Jan. 14, 2018 14:25:58
I'm following the “Channel Packing Pryro Data Using the Texture Sheets ROP” by Mike Lyndon. Its a great video and seems relatively easy to follow. However, the results i'm getting for the packed textures aren't coming out as detailed in the R&G channels as shown by the tutorial. Here is a pic of what I mean:Render Image
As you can see from the left to right, the left has some very crisp details as the right is washed out. I thought it had something to do with the RGB rendering not coming out correctly, but i'm not able to modulate the problem with all the setting i've tried.
Any ideas or suggestions on this would be helpful. Thanks!
Jan. 17, 2018 16:24:39
What does your beauty render look like? Is there enough detail in the volume to give you the crisp details you're looking for?
Also, I'm curious as to why the background is black. Did you change the Image Sequence file format to something other than exr? That could be causing a problem.
Jan. 24, 2018 11:04:39
Thanks for the response. The reason why the background was black was because I had the format at targa instead of exr. I think my problem was caused by me trying to figure out why I was getting vertical lines (found out it was a software bug), but I had changed alot of settings in my file. I ended up installing a new version of houdini, installed the latest 1.8 gamedev from github, and started with a new project file. Everything is working correctly now for the renders.
Is there any way to make the normal map generated less pixelated or add a configurable blur node to just that piece of the blending process.
Thanks for the great tools!
Jan. 24, 2018 12:08:24
You could dive inside the copnet and add a blur if you wish. The sampling of the normal render is relative to the size of the rendered image. It shouldn't be pixelated - at least the rgb light normals should be ‘smooth’.
Jan. 24, 2018 16:56:26
Yeah, i see the rgb light normals look great and smooth, its the other normals that are blocky. So if i increase my camera resolution from 512x512 to 1024x1024, it will turn out smoother, then it will use the downsize to fill in the atlas at 512x512 for an 8x8x4096?