Flexible Motion In-betweening with Diffusion Models

์ €์ž: Setareh Cohan, Guy Tevet, Daniele Reda, Xue Bin Peng, Michiel van de Panne | ๋‚ ์งœ: 2024-05-17 | URL: https://arxiv.org/abs/2405.11126 📄 PDF


Essence

Figure 1

Figure 1: Flexible motion in-betweening given a text prompt and spatio-temporally sparse keyframes. From left to right:

CondMDI๋Š” diffusion model ๊ธฐ๋ฐ˜์˜ ํ†ตํ•ฉ๋œ ๋ชจ์…˜ ์ธ-๋น„ํŠธ์œ„๋‹ ๋ฐฉ๋ฒ•์œผ๋กœ, ํ…์ŠคํŠธ ์กฐ๊ฑด๊ณผ ํ•จ๊ป˜ ์œ ์—ฐํ•œ keyframe ์ œ์•ฝ์„ ๋ฐ›์•„ ๋‹ค์–‘ํ•˜๊ณ  ์ •๋ฐ€ํ•œ ์ธ๊ฐ„ ๋ชจ์…˜์„ ์ƒ์„ฑํ•œ๋‹ค.

Motivation

Achievement

Figure 2

Figure 2: Conditional Motion Diffusion In-betweening (CondMDI) overview. The model is fed a noisy motion sequence x๐‘ก,

How

Figure 2

Figure 2: Conditional Motion Diffusion In-betweening (CondMDI) overview. The model is fed a noisy motion sequence x๐‘ก,

Originality

Limitation & Further Study

Evaluation

Novelty: 4/5 Technical Soundness: 4/5 Significance: 4/5 Clarity: 4/5 Overall: 4/5

์ดํ‰: CondMDI๋Š” masked conditional diffusion model์„ ํ†ตํ•ด motion in-betweening์˜ ์˜ค๋žœ ํ•œ๊ณ„๋ฅผ ํšจ๊ณผ์ ์œผ๋กœ ํ•ด๊ฒฐํ•˜๋ฉฐ, ์œ ์—ฐํ•œ ์ œ์•ฝ ์ฒ˜๋ฆฌ์™€ ํ…์ŠคํŠธ ์กฐ๊ฑด์˜ ํ†ตํ•ฉ์œผ๋กœ ์‹ค๋ฌด์  ๊ฐ€์น˜๊ฐ€ ๋†’๊ณ  ๊ธฐ์ˆ ์ ์œผ๋กœ๋„ ์šฐ์ˆ˜ํ•œ ๊ธฐ์—ฌ๋ฅผ ์ œ์‹œํ•œ๋‹ค.

← ๋ชฉ๋ก์œผ๋กœ ๋Œ์•„๊ฐ€๊ธฐ

๐ŸŽง Audio Overview

์ด ๋…ผ๋ฌธ ๋ฆฌ๋ทฐ๋ฅผ ํŒŸ์บ์ŠคํŠธํ˜• ์˜ค๋””์˜ค๋กœ ์ƒ์„ฑํ•ฉ๋‹ˆ๋‹ค. (Gemini ยท ํ‚ค๋Š” ๋ธŒ๋ผ์šฐ์ €์—๋งŒ ์ €์žฅ ยท ์™„์„ฑ๋ณธ์€ ์ด๋ฉ”์ผ๋กœ๋„ ์ „์†ก)
โ–ธ ๊ณ ๊ธ‰: ๊ตฌ์„ฑ ๋ฐฉํ–ฅ(๋Œ€๋ณธ ์ž‘์„ฑ ์ง€์นจ) ์ง์ ‘ ์ˆ˜์ •