Mind-blowing physics in Wan 2.1 video model - reflections, weight & motion
Wan 2.1 Takes Physics to the Next Level ๐ฅ
In this video, we explore how the Wan AI image-to-video model handles physics with surprising accuracy - from a giraffe's weight bending a tree to realistic reflections in sunglasses. The model also keeps separate entities, like fish, consistent without weird morphing. Itโs not perfect yet (the juggling example almost works), but the way it manages physics, detailed features, and consistency is pretty impressive. Check out the breakdown to see what makes this model so cool!
โฑ๏ธ TIMESTAMPS โฑ๏ธ
00:00 - Intro: Wan AI video model examples
00:30 - Intricate details: Hands and fingers
01:01 - Image-to-video: Drawing becomes animated
01:35 - Realistic physics: Giraffe weight & tree bend
02:13 - Facial expressions and hairstyle changes
03:19 - Complex juggling: AI almost gets it right
03:54 - Reflections: Sunglasses in the mirror
04:56 - Staying separated: Fish donโt morph
05:37 - Final thoughts: Seriously cool model ability
Try it out here:
https://replicate.com/wan-video
https://replicate.com/wavespeedai
Watch on YouTube โ
(saves to browser)
Sign in to unlock AI tutor explanation ยท โก30
More on: Generative CV
View skill โRelated AI Lessons
โก
โก
โก
โก
Errors as Infrastructure: Why the first crate in NEXUS wasn't networking.
Dev.to ยท Anatolii Shliakhto
Building Microfrontends with Svelte: A Modern Approach to Scalable Web Apps
Dev.to ยท Sravan Kumar Velangi
Ambassador Pattern
Dev.to ยท Aviral Srivastava
3 Resilience Patterns โ Powering the Most Reliable Microservices in 2025
Medium ยท Programming
Chapters (9)
Intro: Wan AI video model examples
0:30
Intricate details: Hands and fingers
1:01
Image-to-video: Drawing becomes animated
1:35
Realistic physics: Giraffe weight & tree bend
2:13
Facial expressions and hairstyle changes
3:19
Complex juggling: AI almost gets it right
3:54
Reflections: Sunglasses in the mirror
4:56
Staying separated: Fish donโt morph
5:37
Final thoughts: Seriously cool model ability
๐
Tutor Explanation
DeepCamp AI