LTX-2.3_ICLoRA_Union_Control (V2V)

[javawock7618]

I tried V2V using the LTX-2.3_ICLoRA_Union_Control workflow. I get very good results with realistic images, but when I use cartoon or comic-style images, realistic textures are somehow added. Is there a way to resolve this?

I will attach the original image and the first frame of the video generated with V2V, respectively.

[spadepsycho123gtr-jpg]

Hey, quick question 👋

Which ComfyUI version are you currently using (and branch if you know, like stable_ltx or main)? Also which version of ComfyUI-LTXVideo?

I’m running into this error with IC-LoRA / Union IC-LoRA:

LTXVAddGuide.append_keyframe() got an unexpected keyword argument 'causal_fix'

It crashes as soon as I run the workflow.

Could you share:

• your ComfyUI version
• LTXVideo node version (or when you last updated)
• whether IC-LoRA is working for you

If possible, can you also share your workflow JSON or node setup screenshot?

Trying to figure out which version combo is stable 🙏

[javawock7618]

ComfyUI 0.18.1 + ComfyUI_frontend 1.39.19
ComfyUI-LTXVideo is the latest version.

My workflow is published on Civitai:
https://civitai.com/models/2479744/ltx-23-video-to-video-distilled-gguf-workflows

IC-LoRA works effectively for realistic V2V and produces good results. However, when changing the style from realistic to cartoon, it results in the kind of rendering issues described above, which is problematic.

[javawock7618]

Lowering the LoRA strength from 1.0 to 0.5 (as in the example workflow) fixed the issue. Is 0.5 actually the intended/normal value?