Figure 1 | Scientific Reports

Figure 1

From: Single image super-resolution with denoising diffusion GANS

Figure 1

Random SR (8×) samples generated by SRDDGAN using latent variable Z. Our method generates diverse predicted SR images, including differences in facial attributes and hair (e.g., the second hair detail has a different texture than the fourth, and the third tooth being clearly visible while the fourth is not.), while maintaining consistency with the LR images.

Back to article page