正在加载视频...
视频加载失败
Finally got #gaussiansplatting working! I’m really impressed by how well it captured the fine details of the sneaker fabric and shoelaces, as well as the text. This was generated from 75 DSLR photos I shot on a 24mm lens, took about 30 minutes to process on an RTX 3090,... show more
183,486 次观看 • 2 年前 •via X (Twitter)
10 条评论

This looks amazing! Was the setup pretty complex? Were you able to follow the instructions from Github and get it working or did you run into complications? This looks so much better than the NeRFs I’ve been doing.

Thanks so much Matt! And I definitely kept running into different problems setting up the pipeline. The big thing to get this working was this video tutorial I found. Let me know if you still run into issues and I’ll try to help!

Cool! How big is a file like this on disk?

75 DSLR photos - 1.62 GB Camera Position Files from COLAB - 1.57 GB Trained Gaussian Splatter Model - 339 MB

Mouth. Watering.

Thank you! I was pretty surprised by the quality! I’ve been trying it on some other stuff and getting mixed results so trying to understand how to optimize things. Have you played around with the tech yet?

Can it be exported to an obj?

It’s pretty new so there aren’t many tools out other than the viewer. I’m hoping we start seeing more stuff soon!

How do the reflections work on the non fabric parts of the shoe?? That is amazing! Would kill to see this tech implemented into a game

This video might help explain how this all works:
