Post-processing effects test (SSAO/HDR/Bloom)

[zoot]

Did a before-and-after GIF with Shield Bearer's maps just for fun:

(Sorry for any epileptic seizures this may have caused.)

Edited August 18, 2012 by zoot

[historic_bruno]

Thanks, that's very illustrative What's with the AO(?) artifacts on the farmstead though?

[Potter]

@zoot, cool! That really shows the amazing differences ...

[Shield Bearer]

Thanks, that's very illustrative What's with the AO(?) artifacts on the farmstead though?

Fixed, I had used the another AO map for it And I added specularity to the roof tiles too.

[fabio]

I was finally able to test it with my Radeon 1600 under Linux. Unfortunately as soon as I enable preferglsl I still get:

r300: ERROR: FS input generic 20 unassigned, not enough hardware slots (it's not a bug, do not report it).

and I also see the texture flickering.

When enabling all the new options I can see the wind trees and that looks nice! Is there a way to enable it without enabling other effects?

[myconid]

Unfortunately I had to undo one of the changes we'd made, because it broke something completely unrelated on my PC - there seems to be a bug with the liveness analysis in fglrx's GLSL compiler, though I couldn't think of a workaround. Sorry, man.

To see the windy trees, you have to be in GLSL mode (generally, you can use ARB effects in GLSL mode, but not GLSL effects in ARB mode). If you can put up with the flickering, you could try setting materialmgr.PARALLAX_DIST.max = 0, and it may help you a little with the errors.

Anyway, wraitii's ARB versions should be on the way soon, so that should be the end of the problem.

[fabio]

Unfortunately I had to undo one of the changes we'd made, because it broke something completely unrelated on my PC - there seems to be a bug with the liveness analysis in fglrx's GLSL compiler, though I couldn't think of a workaround. Sorry, man.

To see the windy trees, you have to be in GLSL mode (generally, you can use ARB effects in GLSL mode, but not GLSL effects in ARB mode). If you can put up with the flickering, you could try setting materialmgr.PARALLAX_DIST.max = 0, and it may help you a little with the errors.

It doesn't help (also tried addind materialmgr.PARALLAX_HQ_DIST.max = 0 ).

There is also another problem when enabling preferglsl: the whole minimap is visible.

What about the flickering? Do you think it's a driver bug or a 0 A.D. bug?

Anyway, wraitii's ARB versions should be on the way soon, so that should be the end of the problem.

Nice! Do you ever plan to enable some effects by default? Eventually hwdetect.js could be used for forcing enabling/disabling on some card.

[myconid]

There is also another problem when enabling preferglsl: the whole minimap is visible.

I've been told about that before, I think by zoot. I'll have to look into it separately, as it's unrelated to the stuff I've been doing in this thread. A quick look revealed that the minimap/UI use ancient fixed-pipeline functions for everything, which will need to be replaced eventually.

What about the flickering? Do you think it's a driver bug or a 0 A.D. bug?

It might be a bit of both, tbh. It's related to the shadowmap implementation, which works for some people but not others, and it's also inconsistent between ARB/GLSL even though they share most of the code.

Nice! Do you ever plan to enable some effects by default? Eventually hwdetect.js could be used for forcing enabling/disabling on some card.

A combination of hardware detection with a working options menu will be best, I think.

[Wijitmaker]

Did a before-and-after GIF with Shield Bearer's maps just for fun:

(Sorry for any epileptic seizures this may have caused.)

Cool pic - I'm surprised the AO didn't have more of an effect. For example the upper floor of the temple building in the top right. I would think that the interior would be much darker.

[zoot]

Note that "normal" shadows are disabled on that pic, so that particular room should look darker on a proper PC.

[Zaggy1024]

Myconid, I looked at the AO that Shield Bearer committed (I think from your repository), and it looks like you really need to turn your samples up a bit. I set them to 20 for the Roman buildings, and I think that's about the minimum to keep there from being really visible noise.

[myconid]

Myconid, I looked at the AO that Shield Bearer committed (I think from your repository), and it looks like you really need to turn your samples up a bit. I set them to 20 for the Roman buildings, and I think that's about the minimum to keep there being really visible noise.

I was hoping people would use the script to bake new textures. I guess if I modify the script a little, it can then automatically re-bake the textures using the existing UV sets, so it'll be a matter of just replacing a few images...

[Deiz]

Got some graphical weirdness to report. Examples: 1, 2.

I've seen it on a variety of structures (Pureon has seen it too, apparently. Can anyone else confirm?) so I don't think it's an issue with any particular model, but some of the graphics code that's presumably been touched recently.

Presently running with a simple config: windowed, fancywater disabled, vsync on.

It's a rather strange issue. It occurs randomly and persists for a minute or two. Oddly, the textures are mapped correctly when zoomed in, and then promptly become corrupted like the above when zooming back out. Since it's zoom-dependent, it seems to me it might be a mipmapping issue, but I'm wholly unfamiliar with the graphics and rendering code.

[zoot]

Curiosly, in both cases the artifacts are player colored.

[myconid]

Looks like it's something to do with the UV mapping.

(zoot, the alpha channel is used for player colour, so if the UV mapping is weirding out, the player colour also appears in funny places.)

Since this happens with distance, the only things that are relevant are mipmapping (which I haven't touched) and the LOD stuff.

The latter you can mess with by changing the materialmgr.PARALLAX_DIST.max value in the config to something smaller, and see if the artifacts happen closer.

Btw, does this happen in ARB mode too or is it only in GLSL mode? If it's only in GLSL mode, can you roll back to an old revision and see if you get similar problems with GLSL? (that should give me a hint about where to look for a fix)

[Deiz]

default.cfg currently has preferglsl = false, so thus far I've only experienced the issue in ARB mode.

[myconid]

Hmm. And I'll assume gentangents=false, too.

I think I know where to look.

[myconid]

Try that now.

[Zaggy1024]

Myconid, can you post the settings you're using for baking the AO textures? I tried recalculating all the Roman buildings' AO textures and for some reason they're way darker than the ones Shield Bearer committed.

Also, is there some way you can make the script load each actor and its prop actor files, so it can import all the props along with the main model, so the AO includes the props?

[myconid]

I'm just using the defaults for everything: no normalisation, attenuation distance 10, constant QMC, no falloff. Blender v2.63a.

As for loading the props, that's unlikely to happen. The props information is stored in the actor files, which means the script would have to parse the xml, find the props, load them up and attach them in the right places. I think only the Greek buildings have a problem because their main structures are divided up into pieces. The best solution is to store those models in one piece and duplicate some of the textures instead (or combine them in a bigger texture).

[Deiz]

Can't definitively say whether r12507 fixed the aforementioned UV issue, but I played an hour-long match last night with no graphical glitches. There seem to be quite a few people that play SVN builds regularly, so I suspect no further reports of graphical weirdness means the issue's gone.

Edit: Well, so much for that theory. At present I've actually got it occurring at all zoom levels. It flickers through various textures (the two glitched ones are very similar):

Lighter

Darker

Normal

[historic_bruno]

A couple of us encountered buggy water, LOS, status bars, and minimap in today's staff multiplayer game. They all appear to blacken suddenly during the game. At least the minimap will flicker a bit as I change zoom, from dark black to light black. This is notable as I encountered no issues like this last week (same settings, all new graphics options enabled), but thinking about it, those were all land maps and this one was a water map:

[wraitii]

I'm too having buggy GUI when using the developper interface. Didn't happen before, so a recent change must have broken something.

[myconid]

Deiz: At first I thought it might be using the wrong UV set for texturing, though from your new screenshots it looks like it's in fact being completely random. Then there's the lighter/darker thing, where the mapping is similar, but the lighting changes. So it must be something in the modelrenderer - for some reason, the parameters are sometimes not bound correctly to the shaders! Will look into this (again).

historic_bruno/wraitii: I need more info... I can guess what might be up with the water, the minimap was already bugged in GLSL mode (or is this ARB mode). What's wrong with the LOS and the status bars?

[historic_bruno]

historic_bruno/wraitii: I need more info... I can guess what might be up with the water, the minimap was already bugged in GLSL mode (or is this ARB mode). What's wrong with the LOS and the status bars?

This should be GLSL mode since preferglsl=true, all advanced graphics options enabled.

Look closer at the screenshot, you can see the health bars on the selected or hovered units are all black (not in the unit details panel, but in the game view). The LOS problem might actually have been the water, I could see it out of the corner of my eye, maybe there was no LOS problem. Unfortunately there's no video