Meitu MT Lab Presents CFT for Stable Portrait Relighting at ECCV 2026
jiqizhixin · x · 2026-08-27
Meitu Imaging Technology Lab (MT Lab) presented CFT (Consistent Feature Transport), accepted at ECCV 2026. This method reframes portrait relighting as a lighting-consistent feature transport problem. Built on Rectified Flow, CFT jointly models noise, source image, and target image distributions, explicitly learning lighting-specific feature displacement rather than generic image-to-image translation. This approach preserves identity and structure stability while altering only the illumination. CFT achieves state-of-the-art scores across all four relighting metrics. The concept of consistent feature transport also generalizes to style transfer, improving content structure and perceptual fidelity.
More from Multimodal
- Open-source LoRA turns MiniMax H3 into seamless 360-degree orbit videos from a single photo — camenduru · 2026-10-15
- Neon Noir Elegance prompt template goes viral with striking pink-blue two-tone outputs — LudovicCreator · 2026-10-15
- Krea 2 Turbo + px16 LoRA generates game-ready pixel art in 5s, where GPT Image 2.5 fails — Angaisb_ · 2026-10-15
- How Goldfish Made Its Kung Fu Scenes: Real Nanquan Choreography + Seedance 2.5 — JunieLauX · 2026-10-15
- Custom px16 LoRA makes Krea 2 Turbo churn out pixel icons at 5s each — Angaisb_ · 2026-10-15
- Prompt share: 'Crimson Footsteps' makes one red element pop in monochrome AI photos — azed_ai · 2026-10-15