09 March 2019, 16:40 | #261 |
WinUAE developer
Join Date: Aug 2001
Location: Hämeenlinna/Finland
Age: 49
Posts: 26,518
|
Above config in shader feature is now implemented.
Because it is too late to change config when shader gets loaded (screen config is already in use, changing config at this point would cause screen reinit which would be annoying), shader is parsed separately when config is loaded or shader is changed. Only first 15 lines are checked for "string winuae_config : WINUAE_CONFIG =". |
17 March 2019, 15:57 | #262 |
Registered User
Join Date: May 2017
Location: EU
Posts: 342
|
Thanks Toni, i'm sure it will find some good use. Shaderpack for the 4.2 will be posted soon.
Meanwhile i played around a bit with the "Warmer Colors" shader, Retro-Nerd had a good idea to remove tinting and it's implemented now. The shader is a bit heavier, but not too much. |
17 March 2019, 16:03 | #263 |
Missile Command Champion
Join Date: Aug 2005
Location: Germany
Age: 52
Posts: 12,447
|
Awesome, this looks much better now. Many thanks for this update.
|
17 March 2019, 20:51 | #264 |
Registered User
Join Date: Nov 2009
Location: Legoland
Age: 45
Posts: 1,461
|
Thank you too!
|
10 May 2019, 00:47 | #265 | |
Registered User
Join Date: Jul 2008
Location: Netherlands
Posts: 485
|
Quote:
Not sure if it's possible, but is there a chance to include the Color Space options and CRT color profiles in the CRT-Trinitron-HiRes-SmartRes shader for WinUAE? (From post #190 here? ) I'm currently manually hardware switching my monitor from DCI-P3 to sRGB when going to WinUAE. Hopefully it's a possibility... And if it is, could the Gamma Out option be included too then (I like it at 2.25!). Cheers! |
|
10 May 2019, 22:58 | #266 | ||
Registered User
Join Date: May 2017
Location: EU
Posts: 342
|
New shader-pack.
Quote:
Quote:
Last but not least, the new shader-pack is here, works best with WinUAE 4.2 or above as "smart-res", with earlier versions the shaders work as lores compatible. Smart-res means the internal X/Y resolution is sortoff autodetected and games should work with shaders regardless of the settings used. The shaders should be D3D9/11 compatible, which means single pass / PS 3.0 limitations, but the selection is decent. If hires/double line mode is used some garbage might show up at the WB or loading screen, no way around that currently. Edit: fixed color altering shaders. Last edited by guest.r; 13 May 2019 at 14:21. |
||
11 May 2019, 04:09 | #267 |
Registered User
Join Date: Nov 2009
Location: Legoland
Age: 45
Posts: 1,461
|
Awesome this new shader pack, thanks guest.r! It works incredibly well here.
|
12 May 2019, 10:19 | #268 | |
Registered User
Join Date: Jul 2008
Location: Netherlands
Posts: 485
|
Quote:
Thanks for working on this and the new shader pack The color space options unfortunately don't work properly. Choosing DCI-P3 or other non-sRGB space options results in incorrect colors. I tested both DCI-P3 and AdobeRGB. The below picture gives an impression of how it goes wrong. The first is without colorspace.fx (correct). The second below it is with colorspace.fx at position 1 in shader panel and with DCI-P3 (color space = "1") and profile 4 configured in the shader file. FWIW I would describe the error as being desaturated with a very green hue. The saturation should at least be similar to the first picture and the white point should look different too. (click for larger) |
|
12 May 2019, 20:52 | #269 |
Registered User
Join Date: May 2017
Location: EU
Posts: 342
|
Does it correctly work with this version?
http://eab.abime.net/showpost.php?p=...&postcount=235 I don't have an Adobe RGB or DCI-P3 capable display, so i can't do proper testing. The WinUAE shader is basically a copy/paste of the original so i need a clue on this. |
12 May 2019, 21:38 | #270 |
Zone Friend
|
Really like the new shaders guest.r - Workbench seems a little funky though:
Is this related to the "If hires/double line mode is used some garbage might show up at the WB or loading screen" thing you mentioned. I have the WinUAE display set as 'Resolution autoswitch: Always On' so the resolution settings and line mode settings are greyed out but I believe the default workbench resolution is lores anyway (Running v4.2.1 b4 of WinUAE by the way @ 1920x1080 fullscreen) |
12 May 2019, 22:11 | #271 |
Registered User
Join Date: May 2017
Location: EU
Posts: 342
|
I checked and my WB screen seems to run in hi-res with resolution autoswitch. But yes, i had this sort of "garbage" in mind.
It doesn't happen with all shaders though. |
12 May 2019, 23:00 | #272 | |
Registered User
Join Date: Jul 2008
Location: Netherlands
Posts: 485
|
Quote:
Yes it works fine with that version. To illustrate, in the attached zip I included two screenshots, one from Retroarch with PUAE and color space set to DCI-P3 with profile 4 (correct shader colors) and same setting with WinUAE shader (which gives wrong output colors). Should make the difference quite clear, even when viewing on sRGB screen |
|
13 May 2019, 14:22 | #273 |
Registered User
Join Date: May 2017
Location: EU
Posts: 342
|
Thanks for the confirmation, it should be fixed now in the pack couple of posts above.
|
14 May 2019, 23:43 | #274 | |
Registered User
Join Date: Jul 2008
Location: Netherlands
Posts: 485
|
Quote:
I've thoroughly been enjoying the shader, using the Hires version currently. The gamma option works out quite nicely. There's only one thing with the color profiles, there's a noticable differerence in color output between DCI-P3+profile4 from WinUAE, versus DCI-P3+profile4 in RA (ofc, using the correct shader files). The RA version is pretty much identical to the CRT that I have running here, but the WinUAE output for the "same" setting is more washed out. At first I thought it's a saturation thing, but it doesn't help raising saturation in the shader. As an example, I made a comparison shot between the output from the two. Of course for both the colors will look wrong when viewed in sRGB, but it's about the difference. If you compare the Wolfchild title sceen logo for both you can see the red being more washed out in the WinUAE output. This also becomes apparent when running Ruff and Tumble for example, the colors are more washed out than on the CRT or the RA output. Any ideas? |
|
15 May 2019, 07:28 | #275 | |
Registered User
Join Date: Feb 2009
Location: Poland
Posts: 93
|
WinUAE config
Quote:
Wow, could you send me WinUAE config that you use? |
|
15 May 2019, 15:38 | #276 | |
Zone Friend
|
Quote:
Could you recommend the settings you use in the 'display' and 'filter' tabs that work best with your lovely shaders ? |
|
15 May 2019, 20:40 | #277 |
Registered User
Join Date: Jul 2008
Location: Netherlands
Posts: 485
|
@guest.r
To add to my previous post, it seems the largest difference between the DCI-P3 + profile 4 output from RA versus WinUAE is in Reds and Yellows. These are slightly "washed out" in the WinUAE shader. For some reason it's mainly these that stand out. I've attached a better example for the reds, it's the Ignition table (lots of red shades ;-) ) both with the RA shader and the WinUAE shader. Both look really great, but if you switch back and forth you'll notice the RA shader output having much more vivid colors in the reds. Again these are DCI-P3 snaps, so will look slightly different in sRGB, but the difference between the two is still well noticable. Also, if you look at the white / grey on the inside of the "IGNITION" letters, then you'll notice a particular difference between both also. I wonder whether the custom white points are implemented in the same way for both shaders? I'm asking because from the calibration sessions I did back then, I know that especially yellow is very sensitive to the specific custom white point. So if the custom white point implementation is different from the RA one, it could be related.. Just to be sure, I calculated the profile4 matrix according to the formulas on this page Computing RGB-to-XYZ and XYZ-to-RGB matrices. The matrix has some minor deviations from the one in the shader, but I'm guessing this is because of limitations in Excel (apparently INVERSEMAT uses 16 digits in EXCEL). FWIW, this is the table that I get. I assume you have some better tool to do these kind of calculations Anyways, I think I mainly would like to understand the why for the difference in output for both shaders. If you have some idea that would be welcome.. Code:
0,392267594 0,351133585 0,166598808 0,209415298 0,725676075 0,064908626 0,01606135 0,093635623 0,850303005 Code:
0,392268 0,351134 0,166599 0,209415 0,725676 0,064909 0,016061 0,093636 0,850303 |
15 May 2019, 22:29 | #278 | |
Registered User
Join Date: May 2017
Location: EU
Posts: 342
|
Quote:
Regarding 'display' and 'filter' tab settings i strongly recommend the vertical/Y multiplier to be at least 3x with crt shaders for better scanlines. Otherwise it's OK to use your prefered aspects etc. @Dr. Venom: The color profile matrices are a copy/paste from the RA version, so i'm a bit curious about the differences. Can you test it with a different shader (WinUAE) and without masks? The WinUAE "trinitron" shaders use a slightly different scanline formula too (saturation is affected). And gamma is different... I'm almost sure we can exclude matrices or white point issues, because there should be none. |
|
16 May 2019, 11:25 | #279 | |
Registered User
Join Date: Jul 2008
Location: Netherlands
Posts: 485
|
Quote:
Below image compares saturation setting 1.0 versus 2.0 in the CRT-Trinitron-Hi-Res shader and the RA shader. There's no change in RED saturation for the WinUAE shader when going from 1.0 to 2.0 (while blue does, so it's not completely broken..). The RA shader shows the proper response to the saturation increase. (click for large) |
|
16 May 2019, 19:26 | #280 | |
Registered User
Join Date: May 2017
Location: EU
Posts: 342
|
Quote:
|
|
Currently Active Users Viewing This Thread: 1 (0 members and 1 guests) | |
Thread Tools | |
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Multipass shaders | craig64 | support.FS-UAE | 21 | 08 December 2020 11:01 |
HLSL/CG Shaders and xBR | Enverex | support.FS-UAE | 11 | 19 July 2013 16:44 |
fx Shaders in WinUAE 2.6.0 | crazy46guy | support.WinUAE | 8 | 16 June 2013 14:30 |
[FS-UAE] Optimized or corrected Amiga Game Database Entrys | nexusle | support.FS-UAE | 3 | 24 August 2012 21:30 |
Subpixel-corrected lines and polygons on Amiga | Scali | Coders. Asm / Hardware | 9 | 11 January 2012 12:37 |
|
|