A VFX and video stylization tool built around a simple premise: paint or edit a single keyframe, and EbSynth propagates that look across the rest of the shot. It is used for hand-drawn animation, rotoscoping, digital makeup, colorization, and visual retouching, and was notably used in Richard Linklater\'s Apollo 10½ and a Netflix Wednesday sequence . The key technical distinction is that the propagation does not use generative AI—it uses a texture-synthesis algorithm (PatchMatch-based) that only works with your footage and keyframes, preserving pixel-level control and avoiding the \"hallucinations\" of neural models . EbSynth 2 is browser-based (Chrome recommended), adds a timeline, layers, brushes, and real-time preview, and claims roughly 10x faster processing than the original desktop tool . An optional \"Generate Image\" feature uses Stability AI\'s SDXL 1.0 to create keyframes; if you use it, the output falls under SDXL license terms . Pricing: Free ($0, all core functions, 720p MP4 export), **Pro** ($20/month, up to 4K, PNG with alpha, priority processing), Studio ($200/month/seat, local Windows desktop app, footage never leaves your machine, NDA-friendly) . Made by Secret Weapons (Prague/Dover, Delaware), a small team founded around 2018 .
Pros & Cons:
Pros: The non-AI texture-synthesis approach is the core differentiator—it preserves the artist\'s painted detail at pixel level rather than generating approximations, which matters for stylized animation where fidelity to the keyframe is the whole point; the workflow is genuinely fast for the right task: paint one frame, let the algorithm handle in-betweens, instead of painting every frame by hand; EbSynth 2\'s real-time preview, timeline, and layers address the biggest complaint about the original tool (slow, batch-based iteration), making it feel like actual software rather than a research demo ; the free tier is fully functional with no time limit, only capped at 720p export, so you can evaluate the entire workflow before paying; Studio tier keeps footage entirely local, which is the right option for studios with NDA or privacy requirements ; production credits (Apollo 10½, Wednesday) are concrete evidence it works at professional scale .
Cons: It is not a general-purpose video editor—no timeline editing, no color grading suite, no audio tools; it does one thing (keyframe propagation) and expects you to bring the rest from elsewhere; the learning curve is real: keyframe selection, alignment with source frames, and dividing shots into independently moving parts require practice, and bad keyframes produce stretchy artifacts ; it struggles with motion blur, occlusion, and reflections, which often require external correction ; the \"Generate Image\" AI feature is optional but creates a licensing complexity: if you use it, your entire output video falls under SDXL terms ; the Studio tier at $200/month is expensive for individual artists, and there is no mid-tier between Pro ($20) and Studio; the browser-based workflow means uploading footage to their servers (deleted immediately after rendering, per the privacy policy), which some studios will not accept .
Who is it for?:2D animators and motion designers who want painterly or hand-drawn looks without painting every frame; VFX artists doing rotoscoping, digital makeup, cleanup, or colorization who need a fast 2D tracking shortcut; indie filmmakers and creative studios testing stylized looks on a budget (free tier is genuinely usable); production teams that need offline/local processing for confidentiality (Studio tier). Not suited for editors looking for a full NLE, teams that need AI-generated video from text prompts (this is not that), or anyone unwilling to learn the keyframe-driven workflow.