iXit / Mesa-3D

Please use official https://gitlab.freedesktop.org/mesa/mesa/ !
https://github.com/iXit/Mesa-3D/wiki
66 stars 13 forks source link

Texture bug with enabled d3d9 #290

Open junkedboy opened 6 years ago

junkedboy commented 6 years ago

From the last time I have problem with displaying textures in World of Warcraft 3.3.5 (in other games not tested) with d3d9: http://i.piccy.info/i9/a8c6f74e78a0541d49d55ffb9469e4b4/1507565562/134054/1186269/Screenshot_with_enabled_d3d9.jpg without: http://i.piccy.info/i9/651f44f3b1f3a66df21fcb3cf17dd854/1507565695/136128/1186269/Screenshot_without_enabled_d3d9.jpg

Gentoo, wine 2.18, mesa 17.2.3

axeldavy commented 6 years ago

This seems a texture blending bug. Which card ?

junkedboy commented 6 years ago

Notebook Acer V5-552G with AMD 8550G+8750M.

iiv3 commented 6 years ago

I'm sorry but I don't see the bug. There seems to be some difference in the fog, but I can't tell which one is the correct.

May I ask for apitrace of this scene? Then see if the same trace plays different in Nine, WineD3D and real Windows.

If the game uses different paths, because Nine and WineD3D give different capability flags, then the apitrace would hardcode the drawing command so we'll be sure they are the same in all cases.

junkedboy commented 6 years ago

Maybe you don't understand what problem with textures I mean. New screenchots. with enabled d3d9: http://i.piccy.info/i9/92bed11f19746f70b96e5a0863070c56/1509046376/145802/1191034/Screenshot_from_2017_10_26_22_29_56.jpg without enabled d3d9: http://i.piccy.info/i9/286c3cdd63c8658a20817d09cd515a9d/1509046667/131776/1191034/Screenshot_from_2017_10_26_22_30_48.jpg

Tested on arch linux too, wine-gaming-nine 1.8. This is a new bug, appeared ~2 months ago.

Altosk commented 6 years ago

1.8? That's old I don't have that bug on 2.19 staging latest mesa try updating

On Oct 26, 2017 12:45 PM, "junked" notifications@github.com wrote:

Maybe you don't understand what problem with textures I mean. New screenchots. with enabled d3d9: http://i.piccy.info/i9/92bed11f19746f70b96e5a0863070c 56/1509046376/145802/1191034/Screenshot_from_2017_10_26_22_29_56.jpg without enabled d3d9: http://i.piccy.info/i9/286c3cdd63c8658a20817d09cd515a 9d/1509046667/131776/1191034/Screenshot_from_2017_10_26_22_30_48.jpg

Tested on arch linux too, wine-gaming-nine 1.8. This is a new bug, appeared ~2 months ago.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/iXit/Mesa-3D/issues/290#issuecomment-339779592, or mute the thread https://github.com/notifications/unsubscribe-auth/AIkEKuwgaafVDaNEUC6b2Nn1-dDbzQzZks5swOFmgaJpZM4QEtOL .

junkedboy commented 6 years ago

I'm sorry, *wine-gaming-nine 2.18.

Altosk commented 6 years ago

oh didnt see the version of wow i was gonna say that the dx9 was broken with 7.3 but your on 3.3.4 i know that 1.12 wow is fine last i checked it might be radeonsi bug for your series of gpu

axeldavy commented 6 years ago

If this is a regression, as bisect could be done. Either you could make the bisect yourself, or make a trace to help the bisect.

iiv3 commented 6 years ago

Oh, I see the problem now. It seems that pixel sampling uses nearest-neighbor instead of bilinear-interpolation, so the low-res blend textures get sharp edges.

We need an apitrace showing clearly the bug.

If it is regression, try and find the last version that worked. It would help if we know if the problem is in the wine or mesa code. (Mesa code also depends on LLVM)

Also, please mention what is your hardware.

junkedboy commented 6 years ago

Also, please mention what is your hardware.

Notebook model: Acer V5-552G CPU: AMD A8-5557M Video: AMD Radeon HD 8550G + AMD Radeon HD 8750M RAM: 8G

If it is regression, try and find the last version that worked. It would help if we know if the problem is in the wine or mesa code. (Mesa code also depends on LLVM)

Rolled back mesa-17.0.6 and llvm-3.9.1 from gentoo stable, bug not found.

axeldavy commented 6 years ago

Does it work better with csmt_force=0 ?