#OpenSourceHiggsfield closes: five last-hour ComfyUI builds
Entries for ComfyUI's #OpenSourceHiggsfield challenge closed at 9 AM PT on Nov 3, and the hashtag filled with last-minute submissions. Winners are due Nov 10. These five stood out for working video makers: a 3D previs tool, a kinetic-type engine, a MiniMax H3 film studio, a song-sectioned music-video tool and a take-based shot manager.
Comfy Previs by sbobyn (MIT). You block out a scene in 3D in the browser with three.js, give each character and prop a 'cast card' with reference images, and frame shots with keyframed cameras, a gamepad or an iPhone used as a handheld camera. Blockouts become storyboard stills and then AI video on Comfy Cloud: LTX 2.5 from key frames, Seedance 2.5 or Kling O1 edits of the blockout clip, or Wan VACE on your own ComfyUI. Per the README, it shows a cost estimate before each take, nothing paid runs unless you switch it on, and an MCP connector lets you direct scenes from Claude Code or Claude Desktop. There is a read-only live demo.
MOTNGRFX by okkazoo (Apache-2.0). A ComfyUI node pack plus browser app for kinetic-type motion graphics using only open weights. Per the creator, Gemma 4 plans the shots, fonts and music brief, LTX 2.5 makes the footage along with a subject matte and a clean plate, so letters can pass in front of and behind the subject, and ACE-Step produces the soundtrack. The README lists 12, 16 and 24 GB graphs, and the downloads are large: about 78 GiB for the 12 GB tier.
IV Studio by Matthew Hollis (MIT). A desktop app that goes from idea to script, storyboard, filmed shots and a finished short using MiniMax H3 on your own ComfyUI, or on your own Runpod if you don't have a GPU.
DropShift by debaa98. It takes a performance clip, the song and a style for each section, then produces a music video where the look changes at every section boundary. The repo went up minutes before the deadline and has no licence file yet, so treat it as a demo for now.
NodeTool (AGPL-3.0). An existing agent-first creative workspace that its team entered with a filmmaking flow: each shot gets a still and a camera move from a versioned preset library, renders run on Comfy Cloud with pinned ComfyUI workflows, and every render is kept as a take you can reopen and edit as a graph.
Our take: the strongest entries treat the video model as one step in a director's tool rather than the whole product. Previs, beat-synced type and song-structured looks are the jobs Higgsfield-style presets handle for you, rebuilt here in the open. Still no public reply from Higgsfield to the cease-and-desist story. We'll cover the winners on Nov 10.
Related tools on ReelStack
Skills & repos in our directory
- Comfy Previs · Repo, Generation
- MOTNGRFX · ComfyUI, Motion graphics
- IV Studio · Repo, Generation
- YouTube AI Motion Graphics · Repo, Motion graphics
- OpenField · Repo, Generation
- OpenKatana · ComfyUI, Motion graphics
More from the Lab
- #OpenSourceHiggsfield runs to Nov 3, winners Nov 10: four LTX builds to know
- #OpenSourceHiggsfield final day: six open studios to try
- OPENjutsu: Comfy open-sources a character-swap app on MiniMax H3, and a fully local fork follows within hours
- Body Double: recast the performer in a take with open weights in ComfyUI
Topics: ComfyUI · Higgsfield · Audio, voice & music · Open-weight video & audio models · Motion graphics skills