🔥Diffusion Masters Transparent Objects in Videos!
Chongjie Ye
在 X 上查看@ychngji62734 位关注者
28,181 次观看 #DKT
🔥Diffusion Masters Transparent Objects in Videos!
We introduce #DKT, a foundation model that repurposes video diffusion for zero-shot depth and normal estimation on Transparent and Reflective Objects with Superior Temporal Consistency
Demo:
Code:
6 条评论
评论来自 X (Twitter) 上的原帖

@ChrisAtKIRI
Good work brother
2 个赞

@AIWarper
Does that released checkpoint also work on normals? If so I’ll implement it in Comfy.
1 个赞

@AwokeKnowing
what does this do when walking up to a big mirror?

@Almorgand
Oh great stuff!

@NTU_chenyiwen
Nice work!

@codewithimanshu
DKT model processes video, generating depth maps at 25 FPS.
