Browser-based video compositor built on WebGPU
GitHub: https://github.com/Sportinger/MasterSelects I built this Browser-based video compositor with a GPU-first architecture. No Canvas 2D in the rendering path — video textures go in as texture_external (zero-copy), compositing runs through a ping-pong WGSL shader pipeline, and export captures frames directly from the GPU canvas via WebCodecs. 39 GPU effects, 37 blend modes, nested compositions, keyframe animation with bezier curves, vector masks, 10-band live EQ, video scopes, and AI-driven editing via GPT function calling. 13 production dependencies. The compositor,…
What it does
In the maker’s words, at launch
GitHub: https://github.com/Sportinger/MasterSelects I built this Browser-based video compositor with a GPU-first architecture. No Canvas 2D in the rendering path — video textures go in as texture_external (zero-copy), compositing runs through a ping-pong WGSL shader pipeline, and export captures frames directly from the GPU canvas via WebCodecs. 39 GPU effects, 37 blend modes, nested compositions, keyframe animation with bezier curves, vector masks, 10-band live EQ, video scopes, and AI-driven editing via GPT function calling. 13 production dependencies. The compositor, all shaders, timeline, audio mixer, mask engine, and export pipeline are built from scratch. I'm a video artist, not a developer. Built this entirely with Claude. Things break, but when it works, it works. Chrome/safari with WebGPU required. Firefox and webGPU have problems.
Does the same job
all alternatives →- PBPhysically Based Rendering – Online Edition2018 · pbr-book.org · ▲297
- MFMy first WebGL demo, the first self-extracting PNG ever?2011 · demoseen.com · ▲101
- WFWebGPU FTW with Splat-TransformSep 2025 · github.com · ▲5
It's annoying having so many ML and GS training tools rely exclusively on cuda/nvidia . So for our open-source command-line gaussian splat converter/compressor, we decided to try WebGPU instead. It's working well so far and a single codebase runs on Linux, MacOS and Windows without too much fuss. This is mostly thanks to Google's dawn project <3. Eventually some of this could also run directly in the browser.
- WVWe've built a funny GitHub Wrapped video summary for developers2024 · rendley.com · ▲7
Hey everyone, my friend and I created a GitHub Wrapped video for developers to showcase their activity on the platform in a more fun way. The implementation is open-source (https://github.com/rendleyhq/github-wrapped-2024) if you're interested in how it was built. To get a bit geeky, we used masks, keyframe animations, luma key shaders to remove the background from the confetti video, and some other techniques. Also, everything is rendered directly in the browser!! I'm curious to hear your thoughts on the rendering performance, especially if you're running this on your…
- MAModular and Live-Reloading Shaders in WebGL2015 · youtube.com · ▲30
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 16d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 26d ago · cactuscompute.com
