Dumb question: what does 'image-to-video' actually mean and when would I use it?

by @fatimakumarworks in c/newbie-questions · · Question

Question flair. I keep seeing people say 'just use i2v' or 'start from a keyframe' and I genuinely don't know what that means as a beginner. I've only been typing text and getting video.

Can someone explain, like I'm five, what image-to-video is, why it's better, and whether I should be using it? Sorry if this is super basic.

77 points · 8 comments

Comments

  • @nikolaibrowncreative829 · · 76 points

    Not basic at all, great question! Text-to-video: you type words, the model invents everything including what things look like. Image-to-video (i2v): you give it a STARTING PICTURE, and it animates THAT. You control the look; it just adds motion.

  • @chenmartinez_films6702 · · 19 points

    Follow up: so why is i2v 'better'? Is it always better?

  • @leomiller_creates · · 54 points

    ELI5 version: text-to-video is ordering a mystery box. Image-to-video is 'here's exactly what I want it to look like, now just make it move.' The second one is way less of a gamble.

  • @hiroshigarcia_creates · · 38 points

    Most serious work is i2v because the starting image is where you win or lose. A great keyframe animated simply beats a text gamble every time. You'll graduate to it naturally.

  • @hiroshiwilliamslab · · 33 points

    The fact that you asked this instead of silently guessing is exactly why this sub is good. Ask the 'dumb' ones, they're never dumb.