HomeAI News
Introducing: VideoDoodles for Incorporating Dynamic Doodles into Your Videos
19

Introducing: VideoDoodles for Incorporating Dynamic Doodles into Your Videos

Hayo News
Hayo News
July 21st, 2023
View OriginalTranslated by Google

VideoDoodles: Developed by researchers such as Adobe, is an editing technique that incorporates animated doodles into videos.

This technique adds dynamic graffiti to objects in a video. Simultaneously drawn scribbles follow the movement of the subject and are expressed with appropriate perspective and occlusion (a phenomenon in which the subject is obscured by other objects).

So you can create unique and impressive videos without professional skills or a lot of time.

Click me to view the demo video

Added doodles adjust accordingly to changes in objects and environments in the video.

For example, if an object in the video moves into the distance, the doodle will shrink accordingly to simulate perspective. Likewise, if the graffiti is obscured by other objects in the video, the graffiti will be hidden to simulate the occlusion effect.

This technique allows doodles to blend into the video more naturally, as if they were part of the video.

Click me to view the demo video

VideoDoodles: Mainly based on two core technologies: flat canvas and custom tracking algorithm.

Flat Canvas: The user positions a flat canvas in a 3D scene reconstructed from a video. These canvases can be thought of as virtual "papers" on which users can draw. Since canvases are positioned in 3D space, they can be perspective-transformed based on the camera's perspective and position.

Custom Tracking Algorithm: This algorithm allows the user to anchor the canvas to static or dynamic objects in the scene. This means that if the anchored object moves or rotates, the canvas moves and rotates accordingly, allowing the drawing to follow the movement of the object in the video.

Details: https://levtech.jp/media/article/column/detail_272/

Paper: http://www-sop.inria.fr/reves/Basilic/2023/YBNWKB23/

Reprinted from @xiaohugggView Original

Comments

no dataCoffee time! Feel free to comment