View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0001833 | FSSCP | graphics | public | 2008-11-28 10:11 | 2012-04-03 13:45 |
| Reporter | Zacam | Assigned To | Echelon9 | ||
| Priority | normal | Severity | major | Reproducibility | always |
| Status | resolved | Resolution | fixed | ||
| Summary | 0001833: The Return of 286. With a twist. | ||||
| Description | The same "edge of graphic flickering jiggly spot" at the tail end of the thrusters is back and with a vengeance. Quoting Lightspeed from the notes in 286: "It's a content problem, NOT a problem with the code. You'll have to make sure the leftmost pixel remains completely black in the thruster graphics images. See the L-B hybrid effects for reference." Well, the left most side IS black. every pixel for 2 pixels width on the left side is completely black. Mip maps or no mip maps, makes no never mind. DXT1a, DXT3, DXT5, u8888: You can not add an alpha channel to the thrusters. Black does what it is supposed to and gives you a nothing to see (but also no flicker) while white still gives you flicker, but not the thruster is only visible from top/bottom view, never from the side. DXT1, u555, u888, TGA: All look the same, so it's not a format or compression issue. All formats tested with and without Mip maps. Mip mapping makes no difference either. Only capital ships don't seem to be affected by this, possibly because of view distance rendering. Speaking of rendering distance, the ships view distance affects this. And the more to the right you move the effect on the canvas, the further from an affected ship you have to be before you notice it. Now, here is the twist and the right kick in the pants: It didn't used to happen. During the pre- and gamma test phase for the mediavps, I never experienced the issue.....and then suddenly I did, even though absolutely nothing had changed other than my running and exiting the game to do testing on missions. New Pilot: Check, and no difference. VP problem: Remade set into VP, no change. Ran set in \data\effects instead of vp, no change. Tried having in a VP AND in the \data\effects folder: still no change. As it had been postulated that it was being caused by having both DDS and TGA versions present in the VP's, I also mixed formats around to. No change. So, if it is a content format, can some one please explain or provide a solution for creating correct content that will not produce the flicker? | ||||
| Steps To Reproduce | Ship Lab. Render Options: Show Thrusters. Disable Model Rotation. Rotate to a angled-front-side-top view and start zooming out. Or load a quick mission by placing a few ships that don't move and have no skybox or stars and fly along behind them at approx greater than 142 meters. | ||||
| Additional Information | Happens regardless of species. Prior to the Gamma set, some ships had retail model thruster cones, some did not. During Gamma and beyond to current, all ships no longer have the modeled thruster cones. Just as a test, I loaded one that still did, and it still had flicker. So I changed the +Pri_Normal: <none> to +Pri_Normal: thruster02-01 and it still occurs. Attached are the u888 versions and the two tables that control or have any data regarding the thrusters. | ||||
| Tags | No tags attached. | ||||
|
2008-11-28 10:11
|
|
|
|
Moar info: Increasing the buffer zone on the left side of the image does increase the distance at which the flickering edge appears. But when the apparent size of the thruster becomes small enough, the flickers never fail to appear. You rarely see them with capital ships because they are often close enough that the flickers don't appear, but the apparent size of the thrusters is the real deciding factor here, not the size or class of ships. Idle speculation: Even though the primary texture has a buffer zone, the mipmaps generated by engine/GPU drivers might at some point have non-black pixels on the first pixel column from left. I suspect but haven't tested that disabling mip-mapping and possible anisotropic filtering might reduce the flickering, but for obvious reasons that shouldn't be required even if it worked. Personal opinion: Even if it were fixable by mutilating the content, I wouldn't call it a "content issue". I can't see any reason why the code should make the flickers appear even if the thrusters were at the immediate left edge of the texture. In fact I think it's definitely an engine issue if it does that, even if adjusting the content can reduce the phenomenon to varying extent. Nevertheless... if it is fixable by content, I suspect the only way would be to construct a custom mip-mapped DDS file where you can manually edit a black line to the first column of each mip-map. GIMP DDS plugin does not support this, however, and I'm definitely not certain if even this could work. Code based solution would be better though. EDIT: In response to driver profile thingy: GeForce 8800 GT here, tried making a profile for FS2_Open nightly r4956 and I still see the flickers same as with the global profile. EDIT 08-12-05: This just in - NVidia's Full Screen Anti-aliasing seems to be what triggers this behaviour on my system, so it might move it into a driver bug category. I'm still mystified why anti-aliasing would have such an effect on a 2D texture rendered on a plane. Ad far as I knew, shouldn't anti-aliasing just remove jagged edges of models and leave the textures well alone...? |
|
|
Having a "custom" DDS file won't work. I made a mip mapped file that had one black to the smallest, no go. Even stitching (nVidia DDS Tools) hand resized had the same issue. Unchecking the launcher flag "Enable mipmapping" still produces the flicker as well, regardless if the file has a mip map or not. AA and AF on my end have no apparent effect on the issue. Speaking of, and to address "It must be your drivers" replies (though there haven't been any yet): ATI 9800 GT 1024 x2 SLI. Drivers: 177.92 - 180.48 for WinVista32 Driver Settings (Global) Anistropic Filtering: Application Controlled Antialiasing - Gamma Correction: On Antialiasing - Mode: Application Controlled Antialiasing - Transparency: Supersampling Conformant Texture Clamp: Use Hardware Error Reporting: Off Extension Limit: Off Force MipMaps: None Maximum Pre-Rendered Frames: 3 Multi-display/Mixed GPU Acceleration: Single Display Performance Mode Texture Filtering - Anistropic Sample Optimization: Off Texture Filtering - Negative LOD Bias: Clamp Texture Filtering - Quality: Quality Texture Filtering - Trilinear Optimization: Off Threaded Optimization: Auto Triple Buffering: On Vertical Sync: Use the 3D application setting Launcher Video: OpenGL 1680 x 1050 32-bit Trilinear Anistropic Filter: Off Anti-Aliasing: Off 4. High Command line: C:\Games\FreeSpace2\fs2_open_3_6_10r.exe -mod FSU -spec -glow -env -mipmap -nomotiondebris -missile_lighting -normal -cache_bitmaps -dualscanlines -targetinfo -orbradar -rearm_timer -ship_choice_3d -weapon_choice_3d -3dwarp -no_ap_interrupt -snd_preload -nomusic -loadallweps -fps -ambient_factor 10 -no_emissive_light -spec_exp 11.0 -spec_point 0.6 -spec_static 0.8 -spec_tube 0.4 -ogl_spec 30 -clipdist 0.1 fov 0.55 EDIT: But if I make a profile specifically to fs2_open_3_6_10r.exe that is identical to the Global, guess what goes away? Can some one explain something before I start gnawing on electrical cables please? |
|
|
Did some testing with the latest trunk. It will only seem to happen on my laptop Geforce 8600 mobile, if I set the Anisotropic Filtering parameter to something greater than 2x. I don't know if the anisotropic filtering parameter is set through the launcher, since I see the slider, and it's set at 16x, but this issue doesn't appear then. |
|
|
Interestingly enough, I can force my driver settings to a high AF without a problem. I can even set a high AA, providing that I choose to let it "Enhance Application Settings". If it Over Ride or I set the slider in the launcher, flickers occur. |
|
|
I thought the launcher sliders didn't do anything currently. Or is that just the AF one? |
|
|
Few notes: -mipmap only does something when used in conjunction with -img2dds and when the a texture is actually qualifies to be compressed (must be power-of-2 JPG/PCX/TGA/ANI to qualify). The engine will not create mipmaps if they don't exist (aside from the -mipmap/-img2dds thing). I believe it might be possible to force mipmap generation in some graphics drivers, but this creates problem for the engine since it is then unable to set texture filtering properly. If a DDS texture is in use which has mipmaps then they will be used for rendering. You can not disable the use of mipmaps in the game without changing the content. The AA slider in the Launcher isn't hooked into the engine. Other than adjusting a registry setting for the engine to eventually use, it doesn't do anything except take up space on the launcher UI. Only the AF slider actually does anything in the game. Some AA modes that graphics drivers are able to force could account for this problem, however I don't believe that should be typical behavior. If that is the cause then it should be a bug. It might very be just a side-effect in the drivers, or it could have a strange code bug on our side (possible, but less likely). |
|
|
May as well be closed. Since Echelon9's fix in Commit 5328, no amount of effort on my part is able to reproduce this issue. |
|
|
Marking as feedback on Zacams advice. Can anybody reproduce this issue? Recommending closed (fixed) if not (original reporter can no longer reproduce). |
|
|
Reproduced, using 3.6.10 RC3, 3.6.10 Media VPs. Using NVidia Quadro FX 3500, NVidia ForceWare 182.65. :( Didn't think much of it till I saw this report, but there it is. |
|
|
Any chance you could try with the newer 186 driver series? |
|
|
I'm back to having this issue with the 190.38 drivers. I didn't, until I opened up the nVidia Control panel. I previously had a manually created nvapps.xml profile that I saved from my previous driver installation. I'm thinking that this is less an engine issue (though there is SOME issue at hand, and the thruster effects system needs something better than it currently has) and more a driver/CP package issue. *Edit: And what happened was that the AA reverted back to "Over ride Application Settings". Setting it back to "Enhance Application Settings" made it disappear again. |
|
|
The graphics and OpenGL code have had so many changes since 2009, that it's hardly worth keeping a specific, intermittent and possibly resolved bug open. If there are current graphics corruption, please report in new bugs. |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2008-11-28 10:11 | Zacam | New Issue | |
| 2008-11-28 10:11 | Zacam | File Added: effects.7z | |
| 2008-11-28 10:33 | Herra Tohtori | Note Added: 0010295 | |
| 2008-11-28 10:55 | Zacam | Note Added: 0010296 | |
| 2008-11-28 11:16 | Zacam | Note Edited: 0010296 | |
| 2008-11-28 11:18 | Zacam | Note Edited: 0010296 | |
| 2008-11-28 11:18 | Herra Tohtori | Note Edited: 0010295 | |
| 2008-11-28 11:29 | Herra Tohtori | Note Edited: 0010295 | |
| 2008-12-05 20:49 | Herra Tohtori | Note Edited: 0010295 | |
| 2008-12-05 21:07 | Herra Tohtori | Note Edited: 0010295 | |
| 2009-02-10 02:20 | phreak | Note Added: 0010665 | |
| 2009-02-23 10:30 | Zacam | Note Added: 0010687 | |
| 2009-02-25 20:21 | chief1983 | Note Added: 0010693 | |
| 2009-02-26 00:12 | taylor | Note Added: 0010694 | |
| 2009-06-24 02:31 | Zacam | Note Added: 0010989 | |
| 2009-06-25 06:50 | portej05 | Note Added: 0010992 | |
| 2009-06-25 06:50 | portej05 | Status | new => feedback |
| 2009-06-27 08:01 | Galemp | Note Added: 0011007 | |
| 2009-06-27 18:49 | chief1983 | Note Added: 0011008 | |
| 2009-09-12 04:31 | Zacam | Note Added: 0011175 | |
| 2009-09-13 18:56 | Zacam | Note Edited: 0011175 | |
| 2012-04-03 13:45 | Echelon9 | Note Added: 0013450 | |
| 2012-04-03 13:45 | Echelon9 | Status | feedback => resolved |
| 2012-04-03 13:45 | Echelon9 | Resolution | open => fixed |
| 2012-04-03 13:45 | Echelon9 | Assigned To | => Echelon9 |
