search

minar09 / cp-vton-plus

Official implementation for "CP-VTON+: Clothing Shape and Texture Preserving Image-Based Virtual Try-On", CVPRW 2020
https://minar09.github.io/cpvtonplus/
MIT License
343 stars 120 forks source link

result analysis #64

Open amandazw opened 3 years ago

amandazw commented 3 years ago

000008_0 here is my result, we can see the pic's background is blurred,and the result is not good,any idea to improve it? many thx.

minar09 commented 3 years ago

@amandazw , actually this is the expected result for CP-VTON+. As you can see, the try-on cloth is shorter than the one the user was wearing, so the generative network filled the empty pixels with grey colors. This is actually kind of active research area. You might work on the better segmentation generation or try other SOTA approach for this kind of cases. Good luck!

amandazw commented 3 years ago

@amandazw , actually this is the expected result for CP-VTON+. As you can see, the try-on cloth is shorter than the one the user was wearing, so the generative network filled the empty pixels with grey colors. This is actually kind of active research area. You might work on the better segmentation generation or try other SOTA approach for this kind of cases. Good luck!

thanks for your reply,have a nice day!

thaithanhtuan commented 2 years ago

I think this is not the expected result for CP-VTON+. Maybe you got this bad for the wrong warped clothing (maybe from the wrong open pose or segmentation of reference human image). Can you share the warped clothing generated from GMM? And possible the inputs or the GMM?