Starting to play with video, and I found myself realising there wasn't a basic huge implant model for wan2.1, at least that I could find. I suspect this is because if you use img2img you don't need it. However, in my own workflow, I often use VACE and a control video to get the s[ecific action I want. If you use that, the openpose tends to overwrite the specifics of your image.
However, I discovered if you add loras to the mix, you can add back the elements you've lost. Specifically, for my muscle girls, the MuscleUp lora did a lot of work. However, while that did some size, there was none which gave the huge bolt ons my characters occasionally need.
So I made a quick one, and initial tests show it works well enough that I want to share.
Description
FAQ
Comments (1)
Lifesaver! Works like a charm for wan 2.2 i2v. Keeps shapes and proportions exactly as intended when chain generating multiple clips (using last frame of the previous clip as the next first frame).
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.