Author |
Message |
ZaneZimer
Joined: 2016-06-19 21:33:37 Posts: 385 Location: Columbus area, OH, USA
|
I'm wondering if anyone has seen any issue with the Sun texture when PBR is enabled. It seems to be a semi-transparent darker disk in the bright 'corona' at the sun's position. I have also noted, though maybe unrelated, that the cloud texture used during avatar rez from login or teleport has a 'square-ness' about it. The effect on the avatar rez cloud is very subtle but more visible at 'night'. Isn't as clear in my uploaded images, however. :-/ I have used the official build for the image capture:
Last edited by ZaneZimer on 2024-01-15 15:13:26, edited 1 time in total.
|
2024-01-15 15:10:55 |
|
 |
ZaneZimer
Joined: 2016-06-19 21:33:37 Posts: 385 Location: Columbus area, OH, USA
|
Rez cloud images:
|
2024-01-15 15:11:48 |
|
 |
Henri Beauchamp
Joined: 2009-03-17 18:42:51 Posts: 5940
|
These are not "bugs" but render discrepancies between two different renderers (the legacy one, and LL's new PBR one). There is strictly nothing I can (or would) do on my side about them.If you are not happy with how LL's PBR viewer renders SL, do feel free to open JIRA (or rather Canny, now/soon) issues about them, using LL's own viewer for the screen shots in PBR mode (so that they won't take the fact you are using a TPV as an excuse to disregard your issue). LL decided to push to release a viewer that was barely at beta state... They will have to endure the consequences and fix their stuff by themselves now !
|
2024-01-15 15:16:37 |
|
 |
ZaneZimer
Joined: 2016-06-19 21:33:37 Posts: 385 Location: Columbus area, OH, USA
|
I was just about to redownload and look at the official viewer in Wine. I didn't know if these might also be related to other render discrepancies you have noted before, Henri.
|
2024-01-15 15:19:30 |
|
 |
ZaneZimer
Joined: 2016-06-19 21:33:37 Posts: 385 Location: Columbus area, OH, USA
|
As a follow on, installing Second Life Release 7.1.2.7215179142 (64bit) and running via Wine doesn't show the issue with the sun image. It does, however, have the issue with avatar 'rez cloud' particles. This may be related to this, as noted by a Mole: PBR Bug - Alpha Transparancy on Water issue. I have added a comment to note the issue also seems to affect the 'rez cloud' particles.
|
2024-01-15 18:23:44 |
|
 |
Henri Beauchamp
Joined: 2009-03-17 18:42:51 Posts: 5940
|
The Cool VL Viewer got changes that have been done in the viewer sources git, but have not yet been released (I even had to revert part of those changes, since they broke the avatar preview), so yes, you will see differences with the current official release viewer... As for the Sun and Moon discs, one of the recent changes was about their altitude, to make them render in front of the stars; for the Sun disc, this does not make much sense anyway, since when the Sun is showing, the stars are not rendering... This likely also caused the Sun disc to render in front of the Sun glow. To fix this, change the app_settings/shaders/pbr/class1/deferred/sunDiscV.glsl file in the installation directory to read, line 47 and 48: Instead of:
|
2024-01-15 19:03:28 |
|
 |
ZaneZimer
Joined: 2016-06-19 21:33:37 Posts: 385 Location: Columbus area, OH, USA
|
Is it really pos.z versus the original pos.w? If I use the change above, the sun disc now flickers as the camera angle changes. Somewhat like when two object faces intersect at the same coordinates. *edit: using pos.w and the new factor works though.
|
2024-01-15 19:20:38 |
|
 |
Henri Beauchamp
Joined: 2009-03-17 18:42:51 Posts: 5940
|
No, pos.w would be in the 4th dimension... and indeed looks like a bug by LL. Floating point error/rounding issue... Use a slightly larger factor: 1.0001 should do fine with all drivers and GPUs (all GPUs do not use the same floating point representation).
|
2024-01-15 19:32:39 |
|
 |
ZaneZimer
Joined: 2016-06-19 21:33:37 Posts: 385 Location: Columbus area, OH, USA
|
I thought the .w coordinate part would be odd but noted in the .glsl file was a vec4, and just chalked it up to logic I didn't understand. The slightly larger value does work, on the .z part.
|
2024-01-15 19:43:53 |
|
|