Appropriate workflow for Mega I2V in GGUF?
Hi, I apologize as I'm still very much a newbie to this, but I'm unsure how to get the latest Mega models to function correctly. V10 works fine with the provided workflow and other I2V workflows I use, but whenever I try to use the newer Mega, it just ignores the original image and produces an output that's mostly just smeared noise.
I also tried to use the Mega V2 json workflow provided from the original FP Mega releases too, and replacing the single checkpoint module with the multiple GGUF loaders, but it still ignored my starting image and produces noise. Thank you.
Same here, with MegaV5. The default workflow seems to completely ignore the input image. Can someone help, please?
This work, you guys never tried it like that for i2v ?
It works because you're using the V10. Mega checkpoints ignore the original image with this workflow.
any solution to this?
This one works for me. For Safetensors instead of GGUF just exchange the loader.