New CRT shader from Guest + CRT Guest Advanced updates

ynnad4 · 6 April 2022 19:58

Greetings. I have been testing up your whole CRT guest advanced progress within the couple of presets you have already available.

I’d like to know it could ever be possible to add up within any of the current presets or a new one that could contain this particular effect known as Rainbow banding that was present on Sega Genesis console games. There’s a thread already talking about this:

For example in that particular reply, there’s a Sega Genesis demostration with RF connection which exposes the colorful red lines coming along with green and blue. The whole thread overall exposes different variants I did over time with public modded versions from mame_hlsl alone or with some few other presets like ntsc adaptive.

Would be great if that particular effect can be applied to other shader/presets that come along with composite-blurry looking just like some of you already have the capability of dithering presence. I unfortunately couldn’t find any other shader/preset that could apply accurately this effect just like it is for the real console.

Not to mention the fact such mame_hlsl preset along with others like tvout-tweak can allow to tweak intensity of the rainbow effect but also how specifically separate are each colorful line from each other, which helps out greatly in accuracy compared to the already existent NTSC shader/presets that to an extent deliver these Rainbow artifacts but in the same time far from being that accurate just like mame_hlsl.

I hope you can reply and thank you a lot for such a great job within your updated shaders.

Nesguy · 6 April 2022 20:06

Color issues aside that looks very good. Settings, please!

hunterk · 6 April 2022 20:28

you should be able to replace the ntsc-adaptive passes in guest’s presets with the ntsc passes from mame_hlsl with your settings in most cases.

Syh · 6 April 2022 21:55

Honestly quiet a few of MAMEs passes can be used with other presets. (At least like 3-4 off the top of my head.)

ynnad4 · 6 April 2022 22:18

any ideas ? i may appreciate any preset you could know in how to mix those two passes with:

shader0 = "../crt/shaders/mame_hlsl/shaders/mame_ntsc_encode.slang"
filter_linear0 = "true"
wrap_mode0 = "clamp_to_border"
mipmap_input0 = "false"
alias0 = "NTSCPass"
float_framebuffer0 = "false"
srgb_framebuffer0 = "false"
scale_type_x0 = "source"
scale_x0 = "1.000000"
scale_type_y0 = "source"
scale_y0 = "1.000000"

shader1 = "../crt/shaders/mame_hlsl/shaders/mame_ntsc_decode.slang"
filter_linear1 = "true"
wrap_mode1 = "clamp_to_border"
mipmap_input1 = "false"
alias1 = "ColorPass"
float_framebuffer1 = "false"
srgb_framebuffer1 = "false"
scale_type_x1 = "source"
scale_x1 = "1.000000"
scale_type_y1 = "source"
scale_y1 = "1.000000"

ntscsignal = "1.000000"
shadowalpha = "0.100000"
avalue = "0.000000"
bvalue = "0.000000"
pvalue = "1.100000"
scantime = "47.700073"
notch_width = "3.450001"
ifreqresponse = "1.750000"
qfreqresponse = "1.450000"

Nesguy · 7 April 2022 14:25

Here’s gamma 1.0/1.0, blend mode 2 with a few tweaks (mask strength 100%, etc). Looks fantastic! The edges and beam variation look very natural.

What settings did you use to produce the first screenshot you posted?

guest.r · 7 April 2022 15:30

I used blend mode 2.0 and increased ntsc resolution. The downside is that it doesn’t blend dithering with games which use this technique, but should look very nice with nes… games.

Some posts above i replied that ntsc blending looks better with lower gamma and why 2.0 is defaulted instead of 2.4. Circumstances connected with bloom/glow passes which tend to share the same gamma value are such, that if glow/bloom/halation is used it looks quite different with low gamma. So 2.0 is a compromise, but 1.0 can be used np with presets which don’t rely on some lighting effects.

datjocko · 8 April 2022 14:17

Is there any reference to the use case/intention of each mask option? Aesthetically I’m fond of mask 1 but I’m not 100% on when to use which options.

guest.r · 8 April 2022 15:18

Hey,

masks are currently like WYSWYG and imo it shouldn’t be too hard to pick your fauvorite one or change some options.

Nevertheless, a short description:

-1.0: no mask
0.0: a dense magenta-green mask
1.0: Lottes RGB mask with a some sort of slotmask
2.0: Lottes RGB mask
3.0: Lottes - a bit enlarged and shifted RGB mask
4.0: Lottes - 2x enlarged and shifted RGB mask, looks nicer with 4k+
5.0: a dense magenta green mask (trinitron mask controls)
6.0: RGB mask (trinitron mask controls)
7.0: BW mask (trinitron mask controls)
8.0: BWW mask (trinitron mask controls)
9.0: magenta-green-black mask (trinitron mask controls)
10.0: RGBX mask (trinitron mask controls)
11.0: Red-Yellow-Cyan-Blue mask (nice for 1440p+)
12.0: Red-Magenta-Cyan-Green mask (similar to 11.0, but for BGR panel layout)
13.0: RRGGBBX mask (looks nice with 4k+)

Slotmask options are a bit more complicated to describe, it’s desired to match the slotmask width with the used regular mask width and then increase both strength values.

Some parameters are a bit self explaining i guess, don’t be afraid to experiment a bit…

RealNC · 8 April 2022 17:00

Also note that if you change the “mask shift/stagger” value for a mask, it changes dramatically. For example mask 2 with stagger 1 looks similar to mask 3 with stagger 0, but a bit better for 480p content.

Nesguy · 8 April 2022 22:20

Mask 0 with stagger set to 1 produces an excellent dot mask pattern.

RealNC · 9 April 2022 08:12

My monitor doesn’t like that pattern. The inversion artifacts are the most extreme I’ve ever seen:

Nesguy · 9 April 2022 14:44

What kind of monitor is that? That’s typically only a problem for very old displays, even a cheap 1080p work monitor shouldn’t have any such problems these days. EDIT okay so this can still be a thing on recent monitors, apparently, particularly the ones that boast 120+ Hz refresh rates.

Probably won’t help, but have you tried flipping the mask?

RealNC · 9 April 2022 14:21

All LCD monitors have inversion artifacts. There doesn’t exist any LCD panels that don’t have it. What kind of artifacts depends on which inversion scheme the monitor uses. Mine (a 165Hz IPS) just happens to exhibit artifacts with that particular mask pattern, that’s all.

I don’t know what that means

Nesguy · 9 April 2022 14:43

Just saying that the only time I’ve ever encountered anything like that was with a 10+ year old display. Guess I’ve had good luck in that regard.

try adding this line

mask_layout = "1.000000"

RealNC · 9 April 2022 14:48

Oh you meant BGR. No it doesn’t fix it. It reduces it a bit though.

Nesguy · 9 April 2022 14:52

Yeah that was a shot in the dark. IIRC the only way to eliminate it completely is to reduce the mask strength until the inversion artifacts go away.

rafan · 10 April 2022 15:04

I like to decrease the step value in the shader for some parameters, but when I do this the shader just quits (doesn’t get applied).

for example for bloom default is:

#pragma parameter bloom "          Bloom Strength" 0.0 -2.0 2.0 0.05
#define bloom         global.bloom     // bloom effect

but when changing the step value to 0.01, like below, the shader doesn’t get applied any more:

#pragma parameter bloom "          Bloom Strength" 0.0 -2.0 2.0 0.01
#define bloom         global.bloom     // bloom effect

this also happens with some other parameters, like bright boost dark pixels, lowering step value isn’t possible. In older shaders this used to cause no issues.

Any idea what’s happening? Above was tested with hires shader and vulkan driver.

guest.r · 10 April 2022 16:25

This is a parameter duplication issue, when a parameter line resides in two separate shader files. You need to adjust the step value in the main shader file as well as in the deconvergence pass. Then it should work.

PizzaLaser25 · 12 April 2022 18:36

Hey guys - hope everyone is having a nice week. I read through most of this chain just to increase my own knowledge of shader stuff in general, and came away with some questions. I hope these aren’t dumb questions but maybe there are other lurkers or readers who might have similar questions, so I’ll go ahead and ask.

I am struggling to understand: what are the key differences between what (I think) are the three main shaders being discussed here:

SM
Advanced
Dr. Venom

I suppose what I am asking is, does each have a specific “character” or “quality” that differs from the others? Like were each of them designed independently (by Guest) for different purposes/intents? I know they are all configurable. I know that the scanline appearance can be changed in all of them, and the masks - is it possible to use SM and get very similar results to DV? Or do they have inherent qualities to them? Like SM blends things in a different way than Advanced than Dr. Venom, etc.

Said another way, is there a reason why someone would or should pick one of these over the others?

Follow-up - I understand that the NTSC (incorporating NTSC effects) and hires (intended for use on content above 480p) versions of the shaders are variations of the original underlying shader. What about the fast and fastest variations - do those just take the shader in question and strip out some configuration options in order to reduce the passes and improve performance?

Fingers crossed these aren’t dumb questions!