inbarhub/DDPM_inversion
Official pytorch implementation of the paper: "An Edit Friendly DDPM Noise Space: Inversion and Manipulations". CVPR 2024.
PythonMIT
Issues
- 1
About the init_image
#11 opened by XiruiTeng - 1
Question about Algorithm 1
#15 opened by alimohammadiamirhossein - 1
why noise maps should be statistically independent and have a standard normal distribution?
#14 opened by maxin-cn - 1
Why z1 = 0 in DDPM?
#13 opened by lihuining - 1
Estimation of `pred_sample_direction`
#12 opened by markub3327 - 2
about the Color manipulation.
#3 opened by segatecm - 5
Application to other schedulers
#1 opened by bonlime - 1
Dependence between epsilons
#5 opened by josejhlee - 1
Asyrp Integration
#6 opened by sidney1505 - 1
imagenet-r-ti2i source prompt
#9 opened by wthu12138 - 1
Question about error accumulation
#2 opened by BossunWang - 2
SDXL or SDXL turo
#7 opened by betterze