Mega Bezel Reflection Shader! - Feedback and Updates

[Cyber]

@HyperspaceMadness

It probably would be good to try your settings straight in the latest guest shader to see if there is some difference inside the Mega Bezel or if it’s just what the shader looks like.

If you want to do a comparison it would be best to set the Mega Bezel ‘s aspect ratio to full, so that the scaling is similar when you load up guest’ s shaders with no Mega Bezel

I suppose I could do that when I get a chance. Not to sound ungrateful but when I have my preset settings dialed in and fine tuned, I almost always cringe when I see other users request or complain about certain features in the CRT-Guest Advance Shader.

Guest usually goes to work and comes back with results really quickly, meanwhile I continue fine tuning using the incumbent version, hoping that somehow, just maybe, whatever tweaks have been done would automagically enhance the look of my presets.

Then HSM Mega Bezel gets updated to the latest CRT Guest Advance and this sometimes happens.

Since you released V1 and Guest also released V1 Stable of CRT Guest Advanced, I also thought of releasing a V1 stable version of CyberLab Mega Bezel Death To Pixels Shader Preset Pack. This way, once a user matches V1 with V1 of Mega Bezel, the output quality should be locked in and guaranteed to look a certain way.

I think I’ll start adding a recommended version or tested version somewhere in my first post so that if anything starts looking wonky, users would immediately know which version they should probably revert to if they want things to look the way they should.

In this vein, I think it should be easy for users to have access to previous versions of the shader should they require a rollback to a prior revision.

I know the price of progress is high though. Let’s hope we can get to the bottom of this one sooner rather than later.

[guest.r]

@Cyber

Since you released V1 and Guest also released V1 Stable of CRT Guest Advanced, I also thought of releasing a V1 stable version of CyberLab Mega Bezel Death To Pixels Shader Preset Pack. This way, once a user matches V1 with V1 of Mega Bezel, the output quality should be locked in and guaranteed to look a certain way.

I understand your concerns, but testing is also important. I tested your preset with two versions:

Presets:

CyberLab__Ultimate_Virtual_Slot_Mask_CRT-1P2RTA__ADV.slangp

CyberLab__SNES__ADV.slangp

crt-guest-advanced-2022-02-10-release1 vs. current version :

Looks almost identical.

[Cyber]

@guest.r

I understand your concerns, but testing is also important. I tested your preset with two versions:

@guest.r

I didn’t notice notable differences though. Maybe you can test another preset which uses other features.

The 2 presets I used in my testing are my current

CyberLab__Slot_Mask_II_for_CyberLab_or_Core_Blargg_Composite_or_S-Video_Filter.slangp

preset and an unreleased version version that I’m currently working on.

Both are running with my custom

Blargg_NTSC_Turbo_Duo_SNES_PSX_S-Video_CyberLab_Special_Edition.filt

video filter preset.

Thanks for taking the time to test.

You can try it with the same preset and filter combination I used and see if you notice it.

All of my testing was done at 3840 x 2160 with presets in the

"MBZ__1__Advance_Full_Reflections\4K_Optimized_Presets" folder.

[bloodtype]

Anyone with the new M2 macbook systems try these shaders out yet? Would love to know how they perform. Some of the previous posts regarding using these shaders on apple hardware seemed like people were running into issues.

[JHorbach1]

@HyperspaceMadness Hello! Could you tell if I can replicate latest @guest.r settings with the MegaBezel? I tried to put the parameters in the NTSC shader but it won’t look the same as in MegaBezel, it works with guest standalone.

[HyperspaceMadness]

The latest release in this thread 1.0.002 has the latest guest integrated.

If you are getting discrepancies let me know, to test them set the Mega Bezel aspect ratio to full.

[guest.r]

From my observings the halation is applied quite strongly even with small parameter values, much stronger than in standalone ‘GDV’. Otherwise it’s appearing correctly.

Edit: seems to apply to bloom also.

[Cyber]

@HyperspaceMadness & @guest.r

It was very difficult if not impossible for me to recreate the look of my presets using even my same settings applied to standalone CRT-GUEST-ADVANCE but I did make some progress in narrowing down the issue.

Simply increasing the GSL Scanline Value from -1 to 1 or 2 seemed to bring back the woven slot mask pattern with every other dot being darker than the one above and below it and the slight “spikiness” returning to the white text on the blue background and also the contrast and sharpness returning.

On further analysis it seems as though the darker “dots” I’m referring to are the ones that lie near the outer (upper and lower) edge of the scanline, while the brighter “dot” in the middle is the one that lies near the center of the scanline. So it seems like a scanline dynamics or scanline type issue.

I took some screenshots so I’ll be cropping and uploading them as soon as I get a chance. I have a strong feeling that these might turn out slightly darker as well as smaller in file sizes than the ones that looked wrong.

It’s also interesting that @guest.r noticed virtually no difference between the versions he tested using my

CyberLab__Ultimate_Virtual_Slot_Mask_CRT-1P2RTA__ADV.slangp

preset because this preset features almost indistinguishable scanlines akin to a high resolution slot mask computer monitor.

In fact, @Nesguy even described that preset as “looks good, but where are the scanlines?”

My subsequent Slot Mask II presets sought to address that question.

So wherever and whenever the change in behaviour of the shader occurred, it seemed to affect the behaviour and brightness of the various scanline types.

CyberLab__Slot_Mask_II_for_CyberLab_or_Core_Blargg_Composite_or_S-Video_Filter.slangp + Blargg_NTSC_Turbo_Duo_SNES_PSX_S-Video_CyberLab_Special_Edition.filt

GLS 1

GLS 2

While this is definitely an improvement, the older version shader is still slightly sharper and clearer with the mask pattern being much more easily defined on the white text and the spacing inside of the letters and numerals appearing sharper, finer and tighter. To observe this zoom in and pay close attention to the number “9”.

There could have been a discrepancy in my testing because before I started my latest round of tests I noticed my RGB Video Filter was being auto loaded and not my S-Video Filter. I don’t remember switching filters between versions of HSM Mega Bezel Reflection Shader when comparing the two versions though so I will go back and verify but with the RGB Video Filter installed the sharpness is even closer to the older version.

Updated: 16-07-22

I redid all of my testing and the conclusions remain the same. In the new version of HSM Mega Bezel the GSL Scanline levels changed. They became much brighter.

I intended to upload the new screenshots later but for the time being, I’ll just try to reorganize the existing screenshots to make things clearer.

Whether my RGB or S-Video Blargg Filter preset was loaded doesn’t make a difference.

I hope that this one would be looked into @guest.r & @HyperspaceMadness and if it wasn’t intentional, reverted to the old GSL Scanline behaviour.

[JHorbach1]

I don’t know what my problem is, but the shader (with your new settings) is working right only if I set RetroArch resolution to 4:3 (which give black borders to MegaBezel), if it is set 16:9 the mask get’s “wrong”, even thought the “gaming area” is the same in both cases. What it seems to be happening is that the mask width is being altered when changing aspect ratio.

I’m using this Duimon background, I altered it to use the NTSC version. (Sorry, I don’t know how to format these parameters)

#reference "Mega_Bezel_Community_Collections/Duimon-Mega-Bezel/Presets/Advanced/PVM_PVM20/PVM20_03-[ADV]-[Guest] NTSC.slangp"
HSM_INTERLACE_TRIGGER_RES = "400.000000"
CS = "1.000000"
ntsc_sharp = "10.000000"
mask_bloom = "0.750000"
halation = "0.100000"
gamma_c = "1.200000"
brightboost1 = "1.000000"
gsl = "-1.000000"
shadowMask = "9.000000"
maskstr = "1.000000"
slotmask = "1.000000"
slotmask1 = "1.000000"
slotwidth = "3.000000"
double_slot = "2.000000"
deconrr = "0.750000"
deconrb = "-0.750000"
deconrry = "0.750000"
deconrby = "-0.750000"
decons = "1.500000"

[HyperspaceMadness]

@JHorbach1

but the shader (with your new settings) is working right only if I set RetroArch resolution to 4:3 (which give black borders to MegaBezel), if it is set 16:9 the mask get’s “wrong”, even thought the “gaming area” is the same in both cases. What it seems to be happening is that the mask width is being altered when changing aspect ratio.

Ok, thanks for this testing , this gives me a good clue as to why you might be seeing a difference in the ntsc shader. I’ll look into it, I’m sure I can find out what’s going on now.

@guest.r

From my observings the halation is applied quite strongly even with small parameter values, much stronger than in standalone ‘GDV’. Otherwise it’s appearing correctly.

Edit: seems to apply to bloom also.

I’ll check out what is going on with these too

[Cyber]

HSM Mega Bezel Reflection Shader Scanline Type Possible Bug - Scanline Type 2 is now brighter than Scanline Type -1 in pre V1.0 versions.

Version 0.9.107 2022-04-17 Rev 1 - GSL Scanlines Type -1 (Correct Output)

Version 1.0.001 2022-06-24 Rev 1 - GSL Scanlines Type -1 (Incorrect Output)

Version 0.9.107 2022-04-17 Rev 1 - GSL Scanlines Type 2 (Correct Output)

Version 1.0.001 2022-06-24 Rev 1 - GSL Scanlines Type 2 (Incorrect Output)

CyberLab__Slot_Mask_II_for_CyberLab_or_Core_Blargg_Composite_or_S-Video_Filter.slangp + Blargg_NTSC_Turbo_Duo_SNES_PSX_S-Video_CyberLab_Special_Edition.filt

4K_Optimized presets from

CyberLab Mega Bezel Death To Pixels Shader Preset Pack 13-07-22

were used in these tests.

[HyperspaceMadness]

One thing which might be different between the old version and newer versions with the new ntsc stuff was integrated is that I took out the scanline gamma offset parameter from the non-ntsc presets. This was so that the shader did not have an inconsistent behaviour from guests standard shader. This was something that I had previously hacked in.

The NTSC preset now has Guest’s standard scanline gamma parameter instead of my hacked offset. This means that standard guest presets should map better to the Mega Bezel.

@guest.r I can’t remember, is there a reason that the scanline gamma wouldn’t be good in the NON-NTSC version of the shader?

[Cyber]

@HyperspaceMadness

I took out the scanline gamma offset parameter from the non-ntsc presets

CyberLab__Slot_Mask_II_for_CyberLab_or_Core_Blargg_Composite_or_S-Video_Filter.slangp

#reference "CyberLab__Ultimate_Virtual_Slot_Mask_CRT-1P2RTA__ADV.slangp"

scangammaoffset = "0.400000"
double_slot = "1.000000"

GAMMA_INPUT = "2.100000"
post_br = "2.399999"
signalResolution = "1024.000000"
h_sharp = "15.000000"
s_sharp = "1.500000"

HSM_CROP_PERCENT_LEFT = "0.500000"
HSM_CROP_PERCENT_RIGHT = "0.300000"

HSM_SCALEFX_ON = "0.000000"
gamma_c = "1.200000"

SHARPEN = "1.000000"

CyberLab__Composite-Pure__Soqueroeu-TV_Special_Edition-NES__STD.slangp

#reference "../../../../Mega_Bezel_Community_Collections/Soqueroeu-TV-Backgrounds_V2.0/presets/TV-Console/Nintendo_NES.slangp"

scangammaoffset = "2.500000"

g_hue_degrees = "-7.000000"
wp_temperature = "6500.000000"



GDV_DECONVERGENCE_ON = "0.000000"



g_sat = "0.020000"



post_br = "1.310000"

h_sharp = "5.200000"

GAMMA_INPUT = "2.100000"
gamma_out = "2.500000"


bloom = "0.150000"
halation = "0.000000"



scan_falloff = "0.250000"


g_crtgamut = "1.000000"


scans = "0.000000"
GDV_NOISE_ON = "1.000000"

glow = "0.000000"

masksize = "2.000000"



gsl = "2.000000"
scanline1 = "9.000000"
scanline2 = "9.000000"
beam_min = "1.350000"
beam_max = "1.300000"

shadowMask = "6.000000"
maskstr = "1.000000"


addnoised = "0.180000"
noiseresd = "1.000000"


HSM_NON_INTEGER_SCALE = "61.570278"
HSM_CURVATURE_2D_SCALE_LONG_AXIS = "10.000000"
HSM_CURVATURE_2D_SCALE_SHORT_AXIS = "5.000000"
HSM_BZL_USE_INDEPENDENT_SCALE = "1.000000"
HSM_INT_SCALE_MODE = "1.000000"
HSM_INT_SCALE_MULTIPLE_OFFSET = "-2.000000"
HSM_BZL_INDEPENDENT_SCALE = "61.470329"
HSM_BZL_USE_INDEPENDENT_CURVATURE = "1.000000"


HSM_REFLECT_CORNER_ROTATION_OFFSET_TOP = "-4.000000"
HSM_REFLECT_CORNER_ROTATION_OFFSET_BOTTOM = "-4.000000"

Those are just 2 of my preset examples which use Scanline Gamma Offset. I use it for different things. It’s one of my main tools in reducing moire patterns.

[HyperspaceMadness]

@Cyber

scangammaoffset = "2.500000"

So this parameter is no longer doing anything in all the recent versions since Mega Bezel V 0.9.109 2022-06-24 Rev 1

[Cyber]

@HyperspaceMadness

So this parameter is no longer doing anything in all the recent versions since Mega Bezel V 0.9.109 2022-06-24 Rev 1

This explains a lot. I think I noticed that this setting might have been removed from at least one of the presets I submitted for inclusion in the Community CRT Variations folder.

Also very recently I loaded up some of my Soqueroeu-TV Special Edition presets and I noticed they looked a bit strange. I couldn’t put a finger on it though.

Since I was in mid development of some presets when v1 was released, I reverted to the version I was using until I recently tried the newest release.

It really threw me off as right now I’m just aiming for perfection. I’m already pretty happy about the way things look but I still challenge myself to see if I can do better.

If Scanline Gamma Offset can be of use in Guest-NTSC Shaders then perhaps it can see a comeback for creaters like myself who currently employ the Blargg-NTSC variety, which is still included in many cores.

It just handles very differently from Guest-NTSC and I like the fact that it gives my presets a unique touch and it offloads some of the work to the CPU thus freeing up some valuable GPU resources while taking advantage of often underutilized CPU resources.

Even though it’s “old”, it’s still a benchmark that at least some shader creators seem to aim to match/surpass with their own efforts.

It’s also included in RetroArch so why not have the ability to fully tap into such a useful and valuable resource?

[HyperspaceMadness]

@JHorbach1

I don’t know what my problem is, but the shader (with your new settings) is working right only if I set RetroArch resolution to 4:3 (which give black borders to MegaBezel), if it is set 16:9 the mask get’s “wrong”, even thought the “gaming area” is the same in both cases. What it seems to be happening is that the mask width is being altered when changing aspect ratio.

So I tried out your preset, using Duimon’s preset but referencing the NTSC base preset, and I can’t see any difference between Retroarch video scaling aspect ratio 4:3 vs 16:9, here are my tests

16x9

4:3

Closeup

16:9

4:3

[guest.r]

@HyperspaceMadness

@guest.r I can’t remember, is there a reason that the scanline gamma wouldn’t be good in the NON-NTSC version of the shader?

It could be added, yes. The main reason i implemented it in the NTSC version is because it looks better with main gamma value lowered to affect the horizontal filtering, while keeping the scanlines intact with scanline gamma.

[HyperspaceMadness]

@guest.r

It could be added, yes. The main reason i implemented it in the NTSC version is because it looks better with main gamma value lowered to affect the horizontal filtering, while keeping the scanlines intact with scanline gamma.

That would be awesome if you did choose to add it I think you would make @cyber’s day

[Cyber]

@HyperspaceMadness

That would be awesome if you did choose to add it I think you would make @cyber’s day

It most definitely did!! Thanks again guys!

This is what your work helps create!

CyberLab__Slot_Mask_III_for_CyberLab_or_Core_Blargg_S-Video_or_RGB_Filter.slangp + Blargg_NTSC_Turbo_Duo_SNES_PSX_RGB_CyberLab_Special_Edition.filt

Featuring Scanline Gamma Offset!

[JHorbach1]

Interesting… This is how it looks to me, 4K resolution. P.S.: Second image has the same mask pattern I get with standalone Guest.