Morning coffee [interpolation, ML/Stable Diffusion generated]

Описание к видео Morning coffee [interpolation, ML/Stable Diffusion generated]

Another interpolation on an SD 1.5-based LDM, again just a single text prompt where we vary the noise, with a little better training. You could add rain sounds and it'd basically an ASMR video. How about that. We're going to see some things.

I strongly suspect that if I varied the interpolation steps according to the magnitude of the difference in VAE feature space (or, y'know, just walked carefully on steeper ground) the 'animation' could be more evenly smooth, but that feels like work and I'm not working today. Maybe tomorrow.

Комментарии

Информация по комментариям в разработке

Похожие видео