IC-1486HiFaceGAN's FID degrades from 34.83 to 320.41 under Gaussian noise at the highest tested intensity for face super-resolution

Chaohua Shi, Kexin Huang, Lu GAN, Hongqing Liu, Mingrui Zhu, Nannan Wang, Xinbo Gao

SourceOn the Analysis of GAN-based Image-to-Image Translation with Gaussian Noise Injection

The paper evaluates the released HiFaceGAN on a 16x face super-resolution task (32 to 512 pixels) using 1000 test images from FFHQ. Under five noise types at six intensity levels, FID degrades sharply with noise. For Gaussian noise, FID rises from 34.83 (clean) to 320.41 (s6), while PSNR drops from 22.09 to 15.11 dB. The degradation pattern is consistent across all noise types, with color noise producing the worst FID (328.49 at s6).

Evidence
correlational
Key metric
FID: 34.83 (clean) to 320.41 (Gaussian s6); PSNR: 22.09 (clean) to 15.11 dB (Gaussian s6); 5 noise types x 6 intensity levels
Model
HiFaceGAN
Concepts
Failure mode
Datasets
FFHQ [eval]
Methods
FID [eval], PSNR [eval]
Related findings
IC-1484, IC-1485, IC-1487
Extraction
automatic-extraction