Thanks for pointing this out. I would have missed it because I never look in that section of the forums.
AMD RX 6000 series driver issues & workarounds
Actually, i'm not sure it's an aliasing problem. If you disable elite's software AA (FXAA, SMAA, etc), there's no AA running.
It's more of a rendering difference with NVIDIA. That would not be the first time.
The line is actually a 3d "ribbon" with a very thin width. If you extend that width, you see the triangles composing the ribbon.
Each point of the "ribbon" is moved/calculated by the shader.
In short, after a few hours of searching, I had to assign the z coordinate of those points to 0 (in the shader's output).
And I got my lines back
It's more of a rendering difference with NVIDIA. That would not be the first time.
The line is actually a 3d "ribbon" with a very thin width. If you extend that width, you see the triangles composing the ribbon.
Each point of the "ribbon" is moved/calculated by the shader.
In short, after a few hours of searching, I had to assign the z coordinate of those points to 0 (in the shader's output).
And I got my lines back
Nice work. Means it can be fixed "relatively easy" when they get around to it.Actually, i'm not sure it's an aliasing problem. If you disable elite's software AA (FXAA, SMAA, etc), there's no AA running.
It's more of a rendering difference with NVIDIA. That would not be the first time.
The line is actually a 3d "ribbon" with a very thin width. If you extend that width, you see the triangles composing the ribbon.
Each point of the "ribbon" is moved/calculated by the shader.
In short, after a few hours of searching, I had to assign the z coordinate of those points to 0 (in the shader's output).
And I got my lines back
("relatively easy" bc coding rarely is)
Actually, i'm not sure it's an aliasing problem. If you disable elite's software AA (FXAA, SMAA, etc), there's no AA running.
It's more of a rendering difference with NVIDIA. That would not be the first time.
The line is actually a 3d "ribbon" with a very thin width. If you extend that width, you see the triangles composing the ribbon.
Each point of the "ribbon" is moved/calculated by the shader.
In short, after a few hours of searching, I had to assign the z coordinate of those points to 0 (in the shader's output).
And I got my lines back
Anything that results in a feature being more aliased than it should be, or more aliased relative to the norm (i.e. these lines on everything that's not an RDNA2 or 3 part), is an (anti)aliasing problem, including incorrect weights being given to underlying geometry, resulting in the color of a pixel being lost to the background.
Anyway, it's interesting to note that these lines are a 3D ribbon rather than a 2D line. I had assumed it was a line and that the problem was a difference in how the game, or drivers, handled line specific AA (which almost always exists in some form, independent of either polygonal AA or post-process filters). I didn't think it was a hardware issue as the line AA looks equivalent in other applications, but it not actually being a line explains a lot...while raising more questions.
Chapter 22. Fast Prefiltered Lines
Chapter 22. Fast Prefiltered Lines Eric Chan Massachusetts Institute of Technology Frédo Durand Massachusetts Institute of Technology This chapter presents an antialiasing technique for lines. Aliased lines appear to have jagged edges, and these "jaggies" are especially noticeable when the lines...
I've seen quite a few report that it works now.
I am very disappointed in this. AMD even know about the issue, so does Fdev.
Can anyone with an Intel 750 or 770 card verify if these issues also affect those cards?
Practical video card choice is limited to one company right now.
Can anyone with an Intel 750 or 770 card verify if these issues also affect those cards?
Practical video card choice is limited to one company right now.
I started getting the Orange Sidewinder in Odyssey when trying to approach a settlement again.
I thought this was already fixed. I'm currently on
Driver Version 22.20.27.09-230330a-390451C
AMD Windows Driver Version 31.0.12029.4013
Does the workaround by replacing files still work?
EDIT: Nope, it doesn't. Maybe it's a brand new error now?
EDIT2: Ignore me, I'm being a bit slow and missed the 4th point in the Workaround #2. Once I edited the file and changed the parameter to false I am again able to bypass the error.
Out of curiosity - didn't AMD already fix this issue? Are they fixing bugs "Frontier style" now?
I thought this was already fixed. I'm currently on
Driver Version 22.20.27.09-230330a-390451C
AMD Windows Driver Version 31.0.12029.4013
Does the workaround by replacing files still work?
EDIT2: Ignore me, I'm being a bit slow and missed the 4th point in the Workaround #2. Once I edited the file and changed the parameter to false I am again able to bypass the error.
Out of curiosity - didn't AMD already fix this issue? Are they fixing bugs "Frontier style" now?
Last edited:
They did, but it's entirely possible Frontier broke it again with one of the updates that happened since the 22.10.3 driver, which is what you seem to be using. That driver came out when U13 was current.
Current driver version is 23.5.2 (31.0.14057.5006), which seems to work fine without having to disable double precision floats, as did the last several versions, at least.
Not having any issues with todays 23.5.2 drivers and vanilla game files - could it now be that workarounds are causing more issues than they solved?
No workarounds for the issues this thread was created to address should be needed post 22.10.3 and U14. ferus is apparently getting Orange Sidewinder crashes with 22.10.3, with no workarounds applied.
As for the workarounds themselves, these would have been reverted when the game was patched and there would be no reason for most people reapply them, if they have recent drivers. However, disabling native DP floats has never had any appreciable downsides, and shouldn't even have a negative performance impact on most GPUs released in the last decade.
Thank you. I thought I had auto update of drivers enabled (after the original issue was fixed by AMD the very fist time) and I was running the latest drivers but it turns out I hadn't.
I'm on 23.5.2 and planets are broken for me in VR, with and without game patches. If anything the issue is more regular now.
I'm trying to avoid surfaces in general (yay Odyssey) but if I need to get planetside, I:
1. Approach the broken planet
2. Glide towards the "surface" (this usually causes POIs to drop to the center of the planet).
3. Quit the game and restart.
4. Sometimes this works, the surface is now intact, and the POI is on it. Sometimes the POI is further away and when flying towards it, I enter a new quadrant of the surface, which is broken more often than not. Goto 3., repeat. Sometimes the glide had me end up inside of a mountain and the game crashes on load. "Redeploy in space" from the main menu, goto 1., repeat.
The statue of brokenness is hilarious, and FDev's silence is deafening.
Yes, I probably should have specified that VR in 4.0.x is still broken on recent AMD GPUs.
Can sympathize with it not working; i had the most weird issue with graphics on Odyssey weapons which wasn't fixed until a complete OS reinstall.
-couple of squadmates did report the "exploding planets" fixed after the last update, however that is not to say exceptions aren't possible.
Currently i'm having some flickering issues with shadows and a super weird one where DSS renders ship decals on planets.
-i'm not going to blame elite for that however bc i know how many other things it can be.
Radeon RX 6750 XT driver 22.5.1
With the latest versions of Elite Dangerous Odyssey, performance has dropped and occasional stuttering has started to occur, especially at the end of the jump when dodging the star.
The orbital line as usual when it's nearly horizontal becomes tremendous.
After long investigation I updated to the driver 23.2.2
The situation has not changed, indeed the stuttering has become more evident.
Intrigued by the solution to fix the orbital lines I wanted to try - 3Dmigoto-1.3.16.zip + EDO Orbit Lines Fix v0.3.zip - https://forums.frontier.co.uk/threads/orbit-lines-shader-fix-for-amd-6000-graphic-cards.616882/
Unbelievable suddenly the graphic card looks transformed.
Perfect orbital lines, completely eliminated the stuttering, the fps almost doubled, the 4K resolution I was able to bring it to ultra and have 60fps synchronized.
Never had such detail and fluidity in conflict zone battles.
Really, many compliments to whoever worked on this solution.
I wonder, why doesn't FD use these libraries in the program?
With the latest versions of Elite Dangerous Odyssey, performance has dropped and occasional stuttering has started to occur, especially at the end of the jump when dodging the star.
The orbital line as usual when it's nearly horizontal becomes tremendous.
After long investigation I updated to the driver 23.2.2
The situation has not changed, indeed the stuttering has become more evident.
Intrigued by the solution to fix the orbital lines I wanted to try - 3Dmigoto-1.3.16.zip + EDO Orbit Lines Fix v0.3.zip - https://forums.frontier.co.uk/threads/orbit-lines-shader-fix-for-amd-6000-graphic-cards.616882/
Unbelievable suddenly the graphic card looks transformed.
Perfect orbital lines, completely eliminated the stuttering, the fps almost doubled, the 4K resolution I was able to bring it to ultra and have 60fps synchronized.
Never had such detail and fluidity in conflict zone battles.
Really, many compliments to whoever worked on this solution.
I wonder, why doesn't FD use these libraries in the program?
Last edited:
I'm glad it works for you, I did that orbit lines fixRadeon RX 6750 XT driver 22.5.1
With the latest versions of Elite Dangerous Odyssey, performance has dropped and occasional stuttering has started to occur, especially at the end of the jump when dodging the star.
The orbital line as usual when it's nearly horizontal becomes tremendous.
After long investigation I updated to the driver 23.2.2
The situation has not changed, indeed the stuttering has become more evident.
Intrigued by the solution to fix the orbital lines I wanted to try - 3Dmigoto-1.3.16.zip + EDO Orbit Lines Fix v0.3.zip - https://forums.frontier.co.uk/threads/orbit-lines-shader-fix-for-amd-6000-graphic-cards.616882/
Unbelievable suddenly the graphic card looks transformed.
Perfect orbital lines, completely eliminated the stuttering, the fps almost doubled, the 4K resolution I was able to bring it to ultra and have 60fps synchronized.
Never had such detail and fluidity in conflict zone battles.
Really, many compliments to whoever worked on this solution.
I wonder, why doesn't FD use these libraries in the program?
But it shouldn't affect performance :
- 3dmigoto mainly allows to edit shaders.
- I've only edited the space lines shader (orbit, exclusion etc).
- And in that shader, i only edited the 3d coordinates used for the lines.
Because it is most likely "not that simple".I wonder, why doesn't FD use these libraries in the program?
-they may not even own the libraries that the calls are done from, or it ties into something else in the backend which in turn will need re-adjusting and if messed up could affect the whole game, not just orbit line shaders. We don't know.
If they can, i'm sure they will implement something similar to the fix asap.