New animation model dropped yesterday. This is from the Text-to-Video workflow, prompting for Unicorn Twilight Sparkle, but otherwise just mimicking the default prompt in ComfyUI for this model after doing some basic stylistic rewrites. This 5 second clip took 130 seconds to generate on a 5090 and used about 24 GB of VRAM. A 1080p video took about 27 minutes to generate on the same hardware. It is capable of 15 second videos in a single prompt.

I’m going to be doing some further testing today.