-
Notifications
You must be signed in to change notification settings - Fork 0
OGSMOD-8325 - Investigation of SkyDome utest failure #122
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Changes from all commits
c802b09
ce98a1c
f8e105c
8491ff5
73439db
93c309a
7fb7442
9e2ba52
3536391
4349467
File filter
Filter by extension
Conversations
Jump to
Diff view
Diff view
There are no files selected for viewing
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -74,14 +74,18 @@ target_compile_definitions(${_TARGET} | |
| ) | ||
|
|
||
| # Where to find all the data files for the unit tests. | ||
| cmake_path(SET hvt_test_data_normalized NORMALIZE "${CMAKE_CURRENT_SOURCE_DIR}/..") | ||
| cmake_path(SET hvt_test_data_normalized NORMALIZE "${CMAKE_CURRENT_SOURCE_DIR}/../test") | ||
| # Remove the last '/' if any. | ||
| get_filename_component(hvt_test_data_normalized "${hvt_test_data_normalized}" DIRECTORY) | ||
| target_compile_definitions(${_TARGET} | ||
| PRIVATE | ||
| HVT_TEST_DATA_PATH=${hvt_test_data_normalized} | ||
| ) | ||
|
|
||
| # Where to output any processing results. | ||
| cmake_path(SET test_data_output_normalized NORMALIZE "${CMAKE_CURRENT_BINARY_DIR}/..") | ||
| # Remove the last '/' if any. | ||
| get_filename_component(test_data_output_normalized "${test_data_output_normalized}" DIRECTORY) | ||
|
Contributor
Author
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. Avoid having a path with some platform specific issues like:
|
||
| target_compile_definitions(${_TARGET} | ||
| PRIVATE | ||
| TEST_DATA_OUTPUT_PATH=${test_data_output_normalized} | ||
|
|
||
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -66,7 +66,7 @@ const std::filesystem::path inAssetsPath = getenv("HVT_TEST_ASSETS"); | |
| const std::string resFullpath = getenv("HVT_RESOURCES"); | ||
| std::filesystem::path inBaselinePath = getenv("HVT_BASELINES"); | ||
| #else | ||
| const auto outFullpath = std::filesystem::path(TOSTRING(TEST_DATA_OUTPUT_PATH)) / "computed"; | ||
| const std::filesystem::path outFullpath = TOSTRING(TEST_DATA_OUTPUT_PATH) + "/computed"; | ||
|
Contributor
Author
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. Change to have a consistent path between platforms (i.e. *nix vs. Windows) |
||
| const std::filesystem::path inAssetsPath = TOSTRING(HVT_TEST_DATA_PATH) + "/data/assets"; | ||
| const std::string resFullpath = TOSTRING(HVT_RESOURCE_PATH); | ||
| std::filesystem::path inBaselinePath = TOSTRING(HVT_TEST_DATA_PATH) + "/data/baselines"; | ||
|
|
||
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -139,6 +139,10 @@ HVT_TEST(howTo, useSkyDomeTask) | |
|
|
||
| // Validates the rendering result. | ||
|
|
||
| // OGSMOD-8325 - WebGPU & Linux needs a small threshold to use baseline images. | ||
| ASSERT_TRUE(context->validateImages(computedImageName, imageFile, 20)); | ||
| std::string newImageName = imageFile; | ||
| #if PXR_VERSION >= 2511 && !defined(ENABLE_ADSK_OPENUSD_PENDING) | ||
| newImageName = "origin_dev/02511/" + imageFile; | ||
| #endif | ||
|
Contributor
Author
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. There is a light difference from
Contributor
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. I just had a look at the new image, and it looks like it is "sharper" (less blurry). I am not sure what the change is between 25.08 & 25.11 to have this effect. Perhaps a difference in how the image samples are averaged? (MipLevel or MSAA related change?) The lighting from the DomeLight (Image-Based lighting using the SkyDome) on the 3D object is also slightly different, but to a lesser degree. I approve adding a new baseline image, but it would be interesting to know what causes this change, and if the baseline images from OpenUSD repo have also been changed recently. We can answer that question later, just adding my thoughts here, in case we need to investigate some of this later.
Contributor
Author
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. There is also some issue/change with the MSAA introduced by the change to |
||
|
|
||
| ASSERT_TRUE(context->validateImages(computedImageName, newImageName)); | ||
| } | ||
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Although this change does not appear to be related to the failure, I am fine with the change from
HdxRenderTaskParams*toHdxRenderTaskParams&, it is more consistent.For the record, the only reason a pointer was used before is because this function was originally from TaskController.cpp.
See:
Open USD repo link:
https://github.com/PixarAnimationStudios/OpenUSD/blob/48cb715a00097e93fa857810d2e948a2c1ad826b/pxr/imaging/hdx/taskController.cpp#L354C1-L392C3