Jutsu Studio#OpenSourceHiggsfield
Open Source Higgsfield, every project on GitHub.
In October 2026 ComfyUI challenged builders to rebuild Higgsfield's apps in the open. These are the 100 entries worth opening, from OPENjutsu's character swap to one-tap effects, camera re-shoots and whole studios on open-weights models, plus the older alternatives. Stars as of October 31, 2026. The same list is a GitHub repository you can star or add to.
Most starred.
Challenge entries.
| Project | What it does | Stars | License |
|---|---|---|---|
| OPENjutsuD3voz/openjutsu | The Comfy team's open-source Genjutsu-style character and object swap: drop a video and a photo, press Generate. Also changes hair or clothes, removes objects or moves the person to a new place. | 651 | MIT |
| ComfyArtjtydhr88/ComfyArt | Free design canvas on your own ComfyUI: an infinite tldraw canvas with a design agent; image, video, audio and 3D run locally. | 405 | MIT |
| ComfyUI-Continuityroadmaus/ComfyUI-Continuity | One node for AI video and stills: write a prompt, attach media with @ and press Render; six model families, all local through ComfyUI core. | 138 | MIT |
| OpenFieldwexloops/openfield | Studio for video, stills and 3D on your own GPU: every app is a ComfyUI workflow with a screen on top, open weights only. | 116 | none |
| Open Source Canvasraph651/open-source-canvas | Node canvas that runs ComfyUI workflows. | 89 | MIT |
| comfy-openjutsuComfy-Org/comfy-openjutsu | Official clonable character-swap app and development guide from Comfy: upload a clip and a character image, get a 5 to 15 second part back with the character in it. | 68 | MIT |
| CreatorOS StudiothePloggy/CreatorOS-Studio | Self-hosted studio for one recurring character, from a one-line brief to a finished reel: camera moves, effects, talking avatars, motion transfer and a video editor. | 42 | Apache-2.0 |
| agentYszprivate/agentY | AI agent inside ComfyUI: describe the result and it picks or builds the workflow, runs it and puts the output on your canvas; any LLM, fully local with Ollama, plus a built-in image, video and 3D viewer, Atlas worlds and Gaussian splats. | 39 | MIT |
| H3 RecastBambushu/h3-recast | Swaps up to three people in an existing clip in one pass with MiniMax H3 and a BFS LoRA, keeping the motion, set and audio; optional FaceRefine lip pass, 16 GB VRAM and 64 GB RAM. | 36 | MIT |
| Body Doublelucataco/open-source-higgsfield | Same take, new cast: swaps the actors in a clip with open-weights models only. | 25 | Apache-2.0 |
What happened.
October 16
ComfyUI announced the challenge: Higgsfield had promised to open-source its platform and hadn't, so Comfy and its partners put up $25,000 for open-source builds.
October 26
The Comfy team shipped its own entry, OPENjutsu, an open-source Genjutsu-style character swap, and a clonable app template.
October 29
ComfyUI said Higgsfield had sent it a cease and desist asking it to shut the challenge down.
October 30
Submissions closed. Entries were found through the #OpenSourceHiggsfield tag on X and GitHub search; empty repositories and clones are left out.
Character swap (Openjutsu).
Open-source takes on Higgsfield Genjutsu: replace the person in a video with a character from one image and keep the camera, motion and sound.
| Project | What it does | Stars | License |
|---|---|---|---|
| OPENjutsuD3voz/openjutsu | The Comfy team's open-source Genjutsu-style character and object swap: drop a video and a photo, press Generate. Also changes hair or clothes, removes objects or moves the person to a new place. Launch post | 651 | MIT |
| comfy-openjutsuComfy-Org/comfy-openjutsu | Official clonable character-swap app and development guide from Comfy: upload a clip and a character image, get a 5 to 15 second part back with the character in it. | 68 | MIT |
| H3 RecastBambushu/h3-recast | Swaps up to three people in an existing clip in one pass with MiniMax H3 and a BFS LoRA, keeping the motion, set and audio; optional FaceRefine lip pass, 16 GB VRAM and 64 GB RAM. Launch post | 36 | MIT |
| Body Doublelucataco/open-source-higgsfield | Same take, new cast: swaps the actors in a clip with open-weights models only. Launch post | 25 | Apache-2.0 |
| Swapjutsujutsuhq/swapjutsu-seedance-character-swap | Free Mac app that swaps a character into any video: prepares the clip on the Mac with Vision, Core ML and MLX, then renders on Seedance 2.5 through Comfy or on your own ComfyUI with Wan 2.2 Animate. Launch post | 6 | MIT |
| Open Jutsu Studiolvsas/OpenJutsuStudio | Video studio that puts your characters into real footage on your own ComfyUI, reopens an H3 take to continue the story and exports a vertical before/after reel. Launch post | 3 | Apache-2.0 |
| ComfyUI H3 one-click swapD3voz/comfyui-h3-oneclick-swap | ComfyUI nodes for MiniMax H3 character and object swaps with tracked-person matching. | 1 | MIT |
| Kagamidemoncase/kagami | Open-source version of Higgsfield Genjutsu. Launch post | 1 | MIT |
| openjutsu (SydFloyd)SydFloyd/openjutsu | Genjutsu alternative on ComfyUI with open-weights models only. | 1 | Apache-2.0 |
| openjutsu (Ex8-ca)Ex8-ca/openjutsu | Genjutsu Motion Transfer alternative running MiniMax H3 locally. | 0 | MIT |
| OpenJutsu (fetchloom)fetchloom/OpenJutsu | Web-based video-to-video studio inspired by Genjutsu, self-hostable, with motion transfer. | 0 | MIT |
| openjutsu-catSvetimFM/openjutsu-cat | Openjutsu-style motion transfer driven by depth performances from a Kinect or any video. | 0 | MIT |
| OSH Character StudioLLannnn/OSHCharacterStudio | ComfyUI replacement for Genjutsu character work. | 0 | GPL-3.0 |
| Utsushiregiellis/utsushi | Video studio for your own ComfyUI, written in Go: motion transfer, object swap and style. | 0 | MIT |
All-in-one studios.
Higgsfield-style apps with many tools behind one screen: effects, camera moves, characters and editing, on your own GPU or Comfy Cloud.
| Project | What it does | Stars | License |
|---|---|---|---|
| OpenFieldwexloops/openfield | Studio for video, stills and 3D on your own GPU: every app is a ComfyUI workflow with a screen on top, open weights only. Launch post | 116 | none |
| CreatorOS StudiothePloggy/CreatorOS-Studio | Self-hosted studio for one recurring character, from a one-line brief to a finished reel: camera moves, effects, talking avatars, motion transfer and a video editor. | 42 | Apache-2.0 |
| agentYszprivate/agentY | AI agent inside ComfyUI: describe the result and it picks or builds the workflow, runs it and puts the output on your canvas; any LLM, fully local with Ollama, plus a built-in image, video and 3D viewer, Atlas worlds and Gaussian splats. Launch post | 39 | MIT |
| OpenKatanark0neki/OpenKatana | Agent-driven video editing with a studio UI, ComfyUI Apps, a CLI and an MCP server. Launch post | 17 | Apache-2.0 |
| Director's Cutleonardofrito/Directors-Cut | Local asset library and reference-to-video studio for ComfyUI. Launch post | 15 | MIT |
| OpenSourceHiggsfieldkevin9327/OpenSourceHiggsfield | Three-in-one cinema and video studio on open weights. | 9 | MIT |
| Comfieldsynthetic-ape/comfield | Higgsfield-style creative studio for AI images and video on ComfyUI. Launch post | 6 | MIT |
| HyperKut StudioMajoorWaldi/HyperKut-Studio | Local ComfyUI video studio for creating, remixing and editing video. | 5 | MIT |
| BacklotArfolomew/backlot | Open-source Higgsfield that runs on a 16 GB card with local open-weights video. Launch post | 4 | MIT |
| EasyComfeesintetizacom/easycomfee | Higgsfield-style studio on ComfyUI: image, video and LTX VFX. Launch post | 3 | MIT |
| FrameForge StudioRahmelJackson/frameforge-studio | Creative studio inspired by Higgsfield, powered by ComfyUI workflows. Launch post | 3 | MIT |
| OpenIllusionGETHUK-BIGEMPEROR/OpenIllusion | Turns ComfyUI workflows into one-click templates: pick one, drop in a photo, clip or sound, press Run. Launch post | 3 | MIT |
| Cheapsfielddougliell/Cheapsfield | Fully local, open-weights creative studio for Windows with Higgsfield-style apps. | 2 | Apache-2.0 |
| MarcoStudioNeoMarcoPolo/marcostudio | 13 image and video tools on your own ComfyUI with open-weights models. | 2 | MIT |
| Rosesfieldrossaai/rosesfield | Open-source Higgsfield alternative on ComfyUI with 25+ apps for image and video. | 2 | MIT |
| Opensfieldcodyclack/opensfield | Higgsfield rebuilt on ComfyUI: 146 tools in one app. | 1 | MIT |
| Susanonorioversoul/susano | Open-weights studio on ComfyUI with Openjutsu, Tsukuyomi, Cinematic and Influencer apps. | 1 | MIT |
| CineStudioMachineLearningAmateur/cinestudio | Scene control for renders made with LTX 2.5 on ComfyUI. Launch post | 0 | MIT |
| DariusNiutonian/Comfyui_Darius_OS | ComfyUI workspace with a hosted interactive demo. Launch post | 0 | none |
| HaggsIBFolding/haggs | Film studio on Comfy from script to screen, with every shot and its price in view. | 0 | MIT |
| Hildegardbaney75/hildegard | AI studio on your own GPU: video, voice, characters and 3D. | 0 | AGPL-3.0 |
| Hoodsfield AIdesignn2future/hoodsfield-ai | Local-first AI film studio on ComfyUI: idea in, finished video out. Launch post | 0 | Apache-2.0 |
| Liberty CutE5EnclaveInc/liberty-cut | Cinematic studio built only on open weights (LTX-Video). | 0 | MIT |
| MontagioJubzh-29/montagio | Viral video effects and camera moves as an open-source Higgsfield alternative. | 0 | AGPL-3.0 |
| Vision Orchestratorstellartune-studio/vision-orchestrator | Chat with an LLM as the director while it grows a production blueprint: story, characters, materials, shots. Launch post | 0 | none |
Effects and VFX.
One-tap viral effects and plate-preserving VFX, the open version of Higgsfield presets.
| Project | What it does | Stars | License |
|---|---|---|---|
| Seamjdnichollsc/seam | Two shots in, one AI transition out: smoke, water and more. Launch post | 15 | MIT |
| SpEdEBurgstall-labs/ComfyUI-SpEdE | Click an object in a Gaussian splat, then remove, replace or restyle it. Launch post | 15 | AGPL-3.0 |
| Element ForgeZitroffi/comfyui-element-forge | AI VFX elements for your own footage: LTX-2.5 generates fire, lightning, smoke or particles, then Element Forge tracks, lights and layers them into the shot so they stay editable. Launch post | 6 | MIT |
| FreeFieldPhatStraw/freefield | Higgsfield-style viral effects as ComfyUI workflows: one photo in, one shareable clip out. Launch post | 4 | MIT |
| Pickupsthekhoma/pickups | Video app on top of ComfyUI: pick a shot and change the weather, the lighting or what is in the frame. | 3 | MIT |
| LiDAR Labchinesepowered/hack-comfy | LiDAR scan, clones, vanish and bullet-time effects that run on a laptop CPU. Launch post | 2 | Apache-2.0 |
| PlateLockTusharTechs/platelock | Plate-preserving generative VFX: add the effect, keep the original shot. | 2 | MIT |
| Forkfieldmarciano71/forkfield | Fork any video: Higgsfield-style effects where every clip keeps its recipe. | 1 | MIT |
| ConjureOrazen/conjure | One photo, one tap, a whole VFX shot: gold statue, inflate, disintegrate, earth zoom-out on FLUX and LTX. Launch post | 0 | none |
| FXjutsuaudriu/fxjutsu | Higgsfield-style VFX presets on your own footage with LTX-2.5. | 0 | none |
| MOTNGRFXokkazoo/MOTNGRFX | Node pack and browser app that turns a song, a few images or a short brief into a kinetic-type motion graphic, with 3D type landing on the beat, on open weights. Launch post | 0 | Apache-2.0 |
| One-take VFXKainotomic/one-take-vfx | Add disintegrate, levitate, fire, portal or time-freeze to a phone clip with LTX 2.5. Launch post | 0 | MIT |
Camera, reshoot and previs.
Re-shoot a clip from a new angle, plan camera moves in 3D, or check that a render followed the move.
| Project | What it does | Stars | License |
|---|---|---|---|
| Motion Previs Studionavarisun1982/comfyui-motion-previs-studio | 3D camera, character and motion previs, an open take on Higgsfield 3D. | 15 | MIT |
| 360 Scout to FilmAlvinvision/360-Scout-to-Film-Consistency | Camera-controlled 360 location references and multi-shot video inside ComfyUI. | 3 | MIT |
| ReAngleTekholms-LW/ReAngle | Re-shoot any video from a new camera angle with ComfyUI and LTX-2.5. | 3 | Apache-2.0 |
| TimeFlightEmmanuelRiveros/TimeFlight | Drone flights over real places that go back in time, from open geodata. | 3 | Apache-2.0 |
| One-shot Multicamapsolut/one-shot-multicam | Turn one selfie video into a multi-camera edit. Launch post | 2 | MIT |
| Hero Frame Rigmetababe-eth/hero-frame-rig | Change the focus of any clip after it is made. Launch post | 1 | Apache-2.0 |
| Open Higgsfield Reshootpyoonpyoon/open-higgsfield | A 3D virtual film set that re-shoots any photo from a new angle. | 1 | MIT |
| Comfy Dialylchen333/comfy-dial | Draw a camera move in 3D; it renders takes, measures the camera each one used and regenerates only the part that drifted. | 0 | MIT |
| Motion Guardjunco3/motion-guard | Checks AI video against its 3D motion reference before an expensive upscale. | 0 | Apache-2.0 |
| Open CinemaLaytonGott/open-cinema | Direct a shot, then change the lens afterwards: LTX-2 video plus physically based depth of field. | 0 | GPL-3.0 |
| previz2shotMVOlekhno/previz2shot | Blender grey playblast to a finished shot with FLUX.2 Klein and LTX-2.5. Launch post | 0 | MIT |
| Shot BoardArcSolver/shot-board | Write one scene, get six shots, pick a hero take and redraw the cuts from it. Launch post | 0 | MIT |
| Shot Directorbolttmade/shot-director | Free previs, then one real render: a camera-move studio for Comfy. | 0 | MIT |
| ShotChecksoft-jelly/shotcheck | CPU-only camera-motion checker: match, mismatch or inconclusive. Launch post | 0 | MIT |
| Toybox Directormahdi-salmanzade/toybox-director | Block a shot with toys on a 3D stage, then render it with open-weights LTX. Launch post | 0 | MIT |
Films, music videos and stories.
From a logline, a song or a document to a finished video.
| Project | What it does | Stars | License |
|---|---|---|---|
| Pickychenthedesignmaker/ComfyUI-Picky | Render a shot ten ways and keep the one you love. Launch post | 17 | GPL-3.0 |
| Mondo Music Videohodgemann/mondo-music-video | Music-video studio with an actor protocol and optional MCP tools. Launch post | 10 | Apache-2.0 |
| LeafReelEditorenbici/leafreel | Any document into a narrated explainer video in 8 visual styles. Launch post | 1 | Apache-2.0 |
| Onelinerthaohienhomes/oneliner | Type one line, get a cinematic trailer on Comfy Cloud, LTX and ffmpeg. | 1 | MIT |
| OpenReelsmithysw7/openreel | One logline in, a character-consistent short film out. Launch post | 1 | MIT |
| seriesmithbgstratt/seriesmith | A series config and markdown episode scripts in, finished episodes out: any shot on any model, on your own GPU or on Comfy's cloud. | 1 | MIT |
| Xenoformtysonfarrington-cmyk/xenoform | Song in, music video out, on your own GPU with open weights. Launch post | 1 | MIT |
| Cat TapesNemont3/cat-tapes | Lo-fi VHS cat home videos made locally with LTX. | 0 | MIT |
| Crew CutGiveMePrompts/crew-cut | A party game that makes a movie: phones become characters, friends direct scenes. | 0 | MIT |
| Hero's Journeyspiritform/herosjourney | The monomyth as a generative experience starring you. Launch post | 0 | MIT |
| Storyboard StudioDialloWill/storyboard-studio | Open rebuild of Higgsfield Cinema Studio for music-video storyboards. | 0 | MIT |
Ads and UGC.
Product ads, UGC-style clips and personalized spots.
| Project | What it does | Stars | License |
|---|---|---|---|
| ComfyFieldergophobian/comfyfield | Insert your product into proven viral short-form formats and post on schedule. Launch post | 1 | MIT |
| HiggsOpenkingofmit/ComfyUI-HiggsOpen | Rebuild of Higgsfield Diffuse: a topic in, a viral short video out. | 0 | none |
| Open Ad Studiomoonz-1/open-ad-studio | Open-source AI creative agency on ComfyUI. | 0 | MIT |
| Open Ads Studiojhoard/open-ads-studio | Open-weights rebuild of Higgsfield Ads Studio with label-exact, compliance-checked product shots. | 0 | Apache-2.0 |
| Reelmatchvidyabharadwaj03/reelmatch | Describe a campaign, drop in photos, get a UGC-style ad draft. | 0 | MIT |
| Urban Cutsjk-automations/urban-cuts | ComfyUI App Mode clone of Higgsfield Urban Cuts: a photo plus outfits or products. | 0 | Apache-2.0 |
Characters and datasets.
Lock a character look, build LoRA datasets, keep identities consistent.
| Project | What it does | Stars | License |
|---|---|---|---|
| comfyui-samefacefixed-42/comfyui-sameface | Custom node that measures where an image edit moved the parts of a face and moves them back, on open weights. | 6 | Apache-2.0 |
| Character LockZhuEvgeny/Character-lock-comfy | One reference image in, a locked character look out. Launch post | 1 | MIT |
| Boson0xMisterBailey/Boson | Your taste as plain files in git that every next generation reads. | 0 | MIT |
| Dataset ForgeToxicLizard99/Dataset-Forge | One face in, a full character LoRA dataset out. | 0 | MIT |
| Soul Person EnhanceSoussa4535/SoulPersonEnhance | Enhance each person in a photo separately, or only the ones you pick. Launch post | 0 | AGPL-3.0 |
Canvases, front ends and tools.
Node canvases, web UIs and helpers that put a friendlier face on ComfyUI.
| Project | What it does | Stars | License |
|---|---|---|---|
| ComfyArtjtydhr88/ComfyArt | Free design canvas on your own ComfyUI: an infinite tldraw canvas with a design agent; image, video, audio and 3D run locally. Launch post | 405 | MIT |
| ComfyUI-Continuityroadmaus/ComfyUI-Continuity | One node for AI video and stills: write a prompt, attach media with @ and press Render; six model families, all local through ComfyUI core. | 138 | MIT |
| Open Source Canvasraph651/open-source-canvas | Node canvas that runs ComfyUI workflows. Launch post | 89 | MIT |
| KubaNodeskubakubkub/ComfyUI_KubaNodes | Visual toolset for pictures made of parts: projection mapping with regions, per-region diffusion, a director window, and video and 3D scene tools. | 10 | GPL-3.0 |
| FreavyFrameDeptCo/freavy | Local-first node canvas for AI image and video on ComfyUI. | 6 | MIT |
| ComfyUI-Spooldoubletwisted/ComfyUI-Spool | Video timeline for ComfyUI: queue workflows per clip, review the generated takes and assemble a sequence. | 3 | GPL-3.0 |
| Justae FlowHomoLapis/justae-flow | ComfyUI inside the open-source EffectCraft editor. | 2 | GPL-3.0 |
| Ask Higgsymhaihq/hana-comfy-hack | Video-call a small animated guide that helps you pick the right model. Launch post | 1 | MIT |
| FrameCommanderwelski3/framecommander | Web UI for MiniMax H3 video through ComfyUI workflows on RunningHub. Launch post | 1 | MIT |
| RenderRepocrystalattice/CBI_RenderRepo | Local-first package manager for ComfyUI workflows: discover, inspect, install and open workflows while reusing the models and nodes you already have. Launch post | 1 | GPL-3.0 |
| Open Studio for Affinityviriato68/open-studio-for-affinity | Local AI editing inside Affinity, powered by ComfyUI. | 0 | MIT |
| openfield (Blender MCP)obaidiDev/openfield | The Higgsfield Blender MCP stack made local-first with a self-hosted gateway. | 0 | GPL-3.0 |
Beyond the challenge.
Older open-source Higgsfield alternatives on GitHub, most of them cloud front ends that need someone's API key.
| Project | What it does | Stars | License |
|---|---|---|---|
| Open-Generative-AIAnil-matcha/Open-Generative-AI | Formerly Open-Higgsfield-AI. The biggest web and desktop studio; cloud models through Muapi.ai. | 31,898 | MIT |
| open-higgsfieldwide-trace/open-higgsfield | Next.js web app with a hosted demo; needs a platform API key. | 5,121 | none |
| HeliosGenSegFault42/HeliosGen | Tauri desktop app with a Mac build; generations through kie.ai. | 3,220 | none |
| LTX DesktopLightricks/LTX-Desktop | Lightricks' own desktop app for running LTX video models locally. | 2,192 | Apache-2.0 |
| BeatDesignBeatAPI/BeatDesign | Local web workspace on Node.js; generations through BeatAPI. | 281 | Apache-2.0 |
| Kawarimijutsuhq/kawarimi-local-character-swap | Local character swap for Windows and Linux on Wan 2.2 Animate through ComfyUI. Made by the maintainers of this list. | 2 | MIT |
Stars are a popularity signal, not a quality one, and change daily. "None" means the repository has no license file, so its code isn't licensed for reuse even though it's public. For a hands-on check of the older alternatives, with whose API each one needs, read open-source Higgsfield alternatives on GitHub, checked.
FAQ.
What is #OpenSourceHiggsfield?
A challenge ComfyUI ran from October 16 to October 30, 2026: build open-source apps that do what Higgsfield's paid apps do, on the Comfy Developer Platform, for $25,000 in prizes from Comfy and partners including NVIDIA, RunPod and MiniMax. The suggested tracks were a cinematic studio, re-shooting a video from a new angle, and Openjutsu, a character swap. This page lists 100 of the entries.
What is OPENjutsu?
The Comfy team's own entry, released on October 26: an open-source, Genjutsu-style character and object swap for video at github.com/D3voz/openjutsu, with a clonable app template at github.com/Comfy-Org/comfy-openjutsu. With 651 stars on October 31, 2026, it is the most-starred project of the challenge.
What is the best open-source Higgsfield alternative on GitHub?
It depends on the job. For character swap, OPENjutsu (651 stars). For a design canvas, ComfyArt (405). For a studio with many tools, OpenField (116). The biggest older project, Open-Generative-AI (31,898 stars, formerly Open-Higgsfield-AI), is a front end for cloud models and needs a Muapi.ai key, while the challenge projects run open-weights models through ComfyUI.
Are they free to use?
The code is. Most projects run open-weights models such as LTX-2.5, Wan and FLUX through ComfyUI, which costs nothing per video on your own NVIDIA GPU. Run on Comfy Cloud, or with paid partner models, generations cost money. Check each repository for its license: a public repository without a license file is not licensed for reuse.
Do they run on a Mac?
Mostly not locally. Nearly all of them expect ComfyUI on an NVIDIA GPU with 12 to 24 GB of memory, or Comfy Cloud, which works from any computer with a browser. LiDAR Lab is one of the few that runs on a CPU. For character swap on a Mac, Swapjutsu prepares the clip on the Mac and renders on Comfy or on your own ComfyUI.
Why did Higgsfield send ComfyUI a cease and desist?
ComfyUI said on October 29, 2026 that Higgsfield had sent it a cease and desist the night before, asking it to shut the challenge down. Its own reading was that a closed platform is uncomfortable watching its paid apps rebuilt in the open in two weeks. The entries listed here are public repositories on GitHub.
Is this list made by Higgsfield or Comfy?
No. It is an independent list kept by Jutsu Studio, not affiliated with or endorsed by Higgsfield or Comfy. Swapjutsu, a challenge entry, and Kawarimi, in the last section, are Jutsu Studio projects.
How do I add a project?
Open a pull request on github.com/jutsuhq/open-source-higgsfield that adds it to projects.json with a one-line summary. The project needs a public repository with a README.