24h GitHub PR Update
24h new pull request roundup with 14 new PRs across 3 tracked repos.
https://github.com/Comfy-Org/ComfyUI/tree/master
- #13327 fix: pin pyisolate to 0.10.1 in requirements.txt by @pollockjj: Fixes pin pyisolate to 0.10.1 in requirements.txt to improve reliability in production workflows.
- #13325 feat: Context Windows sampling with LTX2 models and IC-LoRa guides by @drozbay: Adds context windows sampling with ltx2 models and ic-lora guides to expand practical workflow capabilities.
- #13324 feat: process isolation for custom nodes via pyisolate by @pollockjj: Adds process isolation for custom nodes via pyisolate to expand practical workflow capabilities.
- #13322 range type by @jtydhr88: Introduces range type support, enabling cleaner parameter-bound inputs in node workflows.
- #13321 fix: nodes_save_ply/npz import from comfy_api_sealed_worker by @pollockjj: Fixes nodes_save_ply/npz import from comfy_api_sealed_worker to improve reliability in production workflows.
- #13320 feat: process isolation for custom nodes via pyisolate by @pollockjj: Adds process isolation for custom nodes via pyisolate to expand practical workflow capabilities.
- #13318 Fix nested tensor noise mismatch in CFGGuider.sample by @djdarcy: Fixes nested tensor noise mismatch in cfgguider.sample to reduce runtime errors and edge-case failures.
- #13317 Support Ace Step 1.5 XL model by @comfyanonymous: Adds support for Ace Step 1.5 XL model, enabling broader model usage.
- #13316 fix: pin SQLAlchemy>=2.0 in requirements.txt by @octo-patch: Fixes pin sqlalchemy>=2.0 in requirements.txt to improve reliability in production workflows.
- #13315 format blueprint by @jtydhr88: Refines format blueprint for cleaner and more predictable project structure.
- #13314 Add support for small flux.2 decoder by @huemin-art: Adds support for small flux.2 decoder, widening model/workflow compatibility.
- #13313 Ace step empty latent nodes follow intermediate dtype by @comfyanonymous: Aligns Ace Step empty-latent nodes with intermediate dtype handling, reducing precision mismatch issues.
https://github.com/shootthesound/comfyUI-Realtime-Lora
- no new pull requests in last 24h
https://github.com/ostris/ai-toolkit
- #782 Fix weighted timesteps being ignored by @asaf92: Fixes weighted timesteps being ignored to reduce runtime errors and edge-case failures.
https://github.com/AkaneTendo25/musubi-tuner/tree/ltx-2
- no new pull requests in last 24h
https://github.com/ArtDesignAwesome/ai-toolkit_BIG-DADDY-VERSION
- no new pull requests in last 24h
https://github.com/kijai/ComfyUI-KJNodes
- #600 Fix VRAMdebug cleanup ordering and add guarded CUDA synchronize by @xmarre: Fixes vramdebug cleanup ordering and add guarded cuda synchronize to reduce runtime errors and edge-case failures.
https://github.com/Lightricks/LTX-Desktop
- no new pull requests in last 24h
https://github.com/Lightricks/ComfyUI-LTXVideo
- no new pull requests in last 24h
https://github.com/Lightricks/LTX-2
- no new pull requests in last 24h
https://github.com/wildminder/awesome-ltx2
- no new pull requests in last 24h
https://github.com/modelscope/DiffSynth-Studio
- no new pull requests in last 24h
https://github.com/alvinunreal/awesome-autoresearch
- no new pull requests in last 24h