Case Study: Super-Resolution and Photo Restoration
Design the feature that upscales, de-noises and restores users' photos — the edge-vs-cloud split that decides an on-phone NPU model against a pro cloud restoration tool
Model interview answer for super-resolution and restoration (Pixel Super Res Zoom / Adobe Super Resolution / Real-ESRGAN-class systems): two product points designed side by side — on-device 2-4x gallery upscale under a phone NPU budget and cloud 8x+ face/scratch/colour restoration — with NVIDIA DLSS named and scoped out as a third, differently-constrained regime; the perception-distortion trade-off (PSNR-optimized CNNs vs. GAN-based perceptual SR vs. diffusion SR) argued against the GAN-vs-diffusion comparison table from this track's GAN foundation subject; degradation modelling as the data problem that actually decides whether a model works on real photos; RRDB/ESRGAN-style generators, the relativistic discriminator, and the GAN-loss weight as a tunable hallucination dial; a hallucination audit as this case's non-negotiable evaluation gate alongside PSNR/SSIM-vs-LPIPS/NIQE disagreement; and the edge-vs-cloud scalability story — tiling with overlap, INT8 quantization and distillation, a cheap-CNN-everywhere-plus-GAN/diffusion-on-request cascade, and the DLSS lesson that real-time speed is bought by constraints, not by a smaller version of the same model.
Practice questions (6)
-
View →
Two Products, One Prompt: Scoping the On-Device and Cloud Paths
Advanced · Free -
View →
A Model That Improved PSNR and Made Users Angrier
Advanced · Free -
View →
Why the Bicubic-Trained Model Fails on Real Phones
Advanced -
View →
Debugging a Restoration GAN That Won't Sharpen
Advanced -
View →
A Restoration That Scores Well and Is Still Wrong
Advanced