28k steps 2000+ videos - Dataset has been uploaded this time, not super happy tbh.
Anyway all examples are T2V - works better Image to video. but i don't showcase that stuff - Cheers
- Haven't found a happy place workflow yet - use whatever, str 1.0 usually.
Please throw examples out if you have them <3
Description
FAQ
Comments (5)
Nice. this lora is huge but works. Strangely I couldn't get any dialog. I tried twice (my video is posted below, or will be). I didn't use any special audio cancelling lora loader. My usual dialog prompts are included.
rank 128 ? ^^
i want to but i dont want to download 3.66 gb lora
Have you tried the inverse? Like make the smallest example you possibly can, then work your way up? It will significantly reduce the "melding" effect. Once you get a baseline at a low rank, you can increase variety and rank, separately. I honestly think you'll find better outcomes at much smaller datasets and ranks.
why does it weigh 3.66 gb o.O