Evidence receipt / preference
Published · transcript-backedComfyanonymous: preference
4 Jan 2025 Latent Space AI Engineering for Art — with comfyanonymous, of ComfyUI
“There's, yeah, there's actually a few of them, but the one I've implemented in comfy is Mochi, because that seems to be the best one so far.”
Source trail
Everything needed to verify it.
- Speaker
- Comfyanonymous
- Attribution
- Verified speaker
- Claim type
- preference
- Recorded
- 4 Jan 2025
- Publisher
- Latent Space
Transcript context
…Which means you can, like, move through the space, basically. It's the difference. You're not just kind of, like, reorienting. Yeah. And it's also, well, it's also because you have a temporal VAE. Mm-hmm. Also, like, Mochi has a temporal VAE that compresses on, like, the temporal direction, also. So that's something you don't have with, like, yeah, animated diff and stable video diffusion. They only, like, compress spatially, not temporally. Mm-hmm. Right. So, yeah. That's why I call that, like, true video models. There's, yeah, there's actually a few of them, but the one I've implemented in comfy is Mochi, because that seems to be the best one so far. Yeah. We had AJ come and speak at the stable diffusion meetup. The other open one I think I've seen is COG video. Yeah.…
Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.