You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Describe the bug
I've created assets which has two shaders ( material x ). One for preview and one for render. The shaders are identical but for the preview shader I'm using only 1K jpg textures.
If I load in 5-10 of these assets the GPU memory quickly fills up to 10GB.
If I resave these assets and switch out the TX files to JPG textures on the render shader the memory is fine. So it seems that Viewport 2.0/Hydra tries to load the render shader even though we have not yet activated Arnold or even tried to render.
Steps to reproduce
Create an asset following the ASWF guidelines.
asset.usda, payload.usda, shade.usda, geom.usdc
Create a shader material x ( standard surface ). Use only mtlx nodes. Add base color, roughness, normalmap and displacement.
Duplicate the shader so you have two shaders named: proxy_mtl, render_mtl
Use TX files (4K or 8K) in the render_mtl and 1K JPG in the proxy_mtl
Assign the proxy_mtl to /ASSET/geo/proxy
Assign the render_mtl to /ASSET/geo/render
Do this to a few assets.
load them all in as references into a stage. Watch the GPU memory
close Maya
edit the shade.usda file and replace the ".tx" with ".jpg"
load them all in as references into a stage. Watch the GPU memory
Easiest would be to download Megascan assets and use those as a testing ground. I recommend you download assets with 8K textures.
Expected behavior
Hydra/Viewport 2.0 should not care about the render material as its not assigned to the meshes that has purpose:proxy
Attachments
Megascan Boulders, this uses 10GB of VRAM with tx files in the render material
Megascan ID's for the assets are as follows
uisjbbis, ukopdins, ukopeess, ukopeijs, ukopfbks, weimcftga, wenobdn
Specs (if applicable):
Rocky Linux 8.9
128GB Ram
Intel® Core™ i9-9900K CPU @ 3.60GHz × 16
NVIDIA RTX A4000/PCIe/SSE2
NVIDIA Driver Version: 550.54.15
NVIDIA CUDA Version: 12.4
Maya 2025.1
maya_usd-0.28
The text was updated successfully, but these errors were encountered:
Describe the bug
I've created assets which has two shaders ( material x ). One for preview and one for render. The shaders are identical but for the preview shader I'm using only 1K jpg textures.
If I load in 5-10 of these assets the GPU memory quickly fills up to 10GB.
If I resave these assets and switch out the TX files to JPG textures on the render shader the memory is fine. So it seems that Viewport 2.0/Hydra tries to load the render shader even though we have not yet activated Arnold or even tried to render.
Steps to reproduce
Create an asset following the ASWF guidelines.
asset.usda, payload.usda, shade.usda, geom.usdc
Create a shader material x ( standard surface ). Use only mtlx nodes. Add base color, roughness, normalmap and displacement.
Duplicate the shader so you have two shaders named: proxy_mtl, render_mtl
Use TX files (4K or 8K) in the render_mtl and 1K JPG in the proxy_mtl
Assign the proxy_mtl to /ASSET/geo/proxy
Assign the render_mtl to /ASSET/geo/render
Do this to a few assets.
load them all in as references into a stage. Watch the GPU memory
close Maya
edit the shade.usda file and replace the ".tx" with ".jpg"
load them all in as references into a stage. Watch the GPU memory
Easiest would be to download Megascan assets and use those as a testing ground. I recommend you download assets with 8K textures.
Expected behavior
Hydra/Viewport 2.0 should not care about the render material as its not assigned to the meshes that has purpose:proxy
Attachments
Megascan Boulders, this uses 10GB of VRAM with tx files in the render material
Megascan ID's for the assets are as follows
uisjbbis, ukopdins, ukopeess, ukopeijs, ukopfbks, weimcftga, wenobdn
Specs (if applicable):
The text was updated successfully, but these errors were encountered: