Frameforge

• 

  A vibe video platform for everyone

• 

  Insert your idea and generate script

• 

  Create Audio with a variety of voices, moods

• 

  Create your first scene

• 

  Generate all frames, adjust camera angles, transitions

• 

  Generate Thumbnails and Captions

• 

  Review master render and upload to your channel

Inspiration

Every creator has stories worth telling. A teacher wants to explain complex physics. A historian has fascinating anecdotes. A parent discovered a parenting hack. But turning these ideas into engaging video shorts requires skills most people don't have: video editing, motion graphics, voiceover recording, thumbnail design.

What it does

FrameForge acts as a full-stack production studio in your browser. It takes a raw idea or an inspiration image and guides the user through five distinct stages of production:

1. Ideation & Scripting: Generates viral hooks, scripts, and performs deep logic analysis.
2. Audio Synthesis: Converts scripts into multi-lingual, emotive voiceovers.
3. Visual Synthesis: Generates storyboard scenes (images) and animates them into video clips using Veo.
4. Post-Production: Generates thumbnails and burns in stylized captions/subtitles.
5. Distribution: Compiles assets and uploads directly to YouTube as a private draft.

How we built it

The application uses a "5-Gate" Workflow. The user cannot proceed to the next gate until the current stage is approved. This ensures quality control at every step.

Gate 01: Script (The Narrative)

• Input: Topic, Tone, Duration, or an Inspiration Image.
• Process: Uses Gemini 3 Pro to write a script with specific timing constraints. It creates a Hook, Body, and CTA.
• Feature: "Deep Logic Scan" uses the model's thinking capabilities to critique the script for logical fallacies or pacing issues.

Gate 02: Voice (The Sonic Identity)

• Process: Uses Gemini TTS to generate audio.
• Feature: Supports "Scene-Aware Translation". If a target language (e.g., Spanish) is selected, the system translates the script scene-by-scene before generating audio to ensure the subtitles match the visuals later.

Gate 03: Visuals (The Eye)

• Process: Generates 9:16 vertical images for every scene defined in the script.
• Feature: Consistency Checks. It allows users to define a "Character/World Definition". The system flags images that deviate from the established style.
• Animation: Uses Google Veo to transform static scene images into 720p videos.

Gate 04: Post-Production

• Process: Generates a high-contrast thumbnail.
• Captions: Renders captions directly onto the canvas (Classic, Karaoke, or Minimal styles) synchronized with the TTS audio.

Gate 05: Master

• Process: Real-time preview of all assets combined.
• Output: Download as .zip or upload directly to YouTube via OAuth.

Challenges we ran into

Not having known anything about content creation, image and video generation it is extremely difficult to understand whether I am right or wrong. Only after testing it with a number of beta users, I could get real feedback and work on it. Key challenge I ran into is with animating images and transitions. Integrating Veo helped a lot, but it is expensive so need to keep costs in mind

Accomplishments that we're proud of

The fact that we can now enable anyone who is interested in generating content, fact check it, analyse script with contrarian viewpoints is great thing for society.

What we learned

What's next for Frameforge

• Take it from MVP to production ready rollout.
• Implement professional design and UX.
• Offer freemium option to onboard users and give them enough credits to create at-least 2 60 second videos to get them hooked to the platform
• Add Analytics, more integrations with social platforms like Instagram, TikTok etc
• Add support of other models, leverage cost effective options

Built With

• python
• react