Updated Ovi 10 second workflow.
I read the README of the new 1.1 Ovi models uploaded. It said the model was video and audio combined and to update the WanWrapper nodes. I did this thinking the workflow for ovi would be updated too. However it was still looking the same. So I tried solving the problem of how to use the new models myself and managed to get 10 second video to generate. However the audio still generates only 5 seconds. Is it okay to request that updated workflows be provided accompanying updated models such as the new Ovi 1.1 10 second model? So that in the Files and versions there are the models but also a workflow file included that shows how the workflow changes or is updated to support the new uploaded models?
Workflow will be updated when it's done testing of course. You'll have to change the audio latent length to 314, and video num_frames to 241 to generate 10 seconds, this will also use a lot more VRAM so probably have to increase block swap too depending on your GPU. I also just found pretty bad bug that always existing in my Ovi code, it's fixed now so using latest version is important.
Thanks for the information. I updated to the nightly build about an hour ago. Are you saying you discovered a bug and updated it since then and I should update again? It's impressive how hard you are working all the time I hope you know how much we appreciate it!
Thanks for the information. I updated to the nightly build about an hour ago. Are you saying you discovered a bug and updated it since then and I should update again? It's impressive how hard you are working all the time I hope you know how much we appreciate it!
Yeah, 15 mins ago. Bug was that the audio negative prompt wasn't actually used, and it always used the video negative instead, which probably was bad for audio...
Understood. I will update now. Once again thank you so much for your hard work. You are going to go down in history when we look back on the early days of all this stuff. You are a very important contributor to the community.