Edit Anything: Experimental LTX-2.3 Video Editing LoRAs for ComfyUI

news

Alissonerdx releases Edit Anything, an Apache-2.0 open-source video editing LoRA for LTX-2.3, featuring motion transfer, no-reference multitask editing, and reference video-to-video — all through ComfyUI BFSnodes.

On June 30, 2026, Alisson Pereira dos Anjos (@Alissonerdx) released Edit Anything, an experimental set of LTX-2.3 (22B) video editing LoRAs licensed under Apache 2.0. The project includes three training tracks for different video editing tasks, all designed to run through ComfyUI with the BFSnodes custom node set.

Heads up. These LoRAs are research experiments. They are far from production-ready and will fail on many inputs. They are released for the community to play with and break, not as a finished tool.

Three Tracks

1. Edit Anything v0.1 — Motion Transfer LoRA

Trained in two stages: image-only pretraining on ~30,000 image edit pairs, followed by video fine-tune with first-frame conditioning. Enables motion transfer — replace the first frame of a guide video with an edited still and the model propagates the new appearance through the rest of the clip. Available in r128 and r256.

Limitations: hard scene cuts break it, fast motion degrades to blur.

2. No-Reference Multitask LoRA (Rank 256)

Prompt-driven editing without a reference image. Supports object insertion, replacement, removal, and style transfer. Raise CFG to 6–8 for harder prompts.

3. Reference Video-to-Video (Ref V2V) — IC-LoRA

Transfers identity, style, or motion from a reference video. Requires BFSnodes to load sidecar modules.

ComfyUI Integration

All three tracks run through BFSnodes, which provides dedicated nodes: LTXVEditAnythingApply, LTXVEditAnythingModuleLoader, and a looping sampler for handling long videos in chunks.

A ready-to-use workflow is available for motion transfer and reference V2V:

Edit Anything: Experimental LTX-2.3 Video Editing LoRAs for ComfyUI | ComfyUI Wiki