View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0005227 | Spring engine | General | public | 2016-05-05 11:26 | 2016-05-19 16:06 |
| Reporter | MadrMan | Assigned To | MadrMan | ||
| Priority | normal | Severity | major | Reproducibility | always |
| Status | resolved | Resolution | fixed | ||
| Product Version | 101.0+git | ||||
| Summary | 0005227: Graphical glitch obscuring most of the screen | ||||
| Description | During combat (and in particular while units are exploding, but not exclusively), large parts of the screen are obscured by backdrop-coloured polys. See screenshots: https://dl.dropboxusercontent.com/u/314911/spring/screen00002.jpg https://dl.dropboxusercontent.com/u/314911/spring/screen00000.jpg https://dl.dropboxusercontent.com/u/314911/spring/screen00003.jpg (note how some of the units actually clip through the blackness) https://dl.dropboxusercontent.com/u/314911/spring/screen00004.jpg https://dl.dropboxusercontent.com/u/314911/spring/screen00006.jpg Tried this on two GPUs: (Old-ish) AMD gpu - happens a LOT during battles, makes it nearly impossible to play. DebugGL doesn't show anything at all (driver doesn't support it, i guess.) Recent (mobile) NVIDIA gpu - happens very rarely here, but it does happen. DebugGL works fine but doesn't give any errors. I'm guessing there's a vbo underflowing somewhere and the AMD is a little more upset about it than the nvid gpu. This already happened in 100.0 and 101.0 with my AMD gpu, but much much much less. It's practically impossible to look at any combat now. Attached infolog is from the AMD gpu. I tried to check with some validation tools if anything was wrong, but nvidia absolutely refuses to support any of their debug tools on their mobile gpus. *sigh* Turning off LuaUI does not make a difference. | ||||
| Steps To Reproduce | /cheat /give 200 corak control-d | ||||
| Tags | No tags attached. | ||||
| Attached Files | |||||
| Checked infolog.txt for Errors | |||||
|
|
Managed to somewhat reliably reproduce it on the nvidia gpu: https://dl.dropboxusercontent.com/u/314911/spring/screen00009.jpg https://dl.dropboxusercontent.com/u/314911/spring/screen00023.jpg /cheat /give 800 corkrog happy control-d'ing |
|
|
countless AKs and krogs lost their lives but I couldn't reproduce this |
|
|
I'm fairly convinced the bugs I'm having on the AMD gpu and the Nvidia gpu are two distinct bugs; the bug on the Nvidia seems to be one of those tiny explosions effects with an extreme scaling applied. The position and colors seem to be an exact match, plus they actually cast shadows correctly from what I can tell.. The AMD issue seems to be different, but not necessarily unrelated. I'll try a bunch of a different GL tools and see if I can figure out what it's originating from. |
|
|
Right, the AMD issue: I tracked it down to CProjectileDrawer::DrawProjectileModel, which draws a certain piece while rotating it as follows: glRotatef(angle = 6.95593e+22, x = 0.892289, y = -0.292652, z = 0.343766) Applying that insane angle causes the verts to be so warped (due to floating point inaccuracing I guess) that they mess stuff up. Manually changing the angle to '0' for that call 'fixed' the issue, still trying to figure out where the angle is originating from. |
|
|
Fix 9bcf38df22d80e3fc3e3d5a904a05c9adb6c322c committed to develop branch: Fix 0005227: spinAngle was garbage, so affected vertices also turned into garbage., repo: spring changeset id: 7063 |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2016-05-05 11:26 | MadrMan | New Issue | |
| 2016-05-05 11:26 | MadrMan | File Added: infolog.txt | |
| 2016-05-05 11:37 | MadrMan | Note Added: 0016264 | |
| 2016-05-07 21:11 | hokomoko | Note Added: 0016278 | |
| 2016-05-08 13:37 | MadrMan | Note Added: 0016280 | |
| 2016-05-15 15:22 | MadrMan | Note Added: 0016299 | |
| 2016-05-15 16:48 | MadrMan | Changeset attached | => spring develop 9bcf38df |
| 2016-05-15 16:48 | MadrMan | Note Added: 0016302 | |
| 2016-05-15 16:48 | MadrMan | Assigned To | => MadrMan |
| 2016-05-15 16:48 | MadrMan | Status | new => resolved |
| 2016-05-15 16:48 | MadrMan | Resolution | open => fixed |
| 2016-05-19 16:06 | hokomoko | Relationship added | has duplicate 0005167 |
