Haiwen (Haven) Feng (@HavenFeng)

2025-06-12 | โค๏ธ 108 | ๐Ÿ” 20


๐Ÿš€ Introducing GenLit โ€“ Reformulating Single-Image Relighting as Video Generation! We leverage video diffusion models to perform realistic near-field relighting from just a single imageโ€”No explicit 3D reconstruction or ray tracing required! No intermediate graphics buffers, directly in the pixel space! ๐Ÿ“„ Dive into the paper: https://arxiv.org/abs/2412.11224 ๐ŸŽฅ Project page & demos: https://t.co/fbMBoEukXN ๐Ÿ›  Code coming soon! #GenerativeAI ComputerVision Relighting DiffusionModels Graphics ๐Ÿงต 1/5

๋ฏธ๋””์–ด

video


์š”์•ฝ

GenLit์€ ๋‹จ์ผ ์ด๋ฏธ์ง€ ๋ฆฌ๋ผ์ดํŒ…์„ ๋น„๋””์˜ค ์ƒ์„ฑ ๋ฌธ์ œ๋กœ ์žฌ์ •์˜ํ•ด, 3D ๋ณต์›ยท๋ ˆ์ด ํŠธ๋ ˆ์ด์‹ฑ ์—†์ด ํ”ฝ์…€ ๊ณต๊ฐ„์—์„œ ์‚ฌ์‹ค์ ์ธ ๊ทผ๊ฑฐ๋ฆฌ ์กฐ๋ช… ๋ณ€ํ™”๋ฅผ ํ•ฉ์„ฑํ•˜๋Š” ๋ฐฉ๋ฒ•์„ ์ œ์•ˆํ•œ๋‹ค. ์ฒจ๋ถ€ ์ด๋ฏธ์ง€์ฒ˜๋Ÿผ ๊ฐ™์€ ๋ฌผ์ฒด/์žฅ๋ฉด์„ ๋‹ค์–‘ํ•œ ์กฐ๋ช… ์กฐ๊ฑด์œผ๋กœ ์ž์—ฐ์Šค๋Ÿฝ๊ฒŒ ์žฌ์กฐ๋ช…ํ•œ ๊ฒฐ๊ณผ๋ฅผ ๋ณด์—ฌ์ค€๋‹ค.

Tags

domain-vision-3d domain-rendering domain-genai domain-ai-ml domain-dev-tools