Text To Video
topic on 4 shows · 5 statements across 4 episodes
Latent Space 
My First Million 
the Startup Ideas Podcast 
the a16z Podcast
5 statements about Text To Video, every show
He: Audio-video models require precise time alignment unlike text-to-video
“So one important thing is like the alignment. So the model has to know like the video and audio, the it has to have a time based alignment, like at which time step the video and the audio token correspond to each other. We actually don't have these kind of ali…”
Image-to-video AI workflows are cheaper and faster than direct text-to-video
“We like to work with images on the basis that it's a lot easier and cheaper and faster to generate images for the commercial than it is to do everything text to video.”
Genre AI scrapped its first text-to-video ad due to facial morphing
“Initially we did it all in actually text to video, and we didn't like the look. This is when image to video kind of first came out. So we did the ad in text to video, And it just looked too like AI, like the faces were morphe and it looked like shit. So we act…”
Lingelbach: Traditional text-to-video prompting will not remain the primary video AI paradigm
“I think that traditional prompt text to video is like very much not what the paradigm will look like.”
Shah: Prompt-driven text-to-video with director styles will arrive in 2023
“I would not be surprised, let's say by the end of this year, that we have a reasonable way to kind of describe in textual form what we want, who the characters are, what the scene is what kind of stylistic attributes we want. We can point it to, oh, I want thi…”