Ok. OVI is DOA for me. It requires a minimum of 16 GB of VRAM at this point until someone successfully quantizes it. I have tried multiple things, even allowing CPU offload, and my machine just does not have enough regular RAM, either. So I am sticking with WAN 2.2 for now. It runs great on my video card, even if I cannot get the boobs to get much bigger. I will work on my WAN I2V install so I can generate those wonderful gigantic breasts in Chroma, then animate the images through WAN. If you have seen Katsume (I think I spelled the name right) on Deviant Art that is what they do. Generate a static image with really large boobs, then use it as the first frame in an animated WAN-based video. You take the best sequence, then steal the last frame from that video to feed back into the model to make the next segment, then splice them all together to make longer videos. Katsume apparently has the hardware to run OVI as they have the capability of putting voice in their videos. Again, OVI is WAN 2.2 with additional bits to do sound and also sync it. As my PC is barely able to run WAN and I have to use special versions, there is no way I can run the current version of OVI on the same PC.
Soon I may release my portable installs of ComfyUI with the setups for 12 GB nVidia RTX cards so someone here can pick it up and run with what I have started. Benefit from all of my research, trial and error, and tweaking. You just unzip to a location on your hard drive and run it. I will even see if I can include a model and the LoRAs I have to give you a start. Easier than a tutorial.
Anyway, for the time I am going to try and have Grok give me some prompts that I can feed into my WAN and generate some good stuff for you guys. Look for some T2V videos later today. Getting my I2V working well will take some time.
Punchy