hi, im getting an issue where the location of the cameras appears to be incorrect and therefore the splat goes wrong, the XMP values seem ok when I check a couple of them in a text editor.
We ran into similar problems. Reality Capture uses the XMP files as a 'hint', but then the solve can break. We're updating the workflow to use COLMAP, which lets us specify the camera positions rigidly. I'll update the tutorial as soon as we have that in place.
Not really useful to do this with a raw scan. You don't have proper PBR materials so this is a lossy result. Would've been better if the author of the scan processed the original dataset.
very interesting to see gsplats of CG scenes working for visualization. This should help with getting an idea of lighting for sure.
It's a fantastic technique that lets you render the 'lit' CG scene directly in the iPhone on set, instead of hauling around workstations.
Thanks for posting this video. It'll be very very useful
hi, im getting an issue where the location of the cameras appears to be incorrect and therefore the splat goes wrong, the XMP values seem ok when I check a couple of them in a text editor.
We ran into similar problems. Reality Capture uses the XMP files as a 'hint', but then the solve can break. We're updating the workflow to use COLMAP, which lets us specify the camera positions rigidly. I'll update the tutorial as soon as we have that in place.
@@lightcrafttechnology thankyou!
What did you use to take the photogrammetry of the room? Was it an iPhone?
It was originally a download from Sketchfab made by user artfletch: skfb.ly/6QXuP
@ thank you
Not really useful to do this with a raw scan. You don't have proper PBR materials so this is a lossy result. Would've been better if the author of the scan processed the original dataset.