Combining a diffusion model and an image-to-image translation model produces more photorealistic game-engine synthetic images than either alone while keeping semantic labels intact.
Flux.2-4b klein: Text-to-image generation model
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
A Hybrid Approach for Closing the Sim2real Appearance Gap in Game Engine Synthetic Datasets
Combining a diffusion model and an image-to-image translation model produces more photorealistic game-engine synthetic images than either alone while keeping semantic labels intact.