Testing a wgpu application
#4182
-
|
I'm basically brand new to any kind of graphics, so apologies if there are standard answers for this kind of thing. I'm wondering what a typical test setup might look like for a For example, what kinds of tests even make sense in a graphics context? If I want to test a command that says "draw a circle", would the test render the circle in a solid color, read the texture back into CPU memory, then check a pixel inside the circle to see if it's the right color? Is that the general strategy? If I want to take the "middleware" approach (outlined in this repo) for a part of my application, is there a way to determine that I haven't accidentally called Anyway, this isn't a |
Beta Was this translation helpful? Give feedback.
Replies: 4 comments
-
|
I personally can't help you, but I was very impressed with the all-is-cubes renderer test system created by @kpreid, I think taking a look at it might help you. Rust Gamedev Meetup video: |
Beta Was this translation helpful? Give feedback.
-
That's quite a common solution. We do this within wgpu itself and I do it in my library rend3. The devil is in the details though. wgpu and rend3 use bindings to an nvidia library called FLIP which does human-style image comparisons. I personally am quite a fan of it for how well it deals with noisy inputs or inputs with very low variance. |
Beta Was this translation helpful? Give feedback.
-
|
Awesome, thanks both of you |
Beta Was this translation helpful? Give feedback.
-
|
I thought that It's only the non-CUDA binary (none of the Python stuff), but it will let anyone consume the package from Nix. |
Beta Was this translation helpful? Give feedback.
That's quite a common solution. We do this within wgpu itself and I do it in my library rend3. The devil is in the details though. wgpu and rend3 use bindings to an nvidia library called FLIP which does human-style image comparisons. I personally am quite a fan of it for how well it deals with noisy inputs or inputs with very low variance.