Very nice being able to read an actual paper on a powerful text-to-video model. | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		GaggiX 3 months ago \| parent \| context \| favorite \| on: Seedance 1.0 Very nice being able to read an actual paper on a powerful text-to-video model.

liuliu 3 months ago [–]

Yeah. It is great. So apparently separating spatial / temporal attention works if you are careful and train with large enough dataset too!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact