正在加载视频...
视频加载失败
White Light Reference for Machine Learning. Meet our inhouse tech Doggo "Rolo". "More Doggo than Doggo" Since July we've been redesigning our scanning pipeline to work with the new 3D Gaussian Splatting for Real-Time Radiance Field Rendering method from Inria. IR's AeonX capture system has been designed to capture... show more
11 条评论

Is the fur just geometry? Traditional 3d softwares go arround the theme with a huge variety of solutions. Or this just a reference fo then make a proper 3d model (im imagine that what the cam can’t see there are no data / no model)

This is more useful for reference as it's an approximation albeit a very good looking one but I bet the folks at Epic could work some interesting magic with it if they had a commercial license to integrate the method into Unreal.

That's crazy! Great test subject: The hair detail and shading really shows this approach off.

I read the Readme but confusing So you provided photos of this dog trained model and then viewed with viewer? How is the workflow?

The images are aligned in either Reality Capture, or Metashape, or Colmap. They are trained with the Guassian Splatting process then viewed in a modified version of Sibr.

I am currently trying to figure out how to install everything and make my first model. As I figuren you need colmap to feed it to the gaussin splatting? What do you export from Reality Capture? Great results btw :)

Thanks we don't have a defined pipeline that we can share for the moment but I think others have started to document and share their pipeline. I think more resources will come online soon. We plan to create an install write up at some point.

You're showing us that Rolo isn't real... but my brain refuses to listen.

Looks cool!

Hi sorry I dont understand but are these renderings with multiple light conditions?

These are viewed in real-time baked lighting results.

