Thanks. I’m using a GeForce GTX 1070. Shadow Play doesn’t record in HDR on that GPU period. It can do HDR stills in jxr format though. My SDR recordings don’t seem to have issues with Chroma Subsampling though so I’m assuming that it does record SDR footage in 4:4:4.
I’ll check them out. I was just wondering if I could just record full range 4:4:4 using Shadow Play then inverse tonemap that recording in editing or post and have a similar result as Sony Megatron HDR. So I wanted to know if there are any tips you could share regarding which colour format to choose or how to go about tonemapping because I’ve never done anything like that before but I can probably wait and see where things end up however I’m pretty cool with the way things are looking now after my tweaking.
For me, YouTube compatibility is secondary so whether or not it accepts 4:4:4 HDR - which I wouldn’t be surprised if it doesn’t, I’ll still do my videos and share them via regular download links if I need to - which is what I’m used to doing.
I guess I could use the information contained in this post you shared to help guide me. I think I should be able to get this to work because it seems to make sense that it should be doable in theory.
RetroArch’s internal video dumping does RGB 4:4:4 in lossless mode. However, AFAIK it’s not hardware-accelerated, so it encodes (much) slower than real-time. Depending on the core, you may be able to use RetroArch’s input recording/playback to do it unattended. I don’t know whether these dumps would be detected as HDR content on playback, either.
Thanks, I have a question, is my understanding of the attack parameter correct in the pic? Higher number means slower?
And that it requires lower content res than the effective “Vertical TVL” (I know that’s not a thing) of the mask? “4k/600TVL/Shadow” settings gives a 360 (4 pixel tall mask÷1440p) “Vertical TVL” on my 1440p. Meaning that 360p and higher content (without overscale) won’t work well with the attack setting on my monitor, am I correct also in this?
If you put 100 into one of the text box that’s equivalent to putting 1.0 for attack in the shader. The image in the background is a gaussian distribution on the scanline (take the image with a grain of salt though as its no doubt been warped by processing and photography).
Just updated the default and 2730QM presets on Github to be colour accurate with my PVM (at least by eyeballing it). These are the closest I’ve got to matching my 2730QM PVM and pretty spot on. I need to use a spectrophotometers to get closer.
You will still need to adjust your peak and paper white brightness.
I just had time to test dylanraga’s win11hdr-srgb-to-gamma2.2-icm profile and I am very happy with the result. I did some measurements with HCFR and my i1 Display Pro Colorimeter and I can get very good results with Windows 11 set to HDR and the SDR slider at maximum. The icm profile raises the lower steps of the gamma curve (between 0-20 IRE) a bit to prevent the washout, which is present without the profile.
Until these measurements I thought it will be difficult without a LUT to straighten the Gamma curve, but after I activated the dynamic tonemapping function of the LG GX OLED TV, I got a very surprising result with a Gamma curve, which measures between 2.2 and 2.3:
As you can see, the results are just great. For me personally, there is no need to use any of the HDR tonemapping solutions, which we discussed before. Just Windows 11 set to HDR, the SDR slider to maximum, the sRGB to gamma 2.2 profile installed and dynamic tonemapping activated on my OLED is all what’s needed to get a bright enough and pretty accurate picture with the Megatron shader, which is set to SDR. And this solution is thankfully compatible with any emulator using the reshade port of Megatron.
Thank you again @ Azurfel for the hint with the sRGB to Gamma 2.2 profile
The only thing I am curious about is how a UHD LCD IPS panel with RGB subpixels or VA Panel with BGR subpixels would perform with the Megatron shader, as the “400” lines setting (resolution set to 8K and 800 TVL) within Megatron, which is my favourite, has some minor issues with overlapping phosphors on my LG OLED as you can see here:
But I think the difference will be so minor, that it will not be visible to the eye from the distance I play games. I think those new Mini LED TV’s with good black levels are a good choice too for Retrogaming and CRT emulation and these bright TV’s should even work in normal SDR mode without any of the workarounds the OLED TV’s need.
I have also the idea to buy a Sony Trinitron CRT in the future, just for the purpuse of comparing to the Megatron shader and perfecting the settings. The Megatron shader just rocks !!!
I have to edit my post, because I have mistaken dynamic tonemapping with HGIG: This measurement is with the LG OLED set to HGIG and not Dynamic under the Dynamic Tone Mapping section of the TV:
Tonemapping Off or Dynamic has those weird gamma curves I posted above which goes toward 1.2 at the higher brightness levels. So HGIG is the way to go for a straight gamma on LG OLED TV’s or at least this specific GX model.
I therefore repeat the statement I made above in a corrected way:
Just Windows 11 set to HDR, the SDR slider to maximum, the sRGB to gamma 2.2 profile installed (I use the 400 nits icm file currently) and HGIG under dynamic tonemapping settings on my OLED is all what’s needed to get a bright enough and pretty accurate picture with the Megatron shader, which is set to SDR.
Next step is to generate an .icm profile by myself, which fits with the SDR slider @ 100 or even overdrive the slider via registry to higher luminance values as Azurfel mentioned, to squeeze out even more brightness of my LG OLED and get the gamma more linear towards 2.2
That’s what the OLED RWBG Subpixel Layout is supposed to resolve. Also, Colour Accurate Mode also introduces the appearance of overlapping subpixels.
I do agree that the Slot Mask Presets might need some work and further testing even with the OLED RWB Subpixel Layout though.
There are also issues with the scanline patterns when using the Slot Mask with Deconvergence enabled.
I’ve been focussing on Aperture Grille for now because that works perfectly. Maybe later on I’ll try to assist with the Slot Mask and Shadow Mask Modes to get them in line with what I’ve used in my presets in the past with other shaders.
I don’t think my older LG OLED TV has any of those tonemapping settings.
This is great work that you’re doing though. Just using Windows and the display to get the extra brightness and perform the correct tonemapping instead of the shader. It makes sense that it would work. Now how do we apply this to a video editor if we were to record this SDR Megatron to get a similar effect but with the display’s HDR mode being triggered and used when the video is played?
I tried all, the WRGB subpixel structure does the same overlapping and the BGR overlaps all and looks even worse. With resolution set to 8K and 800 TVL it doesn’t make any difference if mask or color accurate.
On my LG GX the HGIG setting is under the picture settings and then under advanced controls. What I measure here is, that the HGIG setting actually maps the EOTF in HDR mode back to a SDR gamma curve. Maybe dig again in the menu, your OLED may have it too.
Compared to normal SDR mode in Windows 11, the HDR mode with the SDR slider @ value 100 feels like around 25% brighter. Actually, it would still be nice to have more brightness on my GX, but this workaround at least makes it possible to not have the feeling of a too dim picture. Mini LED TV’s and for example the newer OLED’s like the G3 or Samsung’s QLED may not even need the HDR /SDR workaround and work just fine in SDR. But I will try too “overdrive” the SDR slider in Win11, maybe it will help further. The dynamic tonemapping setting of the LG is even brighter than HGIG, but not accurate anymore as you can see in the gamma measurements above. But it’s usable too, if someone does not focus on absolute accuracy.
Good question, at the moment honestly I have no idea as I currently just use my phone to capture. But maybe one idea is, that people who have an HDR capable display just use the same method I am using, just putting the TV manually through the OS in HDR mode and then setting the SDR slider to the maximum of value 100. This will allow also to see the videos / pictures with more brightness.
I acknowledge this is a very dumb and noobish question on my part! I don’t have a deep understanding of the function of all the parameters in the shader; I’ve just been using it because it looks great on my 4K TV! Megatron in my opinion blows every other CRT shader out of the water.
I’ve been trying to use it now on my tablet though which is near 4K, and the results I get seem to differ dramatically depending on which retroarch core I’m using. For example, this is a Snes9x game (PVM 1910 SDR, only settings change I made was to 600 TVL), which looks good to me:
But when I was trying some PSX games or ffmpeg, the same settings give me what I’d describe as intense color banding (this is ffmpeg, identical settings)
Yeah itll probably be the high resolution of that content. First thing to try is integer scaling. PSX I think is 240p so that really shpuld work without banding/problems so maybe something else is going on there.
What resolution is your tablet? What model is your tablet?
Thanks! You should really update, as Windows 11 has many improvements in handling HDR signals.
It would be really awesome, if you could fix the overlapping phosphor issue. In the close up picture I posted, you can see, that the blue overlaps the green by one single row:
So the blue has to be moved 1 pixel row to the right (this is with the Display resolution in Megatron set to 8K and resolution set to 800 TVL).
The 4K / 300 TVL setting is just too coarse in my eyes, as the overall picture looks too “rough”. Maybe, for the sake of convenience, you could also just include a 400 TVL setting, as many may not know they can set the Display resolution to 8K in the shader with a 4K display.
I would be also interested about what you think how the Display resolution set to 8K and 800 TVL looks to your eyes on your Display’s. For me this combination is absolutely on point in regards to how the phosphors and scanlines look. Not too coarse and also not too fine like a Sony BVM / PVM, just spot on for me.
It would be awesome if @MajorPainTheCactus could do that and have RGB “magically” work on a non RGB layout screen but I don’t think that’s what’s going on there. It’s a case of incorrect layout being used for the layout of the panel which is something that has been tried, tested and proven already using empirical data.
If you switch your layout to RWBG there shouldn’t be any overlapping of a single blue subpixel unless there’s a bug that’s being manifested there.
You said there are no issues with the Aperture Grille at 300TVL using RGB Subpixel Layout but there are glaring issues to me. The spacing is incorrect and uneven. The Red, Green and Blue “phosphors” should be evenly spaced but they’re not. There are 2 subpixel columns of red, then 1 column of black, then 2 columns of green, then 2 columns of blue, then 1 column of black.
If you had taken pictures of the RWBG Subpixel Layout you would have seen 2 columns of blue, 2 columns of green, 2 columns of red, then 1 column of black.
You can see the correct layout of subpixels here:
It’s the red subpixel followed by the white subpixel, followed by the blue subpixel, followed by the green subpixel. The white subpixel is always off so that gives us our black gap after the red subpixel/“phosphor”.
This is why the even grouping starts at the blue, then green, then red then black gap in the Aperture Grille mask.
If you look at the photos in the links below you will see examples of both LED and OLED displays producing proper, evenly spaced aperture grille and slot mask “phosphors”.
What you are seeing there using that RGB Layout, even with the Virtua Fighter preset is far from what has already been achieved using LG OLED TV and many LED displays - which is the standard.
This Slot Mask doesn’t have the same issues of overlapping subpixels and the same uneven spacing as the RGB Virtua Fighter photo you shared. It uses the RWBG layout. The major issue with it is that when Deconvergence is used, the scanlines get messed up.
It might look even better as in more evenly aligned if it followed the Aperture Grille RWGB subpixel order of B, G, R, X. Or at least that’s how the Aperture Grille appears to be ordered but it could just be looking that way due to where the black gap (caused by the white subpixel being off) falls. So even if technically it starts at R and is RRXXBBGG to our eyes it’s going to look like groups of BGR because our eyes are looking for the evenly spaced and grouped triads regardless of what’s going on under the hood.
This is the empirical evidence that I’m talking about:
These photos of an acceptable looking aperture grille using RGB Mask Layout might be a little deceiving, not intentionally though. There is still the evidence of the large gap between the red and the green phosphor as well as the extreme closeness and almost overlapping of the green and the blue phosphors.
Again I say though, if @MajorPainTheCactus can pull off that perfectly aligned RGB Slot Mask on LG OLED TVs by shifting the blue subpixel by one subpixel I would be a very happy camper.
Unfortunately he has to work with what the display provides him and I don’t see the subpixels wanting to move to accommodate our desire so we have to work within and around their structure in order to get as close as possible to our desired result.
Feel free to take a look at this post. It shows how you can edit the Subpixel Layouts so you could experiment on your own if you wished.
I understand some people are adamant about perfect spacing, but in my experience having the colors in the wrong order is way more egregious than the spacing being slightly off. Certainly at a playable distance, but also on close inspection. Though I haven’t done the comparison at every single mask size. (Using the 27GR95QE)
So yeah it would be nice if there were more options for shifting pixels and spacing between phosphors without having to resort to blatantly wrong color order. Not just for Megatron but in general.
I don’t know if anyone else has done this yet or is bold enough to try it but if you take one of those LG WRGB/W-OLED TVs and rotate it 180° and then used the RWGB (OLED) Subpixel Layout, you should be able to get the “perfect” RGB Subpixel display.
