Fogsight: a free open-source AI tool that turns text prompts into animated explainers

What is Fogsight

Fogsight is a free, open-source AI animation generator built around one simple idea: you name a topic or concept, and it turns that abstract phrase into an animation with narration, visual elements and motion. The examples it shows are telling — bubble sort, the law of increasing entropy — subjects that are painful to explain in static slides and suddenly obvious once they move.

You do not lay out a timeline or draw keyframes. A large language model reads the topic, writes a script with scene direction and narration, and an animation engine renders it into a finished piece. Your job is closer to directing: describe what you want clearly, then watch the result appear.

If the first cut misses, you do not start over. Fogsight ships a language user interface (LUI) that supports multi-turn conversation, so pacing, wording and framing can all be adjusted by talking to it.

Main features

• Text to animation: enter a topic or concept and get an animation with narration, visual elements and motion that makes the abstract concrete.
• LLM-driven orchestration: the model handles the entire path from written script to finished animation, so you do not design animation details by hand.
• Language user interface (LUI): refine the result through natural-language conversation, round after round, until it matches what you had in mind.
• Multi-model compatibility: connect different LLMs, such as OpenAI GPT and Google Gemini, and switch according to your needs.
• Open source and hackable: the code lives on GitHub (https://github.com/fogsightai/fogsight), ready for self-hosting and further development.

How to use it

1. Get the code: clone the Fogsight repository and install the environment dependencies described in the documentation.
2. Add a model key: obtain an API key for the LLM you want to use, such as OpenAI or Google Gemini, and put it in credentials.json.
3. Enter a topic: type the concept in the interface or the command-line tool. The more concrete the better — for example, "explain in three minutes how bubble sort swaps elements".
4. Generate the animation: the system parses your input, the model writes the script, and the animation engine produces the video.
5. Refine in conversation: keep talking through the LUI, pointing at the specific segment you want changed and how.
6. Export and use: export the result as a video file or another supported format for lessons, short videos or presentation pages.
7. Try before installing: you can open the official site at https://fogsight.ai/ to see the output first and decide whether to self-host.

Where it fits

• Classroom teaching: turn algorithms and physical laws into animation so students can watch the process instead of imagining it.
• Science communication: build short explainers of scientific principles and replace paragraphs of text with moving images.
• Corporate training: animate technical workflows and business rules for onboarding and internal knowledge sharing.
• Content creation: produce early concept clips for short videos, ads or creative pitches to test whether an idea is worth pursuing.
• Self-directed learning: have students animate a difficult concept themselves; building it once usually teaches more than reading it ten times.
• Development: integrate the open-source code into your own product to add custom animation generation.

Limitations

• Some technical setup is required: self-hosting means dependencies and API keys, which can take a while the first time. If that is not your thing, start with the online version on the official site.
• Model costs are yours: the tool is free and open source, but calls to GPT, Gemini or other model APIs are billed by the provider and depend on the model and volume you use.
• Output quality tracks input quality: vague topics produce vague videos, and long scripts with several concepts usually need more rounds of adjustment.
• Visual style is bounded by the animation engine: this is not frame-by-frame animation software, so fine storyboard and art-direction control is limited.
• The project keeps moving: interfaces and configuration in an open-source repository can change, so treat the repository documentation as the source of truth.

FAQ

Is Fogsight free? The tool is free and open source and can be self-hosted. Model usage is separate — calls to third-party LLM APIs are charged by that provider.

Do I have to know how to code? Self-hosting requires configuring the environment and an API key. If you only want to see what it produces, try the official site first.

Which models can I use? It is built to work with multiple LLMs, including OpenAI GPT models and Google Gemini; check the project documentation for the current list and setup.

Can I use the animations commercially? The code is open source, but the output, the provider terms and third-party assets each have their own conditions — verify the licenses before commercial use.