Workflow - V2V inpainting with Inpaint-Masked-T2V Lora
V2V inpainting with Inpaint-Masked-T2V Lora
Mask an area you want to inpaint and prompt what you want to see.
Comes in 2 variants:
- Sam2 Masking : uses a point editor where you simply click and mark the area you want to mask for inpainting
- Sam3 Masking : just prompt what you want to mask ( for example second woman)
The masking variants both have their pro and cons. Sam3 is easy to use if its a clear object/subject to mask.
Sam2 is better if its more vague where you want to mask, for example in an area where there is nothing already (and hence not easy to prompt Sam3 for anything)
The workflows uses Masked Inpaint T2V loras from here https://huggingface.co/Alissonerdx/LTX-LoRAs
(NOT the R2V lora, a different workflow coming soon for that)
Feel free to try it out here:
https://huggingface.co/RuneXX/LTX-2.3-Workflows/tree/main/Video-2-Video/Inpaint
Hi, the SAM3 detection works perfectly, but the reference image using would be absolutely great. Is it possible to add it e.g. as image2 in LTXAaddGuideMulti node?
yes adding a reference lora wf version. Plus trying to see if the above workflow could also work with reference
Can we have like no reference image and text edit to like change moutth movementt in this? or say different words?
what if I give audio custom and then it edits out to what it will say? i have a video having 2 subjects and they keep their tone and all but just talk what i have written in ttext.
Whatt is the best workflow for that?