forums.PPSSPP.org
16bit colorspace rendering? [Somekind of Android bug] - Printable Version

+- forums.PPSSPP.org (https://forums.ppsspp.org)
+-- Forum: PPSSPP - Playstation Portable Simulator Suitable for Playing Portably (/forumdisplay.php?fid=1)
+--- Forum: Development (/forumdisplay.php?fid=3)
+--- Thread: 16bit colorspace rendering? [Somekind of Android bug] (/showthread.php?tid=919)



16bit colorspace rendering? [Somekind of Android bug] - VIRGIN KLM - 02-05-2013 05:31 PM

Hey everybody!

I have that thing that I can't decide if it's a bug or something gone wrong but I noticed that other people don't have it.
It seems that the emulator gets lauched at 16bit colorspace (I can see posterizing effects).
When I say that I notice it in the emulator I mean as soon as I launch the application, Instead of smooth color transitions in this blue-ish gui I have weird posterizing between the shades.
The same issue applies when I launch games which is a bug I don't understand. If you need any images/log tell me, but I think they wouldn't be really usefull.

Some info:
HTC EVO 3D (GSM)
Stock Gingerbread ROM
No Overclocked/Tweaks applied.


RE: 16bit colorspace rendering? [Somekind of Android bug] - m45t3r - 02-10-2013 12:01 AM

(02-05-2013 05:31 PM)VIRGIN KLM Wrote:  Hey everybody!

I have that thing that I can't decide if it's a bug or something gone wrong but I noticed that other people don't have it.
It seems that the emulator gets lauched at 16bit colorspace (I can see posterizing effects).
When I say that I notice it in the emulator I mean as soon as I launch the application, Instead of smooth color transitions in this blue-ish gui I have weird posterizing between the shades.
The same issue applies when I launch games which is a bug I don't understand. If you need any images/log tell me, but I think they wouldn't be really usefull.

Some info:
HTC EVO 3D (GSM)
Stock Gingerbread ROM
No Overclocked/Tweaks applied.

I think this problem is related to Gingerbread and is not a problem of the emulator. Maybe it's related to this http://www.curious-creature.org/2010/12/04/gingerbread-and-32-bits-windows/ ?


RE: 16bit colorspace rendering? [Somekind of Android bug] - thedicemaster - 02-10-2013 11:26 AM

nope.
4.1.2 here and I see exactly the same.


RE: 16bit colorspace rendering? [Somekind of Android bug] - ssshadow - 02-10-2013 01:46 PM

Post a screenshot


RE: 16bit colorspace rendering? [Somekind of Android bug] - VIRGIN KLM - 02-11-2013 08:48 PM

I made a bit more investigation to the matter.

(Sorry for the bad compression of the images, my phone for some reason refuses to take HQ screenshots and compresses them badly but still the issue is noticable, imagine it visualy like 10 times worse realtime.)

Here it how it looks:
[Image: 20130206000748.jpg]
Here is how it SHOULD look (compare the lower left side):
[Image: 20130206000911.jpg]

Looks like there is somekind of a behaviour set by default on the emulator for it's colorspace and it stays to this setting UNLESS the game you load overrides it and requests a 32bit colorspace rendering which also fixes the UI untill you relaunch the emulator. For example, once I launch the emulator on my phone I notice the banding/16bit effects on the UI, if I launch for example Final Fantasy Type 0 the colorspace won't change and it will stay on 16 bit with the banding effect. Now, If I do the same but run Digimon World Re:Digitize, I don't notice any banding, plus if I exit back to the UI the problem is fixed. Very weird and visualy annoying issue, looks you try to run a game on Windows when your video card stops responding due to overheat/failure.


RE: 16bit colorspace rendering? [Somekind of Android bug] - Henrik - 02-11-2013 10:20 PM

Bizarre. No idea how to fix that really, seems a bit like a driver bug?


RE: 16bit colorspace rendering? [Somekind of Android bug] - HD2MAX - 02-11-2013 10:44 PM

For my this issue since the addition color depth remember maybe from v0.5-5xx more or less


RE: 16bit colorspace rendering? [Somekind of Android bug] - VIRGIN KLM - 02-12-2013 11:50 AM

(02-11-2013 10:44 PM)HD2MAX Wrote:  For my this issue since the addition color depth remember maybe from v0.5-5xx more or less

Ha, yes, just re-installed v0.4 and the bug is not there. Hope this helps. I think the version it started happening was v0.5-550.

I have a feeling it has to do with that line in the changelog:
"Add enable/disable dithering. Respect chosen framebuffer color depth."

There may be some devices/hardware that restrict dithering on whatever case but allow 16bit rendering.


RE: 16bit colorspace rendering? [Somekind of Android bug] - Henrik - 02-12-2013 12:15 PM

Oh, yes - dithering could be the explanation. Will make sure it's enabled in menus soon.


RE: 16bit colorspace rendering? [Somekind of Android bug] - VIRGIN KLM - 02-12-2013 12:21 PM

(02-12-2013 12:15 PM)Henrik Wrote:  Oh, yes - dithering could be the explanation. Will make sure it's enabled in menus soon.

Great thanx! Could I ask what would be the advanatges and disadvantages on devices that can't handle dithering/colorspace changes and what games get affected by it? (just from the top of your mind)


RE: 16bit colorspace rendering? [Somekind of Android bug] - VIRGIN KLM - 03-28-2013 08:29 AM

Sorry to bump the thread but the problem is back again, possibly it wasn't actualy properly fixed at all.
I see color banding on Final Fantsy Type-0, not visible on Digimon World Re:Digitize or on the Windows builds.
Also I have a feeling that for some reason my phone's speed gets affected of some useless conversion of 16 bit to 24 bit to 16 bit, I don't have proof but I can, like, sense it.


RE: 16bit colorspace rendering? [Somekind of Android bug] - [Unknown] - 03-28-2013 01:40 PM

Possibly caused by this, then, if it recently came back:

https://github.com/hrydgard/ppsspp/pull/1081

-[Unknown]


RE: 16bit colorspace rendering? [Somekind of Android bug] - VIRGIN KLM - 03-28-2013 05:07 PM

Nope, the build I tested it dates before that merge pull request. Should I post screenshots? For some season I want somebody else with the game to get a savegame from gamefaqs that is 100% completed before the final battle, and go and check the cutscenes if he sees the same issue. I think it's an OGL ES 2.0 related issue.