Video in 8 steps and 4 steps of upscaler
I made this 25-second video in a single take with this model and the H3 OBVPM system and LoRa DMAD. The first sampler was set to 0.5 megapixels, and the upscaler to 1.0 megapixels. It took 50 minutes. I think it improves faces at a distance, and the quality is good.
My specs:
RTX 4070 Super 12 GBs vram
128GB RAM
one hour for 25 seconds for your resolution is not a good speed. post a screen of the workflow, you may have done something wrong. it takes 4 minutes to generate a 6 second video with no upscale at 1280x736 on a rtx pro 4000 which is around 30% slower than what your card can do. get back to basic, use simple workflows as bloated useless workflows full of all the options in the universe will really destroy renders. most people youtuber or whatever have no idea what they are doing and it doesn't help that these complex models creator like this one rarely add ONE comfyui example workflow.