I'm so impressed with how Wan 2.1 14b interpreted this image for the video pic.twitter.com/DdcjisUObp
— fofr (@fofrAI) February 27, 2025
Wan 2.1 image to video.
— Halim Alrasihi (@HalimAlrasihi) March 1, 2025
The only video model that correctly interpreted the image.
Each dimension is shifting independently without changing the whole form. pic.twitter.com/DACfjONBa5
Video inpainting test with Wan 2.1 1.3b model using FlowEdit in ComfyUI 🔥
— Nathan Shipley (@CitizenPlain) March 1, 2025
Prompt: "a burning fire and flames on a blue background"
Weird tentacle version in thread 🐙#aivideo #VFX pic.twitter.com/tGHk2JoDsX
I love where this is going.
— fofr (@fofrAI) February 28, 2025
All of these used the prompt:
> A tiger performs BMX tricks at the olympics
(made with Wan 2.1 14b t2v on Replicate) pic.twitter.com/2eIA3EqeX8
Wan 2.1 left - Sora right 👀
— Walgtech 👨🏻💻 (@StianWalgermo) February 28, 2025
Prompt:
"A realistic video of a norwegian house cat doing cross country skiing in a perfect sunny winter day wearing traditional norwegian costume" pic.twitter.com/pp4IpkzazE
Quick AMV anime-style demo using the new WAN 2.1 'I2V' model on @FAL
— Kabooki AI Video (@KabookiAI) February 27, 2025
Its handling of 2D assets is impressive and fluid. Strong open-source performance from @Alibaba_Wan pic.twitter.com/aGtKVDIlKf
wan 2.1的图生视频很能打 pic.twitter.com/voyj3swwCA
— Datou (@Datou) February 26, 2025
퀄은 veo가 더 좋은 것 같은데 움직임은 이게 더 자연스러운듯
veo2랑 비교하면 어떰?
오픈소스가 veo랑 비비는게 말 안됨
아니 veo2 클링급 오픈소스 나오려면 2년쯤 걸릴줄알았더만 ,,진짜 미친속도네 ㅋㅋ