AdvFD: Boosting Visual Generation via Adversarial Fréchet Distance Loss
Kwai-Kolors · hf · 2026-08-12
AdvFD improves generator post-training by introducing a learnable adversarial feature space to static Fréchet losses. It applies whitening to stabilize the optimization process, enhancing overall visual generation quality.
More from Multimodal
- Flux Depth Control Yields Poor Results, Developers Seek Workflows — VirtualLavishness463 · 2026-08-12
- Running Video Generation on 16GB Macs: Open-Source VPIPE Engine Breakthrough — TgoAI · 2026-08-12
- Meshy V7 Hits fal: Instantly Turns Images into Export-Ready 3D Models — petewoodbridge · 2026-08-12
- MiniMax H3 and WAN2GP Workflow Tested for Long AI Video — ArjanDoge · 2026-08-12
- AI Video Creation Workflow: From Storyboard to Generation — alifcoder · 2026-08-12
- FLUX 3 Video hits #2 on Arena, just 16 points behind Gemini — arena · 2026-08-12