Starstruk Logo MarkStarstruk

v0.3.53

Comfy-Org/ComfyUIv0.3.53Aug 28, 2025by github-actions[bot]

AI Summary

This release introduces audio encoding capabilities with wav2vec2 and fixes critical crashes in EasyCache and LazyCache. It also adds preliminary support for Wan 2.2 S2V (Sound to Video) model.

Key Highlights

  • Implementation of wav2vec2 as an audio encoder model
  • Fix for EasyCache/LazyCache crash on tensor changes
  • WIP support for Wan 2.2 S2V model
  • Addition of DPM++ 2M SDE Heun (RES) sampler

New Features

  • wav2vec2 audio encoder
  • EasyCache and LazyCache implementation
  • Wan 2.2 S2V model support
  • LatentConcat node

Full Release Notes

## What's Changed
* Fix EasyCache/LazyCache crash when tensor shape/dtype/device changes during sampling by @blepping in https://github.com/comfyanonymous/ComfyUI/pull/9528
* Remove models from readme that are not fully implemented. by @comfyanonymous in https://github.com/comfyanonymous/ComfyUI/pull/9535
* Add models/audio_encoders directory. by @comfyanonymous in https://github.com/comfyanonymous/ComfyUI/pull/9548
* Implement wav2vec2 as an audio encoder model. by @comfyanonymous in https://github.com/comfyanonymous/ComfyUI/pull/9549
* Make AudioEncoderOutput usable in v3 node schema. by @comfyanonymous in https://github.com/comfyanonymous/ComfyUI/pull/9554
* Update template to 0.1.66 by @comfyui-wiki in https://github.com/comfyanonymous/ComfyUI/pull/9557
* Adding Google Gemini Image API node by @Kosinkadink in https://github.com/comfyanonymous/ComfyUI/pull/9566
* Update template to 0.1.68 by @comfyui-wiki in https://github.com/comfyanonymous/ComfyUI/pull/9569
* WIP Wan 2.2 S2V model. by @comfyanonymous in https://github.com/comfyanonymous/ComfyUI/pull/9568
* Negative audio in s2v should be zeros. by @comfyanonymous in https://github.com/comfyanonymous/ComfyUI/pull/9578
* Fix #9537 by @KohakuBlueleaf in https://github.com/comfyanonymous/ComfyUI/pull/9576
* Fixes to make controlnet type models work on qwen edit and kontext. by @comfyanonymous in https://github.com/comfyanonymous/ComfyUI/pull/9581
* Improve s2v performance when generating videos longer than 120 frames. by @comfyanonymous in https://github.com/comfyanonymous/ComfyUI/pull/9582
* Better s2v memory estimation. by @comfyanonymous in https://github.com/comfyanonymous/ComfyUI/pull/9584
* Add DPM++ 2M SDE Heun (RES) sampler by @Balladie in https://github.com/comfyanonymous/ComfyUI/pull/9542
* Add a LatentConcat node. by @comfyanonymous in https://github.com/comfyanonymous/ComfyUI/pull/9587
* Adjust flux mem usage factor a bit. by @comfyanonymous in https://github.com/comfyanonymous/ComfyUI/pull/9588

## New Contributors
* @Balladie made their first contribution in https://github.com/comfyanonymous/ComfyUI/pull/9542

**Full Changelog**: https://github.com/comfyanonymous/ComfyUI/compare/v0.3.52...v0.3.53