-publishing” and in the project's autopilot settings.

Open as a separate page →

Auto-publishing and autopilot

Auto-publishing (the planner) is a publication calendar: you see what goes where, can add a post manually (video or photo), reschedule or cancel. Time is in Oslo time (Norway), UTC+1/+2 depending on the season. Autopilot is fully automatic project operation. Enable it in the project settings (voice and language must be set): 1. “What to search” — a description of topics; the AI searches the internet for fresh topics. 2. Schedule: “at set times of day” (weekday and time checkboxes in 30-minute steps; the video is prepared N hours before publishing) or “at equal intervals” (e.g. every 30 minutes around the clock). 3. Social networks to publish to. The system invents a topic, writes the script, makes the video and publishes it. If 3 videos in a row fail, autopilot pauses — press “Resume” after fixing the cause. Costs come from your AI balance: watch the remaining balance on the “My finances” page.

Open as a separate page →

Autopilot with your own scenarios: plan and queue

Besides the “AI agent” mode (the AI finds topics itself), autopilot has an “Own scenarios” mode: you upload ready scenarios in advance and autopilot takes them in turn — on the project's schedule (every day, every N minutes, etc.). How to turn it on: project settings → Autopilot → “Where to take video topics from?” → “Own scenarios” (the button turns green) → save the autopilot settings. How to add a scenario: the “Add scenario” button (on the project page or in the autopilot settings). Enter a title, a topic (optional) and the text. You can add up to 3 ads with their own sound to each scenario — exactly like in a regular video (the [AD_N] marker or a start/middle/end position). You can add as many scenarios as you like. The plan is on the project page, the “Planned” tab: a table of rows — #, title, topic, character count, approximate video length and price. The full text is not shown. Buttons: “Edit” (text, title, ads) and a trash can. Order: the “Change order” button — drag a row up or down (or press ▲ ▼). #1 is created and published first. The order is saved immediately. Filters on the project page: “All”, “Planned”, “Published” (by day — the freshest day on top, then “Yesterday” etc.) and “Errors”. The project card shows how many videos are planned, published and failed. When the plan runs out, no new videos are created — add a scenario and it goes to the nearest schedule slot.

Open as a separate page →

What works: views analysis and a smarter autopilot

The “Statistics” page has a “What works” block. It takes your videos published on YouTube with statistics and calculates average views per day (since publication, so older clips do not win automatically). Then it compares them by parameters: project, mode, language, length, autopilot or manual, weekday and publication time (Oslo time). Next to each group is the deviation from the average in percent. Groups of 2+ clips are shown; while there are fewer than six clips the block warns that conclusions are tentative. The autopilot uses this data too: if a project has at least 5 videos published on YouTube with statistics, when searching for a new topic it gets a hint about what the most successful (and with 6+ clips — the weakest) videos were about, and leans toward what resonates with your audience without repeating what is already done. The “own scripts” mode is not affected — you set the scripts yourself.

Open as a separate page →

Shorts from every long video: how to get your first views

YouTube barely shows long videos from a new channel: it has no history yet, and a video can sit at zero views for weeks. Shorts, however, are shown to new viewers right away, even on an empty channel. So the best free way to promote a long video is to cut a few Shorts from it with a link to the full version. How to turn it on: My projects → project → Settings → “Autopilot” block → “Shorts from every long video”: 1, 2 or 3. Save the autopilot. What happens next, without your involvement: 1. The long video is published on YouTube. 2. The AI picks its most striking moments, 12–58 seconds each. 3. Vertical Shorts are built from them: the full-width frame on a blurred background, a large title at the top and “Full video on the channel” in the last seconds. 4. The Shorts go out on the same channel 2, 6 and 12 hours after the long video — not all at once. If TikTok channels are selected in the project's auto-publishing, they go there at the same time. 5. Each Short has a link to the full video in its description, and after publishing the channel leaves a comment with the same link. All Shorts appear in the Planner — you can move or delete them like any other publication. What you can do by hand for a bigger effect: in YouTube Studio set a “Related video” (the full video) for each Short and pin the comment with the link. The YouTube API does not allow this, only Studio does. Pinning comments and clickable links in descriptions are unlocked after channel verification (selfie or ID) — YouTube Studio → Settings → Channel → “Feature eligibility”. Shorts are cut only from videos built from a script with a voiceover. Each Short is a separate YouTube upload and counts towards the daily upload limit.

Open as a separate page →

3D videos on autopilot: a series without repeats

A project can publish 3D videos on a schedule by itself. Open the project → “Autopilot” tab → “Enable autopilot” → “Where to take topics from” → “3D videos”. Choose a world and write the series topic — a general one, for example “driving test questions: intersections, traffic lights, signs, turns” or “quick breakfast recipes”. For each slot the AI picks a new specific question within the topic and sees the titles of the project's previous videos, so the series does not repeat itself. The video duration is chosen right there (≈30–90 seconds); voice, language, format and music come from the project settings. The schedule (days and times or an interval) and channels work as for regular videos: the finished video appears in “My works” and the Planner by itself and is published to the selected channels. Rendering is charged to the project owner's AI balance (about 1–3 kr per video). If the balance is below 5 kr the slot is skipped, with a clear reason in the project's video list.

Open as a separate page →

Autopilot \"Chronicle\": videos from archives and from a folder

The autopilot can make chronicle videos on its own: find real footage in open archives (or take it from your folder), tell about it in the project's voice and publish on schedule. How to turn it on: project → "Autopilot" → source "Chronicle from archive". Settings: • Video type, archive and video length (from 30 seconds to 10 minutes). • Series topic — the agent searches archives for new videos by it. • A favourites folder (optional): if you choose a folder, each slot takes the next video from it in the order they were added, without repeats. When the folder's videos run out, the project searches archives by the topic. If there is no topic, the autopilot stops and says so in the slot's error message. • The project's voice is required: without it there is no narration. If the project has a HeyGen avatar and split screen, a blogger tells the story and takes half of the frame (costs more than a voice). How to add a video to a project's queue: in "Chronicle" press ☆ on a video and choose "To project queue" — the project gets its own folder the autopilot takes videos from. How the video is built: the AI watches the footage, picks short moments and writes text for each (no pauses), the original sound of the footage and the project's music are mixed in. The source and licence are added to the description. Important: if no suitable video is found for the topic, the slot is marked with an error with a clear reason — the autopilot does not take a random off-topic video. Voice and processing are charged from the AI balance, minutes from the plan limit.

Open as a separate page →

Plans, limits and the AI balance: what a video costs

How limits and finances work in Content Factory. In this lesson: • «My limits»: minutes,...

How much does an AI video cost? Plans and AI balance

How payment works in Content Factory: a plan plus a pay-as-you-go AI balance. In this les...

Earn 20% with the BelkaFood partner program

How the BelkaFood partner program works. In this lesson: • 20% of every subscription paym...

How long finished videos are stored, and how to rebuild them

How video storage works in Content Factory. In this lesson: • finished videos are stored...

AI balance, costs and the launch limit

All paid Content Factory operations (ElevenLabs voice, HeyGen avatar, AI script, images and animation) are charged from your AI balance — the same one used by the recipe AI assistant. You can top it up on the AI balance page. If there are not enough funds, the video will not start — you will see a clear message. The cost of each video (voice, avatar, AI) is shown on its page; summaries by project and period are in “My finances” and “Statistics”. Pexels stock is free. For orientation: voicing a 6-minute clip costs roughly 5–8 kr. Launch limit (may be enabled by the administrator): no more than N videos per sliding window of M minutes per user — protection against endless or mistaken launches. When the limit is reached you will see in how many minutes the next one can be started. Autopilot is not restricted by this limit (it has its own protection — a pause after 3 failures). The queue is shared: rendering runs one video at a time, the rest wait their turn. The AI balance is MONEY (the real cost of the chosen mode: stock is cheap, avatar/studio cost more) — it is not the same as your tier\u2019s monthly minutes quota for finished video (see the \u201cTariffs and limits\u201d article and the \u201cMy limits\u201d section on the left) \u2014 these are two independent counters: even with a full AI balance a video will not be built if the monthly minutes quota is exhausted, and vice versa.

Open as a separate page →

FAQ and troubleshooting

The video does not start. Check: is there enough money on the AI balance; is the launch limit enabled (the message shows the wait time); does the script exceed the word limit; is a voice selected. The video is “in the queue” for a long time. Rendering runs one video at a time; a 6-minute clip takes about 10–15 minutes together with voiceover and footage selection. Status “Error”. The reason is shown on the video page. Common ones: footage could not be found (rephrase the topic), out of money, a temporary network failure (press “Try again” — already paid voiceover is not paid for again). Footage is off-topic. Stock videos are matched by keywords; refine the topic or use “Full AI render”. You do not want subtitles. Untick “Subtitles on video”. Music is too loud — lower the volume in the project (default 0.12). The ad landed in the wrong place. For own-sound ads choose “By marker” and put the marker where you want it in the script, or choose a fixed position. Interface language. The site language is switched in the header; all Content Factory sections are translated into Russian, English and Norwegian.

Open as a separate page →

Tariffs and limits

The Content Factory has two independent gates: the money AI balance (see the \u201cAI balance, costs and the launch limit\u201d article) and your tier\u2019s monthly quota of MINUTES of finished video \u2014 both must allow it before a video is built. Free — included in Belka Food site Pro (after your application is approved, see “How to get access”): 30 minutes of finished video per month, every clip carries a \u201cMade with belkafood.no\u201d watermark in the corner. Only the cheap modes are available (stock footage and the trigger card), no talking avatar / studio / full AI render. Paid tiers \u2014 Starter, Pro, Business: each has its own monthly minutes quota, its own money budget, its own project limit, its own limit of connected accounts (YouTube+TikTok+Telegram combined) and its own daily publication limit \u2014 no watermark. Pro and Business additionally unlock the talking avatar, studio mode, full AI render, the \u201cMy films\u201d section and the button to launch a ready project straight from \u201cYouTube niches\u201d. Business renders get top priority in the shared queue, Pro gets medium priority, Starter and free access render as usual (these clips simply wait their turn if the server is busy with higher-priority ones). The exact current numbers (price, minutes, budget, project/account/publication limits for each tier) always live on the Content Factory pricing page \u2014 the numbers can change, so they are deliberately not repeated here. How much of your monthly quota you have already used (minutes, money, projects, accounts, publications) is in the \u201cMy limits\u201d section on the left, with a progress bar for each item. Autopilot requires an active paid tier \u2014 without one, if the subscription ended or was cancelled, autopilot pauses itself with a clear reason; building clips by hand is still possible (within free access, or until the limit is reached).

Open as a separate page →

Monthly project budget and protection from overspending

Project settings have a “Monthly project budget, kr” field. When the project's spending for the calendar month (voiceover, avatar, images — everything visible in “My finances”) reaches this amount, the autopilot stops creating new clips and a manual launch shows a limit message. Empty — no limit. The budget resets at the start of a new month; you can raise it in project settings at any time. In addition the administrator can set a general monthly spending limit per user and turn on an emergency pause: while paused, the autopilot, scheduled publications and launching new videos are temporarily stopped (renders already running are finished). Alerts: if a project's autopilot is paused (3 clips in a row failed), a clip failed to build or the budget is exhausted — the administrator (and you, if Telegram is linked) gets a message with the reason. Identical repeat messages come no more often than once in a few hours.

Open as a separate page →

How long finished videos are stored

A finished video file is stored for 30 days. The date is shown on the video page: “The file is stored until …”. Three days before deletion you get a notification in the bell. Download the videos you want to keep. Nothing is deleted without that notification. What is deleted: only the video file itself. The script, voiceover, thumbnail, settings and statistics remain, so you can rebuild the video with the “Rebuild” button on its page. Rebuilding renders the video again and uses minutes and balance like a new generation. What is kept as long as needed: • videos waiting to be published in the Planner; • videos in the project's “Ready to publish” queue (if “Publish ready videos first” is on); • videos from the “Examples” section. Published videos stay on YouTube, TikTok and Telegram — deleting our copy of the file does not affect them.

Open as a separate page →

Content Factory: the full course — every lesson in one video

Every Content Factory lesson in one video — from your first video to earning. Watch it str...