Tweet
Hello.
Some old assets in our library have bitmaps with the filtering disabled and the blur value set very low. Also in many cases we have 2 set of maps for the same assets (let's say 3D people) for foreground and distant shots.
I know this is not the best practice and that we should use the HDRI map instead of the max default bitmap loader. Also reading Lele's guide for optimizing RAM I wanted to make some tests using .tex maps.
For this purpose I created 5 identical scenes all rendered locally (they all contain a 3D scanned model turned to proxy and instanced 10000times, one vray plane with simple gray material and vray sun and sky):
1. First test:
- resized maps 1K converted to .jpg
- bitmap loader with blur value set to 1
Render time: 11m 26.5s
2. Second test:
- resized maps 1K converted to .jpg
- bitmap loader with blur value set to 0.2
Render time: 10m 41.4s
3. Third test:
- original .png maps 8K
- bitmap loader with blur value set to 1
Render time: 9m 22.4s
4. Fourth test:
- original .png maps 8K
- bitmap loader with blur value set to 0.2
Render time: 10m 20.5s
5. Fifth test:
- converted tiled .tx maps using the Vray bitmap to Vray HDRI texmap converter
- HDRI loader with blur value set to 1
Render time: 9m 42s
All scenes rendered with default settings (bucket mode) geometry set to static, conserve memory on.
I tried looking at the vraymemtrack file but I find it hard to interpret the results. I'm not really sure what all the numbers mean. (tried also with this tool https://pulze.io/free-stuff/memory-tracker)
My goal was to see the RAM usage for all the situations mentioned above so that we find the most efficient way of dealing with assets and maps.
I will attach the files here and maybe someone from chaos could give me some pointers.
Also I don't fully understand the differences in the shadows... I think it has to do with the normal maps but I'm not quite sure what is causing this.
Thanks.
Some old assets in our library have bitmaps with the filtering disabled and the blur value set very low. Also in many cases we have 2 set of maps for the same assets (let's say 3D people) for foreground and distant shots.
I know this is not the best practice and that we should use the HDRI map instead of the max default bitmap loader. Also reading Lele's guide for optimizing RAM I wanted to make some tests using .tex maps.
For this purpose I created 5 identical scenes all rendered locally (they all contain a 3D scanned model turned to proxy and instanced 10000times, one vray plane with simple gray material and vray sun and sky):
1. First test:
- resized maps 1K converted to .jpg
- bitmap loader with blur value set to 1
Render time: 11m 26.5s
2. Second test:
- resized maps 1K converted to .jpg
- bitmap loader with blur value set to 0.2
Render time: 10m 41.4s
3. Third test:
- original .png maps 8K
- bitmap loader with blur value set to 1
Render time: 9m 22.4s
4. Fourth test:
- original .png maps 8K
- bitmap loader with blur value set to 0.2
Render time: 10m 20.5s
5. Fifth test:
- converted tiled .tx maps using the Vray bitmap to Vray HDRI texmap converter
- HDRI loader with blur value set to 1
Render time: 9m 42s
All scenes rendered with default settings (bucket mode) geometry set to static, conserve memory on.
I tried looking at the vraymemtrack file but I find it hard to interpret the results. I'm not really sure what all the numbers mean. (tried also with this tool https://pulze.io/free-stuff/memory-tracker)
My goal was to see the RAM usage for all the situations mentioned above so that we find the most efficient way of dealing with assets and maps.
I will attach the files here and maybe someone from chaos could give me some pointers.
Also I don't fully understand the differences in the shadows... I think it has to do with the normal maps but I'm not quite sure what is causing this.
Thanks.
Comment