Solari: Improved mirror denoising #22459
Conversation
| let specular_motion_vector = calculate_motion_vector(virtual_position, virtual_previous_frame_position); | ||
|
|
||
| let F0 = calculate_F0(ray_hit.material.base_color, ray_hit.material.metallic, ray_hit.material.reflectance); | ||
| let wo = normalize(view.world_position - virtual_position); |
There was a problem hiding this comment.
Not entirely sure about calculating wo like this for env_brdf_approx2, but it makes sense to me. Let me know what reviewers think.
JMS55
added
A-Rendering
…nts WC from showing up in reflections when you're really close to something)
|
This PR somehow broke Solari when compiling without DLSS. I need to figure out why. |
| let wo = normalize(view.world_position - virtual_position); | ||
| let virtual_normal = normalize(mirror_rotations * ray_hit.world_normal); | ||
|
|
||
| textureStore(specular_motion_vectors, pixel_id, vec4(specular_motion_vector, vec2(0.0))); |
There was a problem hiding this comment.
Are specular motion vectors meant to be used with PSR?
There was a problem hiding this comment.
I think so? It's a little hard to find literature on it.
There was a problem hiding this comment.
Is there a noticeable difference in image quality in motion when they're set vs just defaulted to 0?
There was a problem hiding this comment.
I tested it, and I don't think so. Maybe it matters for more complex situations? Idk really.
alice-i-cecile
added
the
S-Needs-Review
alice-i-cecile
left a comment
alice-i-cecile
left a comment
There was a problem hiding this comment.
Clear improvement in the showcase, and nothing egregious in the code.
alice-i-cecile
added
S-Ready-For-Final-Review
4 participants
Objective
Solution
Showcase
Before
After
