RuneXX

/

LTX-2.3-Workflows

like 835

Is there a workflow that combines v2v (video-to-video) and i2v (image-to-video)?

"(i2v2v / iv2v)"

For example:

I upload a 5-second video clip (a woman is talking in a café).

Then, I upload a single image of another woman (especially a clear face shot).

Now, I want to replace the woman in the café video with the woman from the uploaded image.

The video should keep the same story, motion, and scene, but only the character (face/identity) is replaced.

Is this kind of workflow possible in LTX 2.3 using ComfyUI?

Alternatively, is it possible to mask only the woman in the café video (using a masking/inpainting approach in v2v) and replace just that region with the woman from the uploaded image?

Alternatively, I’d like to experiment by inserting an uploaded image (for example, even an animal) into the masked region of the first frame of the video and see it propagate through the clip.

In any case, is image-to-video-to-video (i2v2v / iv2v) workflow possible?

https://www.youtube.com/watch?v=HBp03iu7wLA

https://civitai.com/models/2027766/bfs-best-face-swap

https://huggingface.co/Alissonerdx/BFS-Best-Face-Swap-Video/tree/main/workflows

I’ve seen the workflows provided in these links, but they seem to focus mainly on face swapping.

What I’m looking for is a cleaner, more optimized workflow—something minimal and efficient, similar to the “RuneXX” style.

My ideal workflow would be:

Upload an image (it could be an animal, a person, or an object).
Upload a video, with a simple node that allows masking of a specific region.
Combine them cleanly using a streamlined “RuneXX”-style workflow.

In short, I want a minimal, well-structured pipeline that replaces only the masked region in the video with the uploaded image—not just face swapping, but full subject replacement.