With last week's shipping of the Kinect for Windows v2 devices and the release of the public preview of the SDK, I expect a flood of updates from those who have been involved in the previously closed dev preview process.
For example, I just came across this new project...
3D body models from the Kinect in a snap
Be a part of the 3D revolution. Help us to test this revolutionary ultra-portable scan-to-3D model solution!
With this app — now in beta release — you can use a single Microsoft Kinect to take full body scans anywhere and quickly convert these scans to 3D body models. These body models, of yourself or others, can be posed, animated, measured, and used in a variety of applications.
How does it work?
1. Body Modeling Made Simple
With a single Kinect, take 4 shots of your subject’s body in a static pose and 2 shots of the face. No turntable needed!
2. Receive Your Body Model
Once you take the scan, click “get model” to upload it to Body Hub, our body model web app. In the same day (and this will get faster!) your 3D body model will be available for download.
What can you do with your body model?
- Custom Clothing / Made-to-measure: With the body model, we can extract measurements and provide you with a highly accurate visualization of your client’s shape.
- Character Creation: Using Mixamo Fuse, you can upload your body model and create and animate your own character. See how easy it is to put yourself in a game or film project here.
- But Wait…There’s More!: Well…there will be soon. In the near future, all kinds of applications — from apparel and equipment design to fitness apps and recommendation engines — will be able to use your 3D body models.
Want to know how we do it? Check out our lab notes below
Body Snap uses a single Kinect sensor at a fixed position, which captures several views of the person from different angles, as they turn. This is in contrast with other systems, where an operator waves a single sensor around the person being scanned, or multiple Kinect sensors simultaneously capture data from different angles.
As a person moves, their body shape changes, so rigid 3D alignment across these views is impossible. To cope with variation in pose across views, we use a parameterized body model which factors 3D body shape and pose variations. This enables the estimation of a single, consistent shape across views, while allowing pose to vary.
Based on monocular, low-resolution image silhouettes and coarse range data, we estimate our parametric model to reconstruct shape. See these papers for more details:
Project Information URL: http://www.bodysnapapp.com/
Project Download URL: http://www.bodysnapapp.com/
scanning the body with Body Snap app using Kinect for Windows v2 for character creation in Mixamo/Fuse
Hey guys, we're beta testing a new application -- Body Snap -- we built that let's you scan anybody at home and create a 3D digital model of that body. These bodies can be used as a base shape for character creation in Mixamo, and can then be used in video games. We are currently working on improving our automatic alignments from the scanner input, and would love some (friendly!) feedback on our bodies, and some general UI/UX stuff. We just released it 2 days ago for a public beta, so it's not completely optimized yet, but we were hoping to get some people who have the Kinect SDK to play around with it and give us some feedback. Thanks! TLDR: Make avatars with kinect, tell us what you think.
Project Information URL: http://www.reddit.com/r/kinect/comments/2b2ziv/scanning_the_body_with_body_snap_app_using_kinect/