Retro display solutions (monitors, TVs, CRTs, flatscreens, upscaling, calibration)

Discussion in 'Retro & Arcade' started by elvis, Jul 24, 2018.

  1. OP
    OP
    elvis

    elvis Old school old fool

    Joined:
    Jun 27, 2001
    Messages:
    37,112
    Location:
    Brisbane
    Maybe ask them if you can watch. Often old systems need large adjustments (like, strip magnets physically placed inside to adjust things), and then after that minor tweaks for the next few years.

    You might be able to get away with the one major service, and then do little ones yourself if you can get some tips from them.
     
  2. flain

    flain Member

    Joined:
    Oct 5, 2005
    Messages:
    2,341
    Location:
    Sydney
    I recently tested out a GBS-8220 with a neogeo board to an LCD monitor and the results were pretty bad. I started looking up some alternatives, custom firmware etc and in my journey I found there is a HDMI version called the SG-VC9900. The board looks very similar to the GBS8220 but with HDMI. I have no idea what kind of input lag this might have.

    I ordered one (yet to arrive) so i could test it out. However the more reading I did lead me down the path of just ponying up and buying an OSSC (which shipped yesterday yay). So i'll have an unused GBS-8220 and SG-VC9900 :rolleyes:

    One other issue i think may be occurring for me on the GBS-8220 is the VGA inputs on modern LCDs are not the best. It's hard to get info on them so there is no way to know what kind of scaling and image manipulation might be occurring when it converts the analogue VGA to digital within the LCD itself.

    The other thing to note is that some consoles and arcade boards have FPGA based HDMI modifications out there that can pull the digital video off before it hits any of the other video out components (neogeo, cps2/3 and gamecube to name a few), however they require soldering and availability is limited to however many the author of these things felt like releasing.
     
  3. OP
    OP
    elvis

    elvis Old school old fool

    Joined:
    Jun 27, 2001
    Messages:
    37,112
    Location:
    Brisbane
    Keen to hear more details. Interference? Bad scaling? Artifacts? Noisy picture? Something else?
     
  4. flain

    flain Member

    Joined:
    Oct 5, 2005
    Messages:
    2,341
    Location:
    Sydney
    I had image artifacts like certain lines on the screen near the top would get distorted/warp. So i played with the options in the menu and whenever i got it fixed something else would be wrong. Sometimes the image would wrap where the far left part of the screen would pop up on the right (this is probably just me playing with the positioning though). The colours were off too and washed out, i only now realize there are pots on the board to adjust with a screwdriver so that can probably be fixed.

    I initially thought maybe it was the neogeo so i plugged my sf2 ce board in and had the same issues (colours off again but in a different way + positioning issues). There was also some kind of smoothing applied that fuzzed the edges of things that i really didn't like, but i think this could actually be the scaler in the LCD screen. All of the output resolutions on the GBS-8220 are 4:3 and while the LCD itself has 4:3 as an option in the menu there's something up with the image processing resulting in my issues (as a best guess). When i get the OSSC i'm going to try feeding the screen HDMI and see what i can get from it. I've heard people pairing up OSSC with DVDOs or other low latency scalers so i'll have to see how it turns out.
     
    elvis likes this.
  5. Vanne

    Vanne Member

    Joined:
    Apr 20, 2003
    Messages:
    774
    I exchanged emails with Mike yesterday in regards to the Retrotink 2X, lol seems my paypal address is wrong, but that's ok, as long as it gets shipped to my correct address. Also having a question to the PAL model of the PS2, as far as I know the US model puts out 480i, which works fine for the NTSC version of the PS2, but I think the PAL version puts out 576i for PAL.

    Does that mean the Retrotink 2X wont work with a pal PS2?
     
  6. OP
    OP
    elvis

    elvis Old school old fool

    Joined:
    Jun 27, 2001
    Messages:
    37,112
    Location:
    Brisbane
    Will spit out a 576p50 signal, which most TVs are fine with.
     
    Vanne likes this.
  7. OP
    OP
    elvis

    elvis Old school old fool

    Joined:
    Jun 27, 2001
    Messages:
    37,112
    Location:
    Brisbane
    Playing around with a high end calibration tool this weekend, here's some comparison shots. Using the opening scene of Silent Hill 2 on the PCSX2 PS2 emulator for PC, running at triple-res out to a HP P1120 PC monitor (Sony Trinitron 21" PC monitor rebadged by HP).

    PC CRT monitor via VGA cable, 1024x768 res, running the emulator, uncalibrated (brightness 50%, contrast 50%, colour set to "6500" setting):

    pc_uncalibrated.jpg

    My cheap 240p/480i Chinese TV, YPbPr input, native original PS2 uncalibrated ("color" set to 50%, "contrast" set to 50%, "brightness" taken down to about 44%):

    tv_uncalibrated.jpg

    PC CRT monitor once again, this time after a colour calibration using a Spectracal C6 (rebranded i1 DisplayPro / ColorMunky Display):

    pc_calibrated.jpg

    So the bottom image is pretty close to how it's supposed to look (assuming whatever display you're looking at right now is also correct). But a pretty dramatic difference in how the image looks, and from there the artistic intent.

    Next step is to try a few calibration techniques for different displays around retro gaming (more similar to calibrating for old school analogue broadcast video than how a home theatre user would calibrate something), using the 240p test suite. Hopefully I can come up with something that allows people to calibrate their displays on the cheap, without needing to buy devices in the hundreds or even thousands of dollars.
     
    Last edited: Sep 10, 2018
    BuuBox, Vanne and WuZMoT like this.
  8. Vanne

    Vanne Member

    Joined:
    Apr 20, 2003
    Messages:
    774
    Holy crap!! bottom image looks SOOOOO much better!! nicely done! :D
     
  9. flain

    flain Member

    Joined:
    Oct 5, 2005
    Messages:
    2,341
    Location:
    Sydney
    I'm convinced the main issue i have with the GBS-8200 is the LCD screens inbuilt scaler. Since none of the resolutions are output at the displays native resolution it's scaling the image and applying the smoothing effect to all the lines. There is also noticeable feint noise/artefacts on black backgrounds through both boards.

    I also received the SG-VC9900 so tested it out along side the GBS-8200 briefly, this time with a CPS2 board plugged in.

    GBS-8200

    Click to view full size!


    One of the issues on the GBS, comes and goes i'm pretty sure playing with the settings could clear it up:

    Click to view full size!


    SG-VC9900, hdmi output but still the same output resolutions as the GBS-8200

    Click to view full size!


    SG-VC9900 - Interestingly it seems to output by default with a boarder (notice the brighter black around the image)

    Click to view full size!


    The two boards next to each other, very similar

    Click to view full size!


    My OSSC shipped last week and i'm waiting on that, i'm looking forward to trying its line 4x and 5x mode which is basically line doubling ie no smoothing. The other advantage is the OSSC in line 5x mode outputs at 1080p native so this should prevent my display from trying to scale the image and apply its smoothing.
     
    Last edited: Sep 11, 2018
  10. OP
    OP
    elvis

    elvis Old school old fool

    Joined:
    Jun 27, 2001
    Messages:
    37,112
    Location:
    Brisbane
    Might just be an integer scaling thing. Like, 240x4 or 480x2 is 960, a little short of a full 1080 lines, so a black border is added on to keep integer scaling and not have to deal with half-line problems.
     
  11. flain

    flain Member

    Joined:
    Oct 5, 2005
    Messages:
    2,341
    Location:
    Sydney
    It's probably something along those lines but keep in mind that these boards don't output 1080 they only output 4:3 resolutions. They both can output only 640x480, 800x600, 1024x768 or 1360x768. There are many dishonest ebay listings out there that claim "1080p" on these boards but don't specify that its an input not an output that it can do. Trying each of the output resolutions the best seems to be 640x480 on my panel. The big issue here is 4:3 output to a 16:9 LCD is there is no escaping the fact that the panel has to run the image though its own image scaling in order to display it.

    I'm pretty interested in the OSSC line 5x because while if you do the math it results in a resolution greater than 1080p, he made the FPGA crop some lines from the top and bottom to make it 1080p. Also of note is that this is in the latest OSSC revision - 1.6, so a lot of info out there is still outdated and unaware that line 5x exists as a thing that OSSC can do. I became interested in line 5x because i love the picture the Super NT is able to produce on the same panel using basically the same method. The gotcha here is that even though OSSC can output a 1080p image non-stretched and everything the sync is still at the mercy of the original system (no frame buffer in OSSC) so its a roll of the dice if your LCD will take the output or not. Fingers crossed here for me :).
     
  12. OP
    OP
    elvis

    elvis Old school old fool

    Joined:
    Jun 27, 2001
    Messages:
    37,112
    Location:
    Brisbane
    That's generally not an issue (speed wise) for 480p and up. Most panels scale that to HD/4K resolutions without too much drama.

    The issue is 15KHz resolutions - 480i and especially 240p (the latter is not a broadcast standard, so many video engineers refuse to acknowledge it, and force it to be incorrectly processed as 480i).
     
    Last edited: Sep 11, 2018
  13. flain

    flain Member

    Joined:
    Oct 5, 2005
    Messages:
    2,341
    Location:
    Sydney
    Speed wise i do agree its a non issue but the look of the image is impacted a lot, after all the scalers in LCDs are geared towards making videos and modern games look nice, but if you run that same scaling algorithm over an old game it looks bad :(.

    One thing i have noticed with these boards is on faster animations - especially the marvel cps2 board i tested, you do see interlacing. This is documented though so not unexpected, but for fans of super fast motion fighters interlacing does show to the naked eye (when the background scrolls fast you see it even more). You probably wont see the interlacing on slower games though.

    Another thing i realized is that me taking pics with my phone then uploading them at a lower resolution also isn't helping (doh!) because when the image was downscaled, thats another scaler! The actual screen looked a lot worse than the pictures i posted :/
     
    elvis likes this.
  14. OP
    OP
    elvis

    elvis Old school old fool

    Joined:
    Jun 27, 2001
    Messages:
    37,112
    Location:
    Brisbane
    Colour calibration

    I've been pondering for a while how to make this post. Colour science is a HUGE topic all by itself, and there's a lot of detail you can easily get caught up in, even for seemingly simple tasks. I'll try and keep it fairly straight forward in this post, but if folks have questions, please ask. As most know, my background is I work for a VFX and Post Production company, so colour science as well as understanding broadcast standards is a big part of my job, so I'll aim to be as accurate as I can and cite documentation where possible.

    Dictionary
    * Luminance - intensity of light, or volume of light photons, from darkest ("black" to, lightest "white")
    * Rods - photoreceptor cells in your eyes that respond to luminance
    * Cones - photoreceptor cells in your eyes that respond to particular electromagnetic wavelengths we perceive as colour
    * Gamut - the range of colour something can support. "Narrow" gamuts support a smaller range of colour, "Wide" gamuts support more.
    * Colorimeter - a tool designed to measure a specific set of light wavelengths, often limited to a specific gamut
    * Spectrometer - a tool designed to measure a much broader set of wavelengths, often much more expensive than a standard colorimeter

    Basics of colour

    One certainty of colour is that it's always more complex than you think. Some very clever people wrote some very serious documentation on the subject, and if you're interested in much, much more detail than I'm going to cover here, read this for a start:

    http://github.com/jeremyselan/cinematiccolor/raw/master/ves/Cinematic_Color_VES.pdf
    (From here: http://cinematiccolor.org/ )

    And of course, I always recommend Wikipedia for general knowledge on any technical topic:
    https://en.wikipedia.org/wiki/Color_theory

    But to keep it simple, let's state a few facts:

    1) When it comes to measuring the difference in luminance (bright/dark) values, the BEST tool is the human eye. Our eyes are more sensitive and better able to measure differences in luminance values than almost any colorimeter or spectrometer on the market.

    2) When it comes to measuring the difference in colour, our eyes are one of the worst tools. Because our eyes (more realistically, our brains responding to signals sent from our eyes) can so quickly adjust what they consider "normal colour". If you've ever spent a few minutes staring through coloured glasses/lenses/cellophane or some other material, you will know the sensation of taking away that material and perceiving the natural colour around you as being off (typically the opposite colour of what you were looking through - look through a blue filter for a while, remove it, and the world seems more orange/red, for example). Likewise how we perceive colour and different luminance greatly varies.

    3) There is no such thing as "white light". Nor pink or brown or a bunch of other colours for that matter. The colour spectrum is analogue, and therefor infinite in range. However there are set colours that add together to make others. Daylight on planet Earth also changes, and we see that light reflected. A "white" object looks white to us whether the sun is high in the sky (typically emitting bluer, or "cooler" light because of our atmosphere), or low in the sky (a redder or "warmer" light). Our eyes adjust quickly to what we call the "colour temperature" that's being reflected from an object to normalise it for us.

    4) Your eyes don't see all colours directly. The cones in your eyes respond to colours at analogue levels to certain colours, and overlap slightly. When a particular wavelength triggers a signal to your brain, it triggers that in sliding levels. Our cones typically fire on three different "primary" signals - red, green and blue. Hence it makes sense for colour science to represent colours in those same three primaries, as it matches how our eyes work. It's entirely possible to use totally different primary colours, however it's quite pointless in practice, as we generally care about the human eye as the receptor we care about.

    Human eyes have evolved over millions of years, and like most evolution has done so for our safety. Being able to adapt quickly to light/dark situations is helpful when changing from areas of direct sunlight to areas of shade. Perceiving differences in light help is detect shadow and movement, which is necessary for catching prey as well as avoiding being prey. Similarly our eyes are far more sensitive to detail in reds and blues (frequently danger/food colours in nature) than they are greens (foliage, plants in nature). So measuring colour can be tricky.

    So why bother?

    Good question. Given our eyes are pretty crappy at objectively measuring colour, why bother calibrating things at all?

    The technology we use to represent pictures (TVs, monitors, projectors, etc) generally has a fixed gamut (range of colour). Often these colours are limited, so if we "tune" or calibrate them the wrong way, colours end up "clipping" - i.e.: the signal being sent to the device isn't displayed correctly, and fine ranges of colours are lost. This then destroys the artistic value of what we see, and fine details go missing. (Similar to audio clipping, where the high range of a frequency or amplitude is cut flat in poorly tuned audio equipment, and detail is lost, which we sometimes hear as audio crunches or crackles).

    At a more extreme end, very very poor calibration can make some things look utterly terrible. The classic situation for retro gamers is the ancient CRT television showing Super Mario Brothers with the "purple sky". Nobody wants that.

    If your colours are too warm, you can blow out the reds, make skin tones look awful, bring greens closer to yellow and blues closer to magenta. Too cool, and the opposite happens to the other extreme. If brightness and contrast are out, you lose subtle details in things like dark blues on blacks, or bright yellows on whites. Just as good quality signals and sharp pictures are important, so too are colours and levels.

    Enough sciencey crap, how do I make my display look better?

    Calibration is a relative relationship between three things
    * The thing defining the colours and brightness to display (console, computer, Blu-Ray player, Netlix app, etc)
    * The source of light (your display/monitor/TV)
    * The thing receiving the light (your eyes, colorimeter, spectrometer)

    With all of the above in the back of your mind, there are three major things you want to get right, in order of both importance and complexity, when colour calibrating your displays. This goes equally for any display technology on the market - CRT, LCD or OLED.

    1) Your black levels and white levels. As above, human eyes are most sensitive to differences in luminance. Getting this wrong on your display can cause "crushing", or a loss of detail when the dark and light values go beyond what your display can handle, and all end up looking like a single colour or shade. It's worth noting that "HDR" in home theatre speak often refers to more detail in these high ranges, unlike the simulated "HDR" that's been around in video games for a while now, which is about some clever colour science that makes a scene look more like real life despite being on a crappy LCD monitor.

    Somewhat related to this is gamma correction, or the fact that humans prefer brightness to be non-linear. We prefer more detail in the mid ranges (i.e.: not the extreme light, or extreme dark) on limited gamut displays. A "higher gamma" setting means the "gamma curve", or amount that the brightness changes in the middle range is greater. Higher gamma is often preferred for when you view your display in light conditions (daytime viewing in a bright room), and a lower gamma curve is preferred for darker conditions (night time viewing, dark room, cinema, etc).

    2) White point, or white balance. Again remember our sciencey crap above - there is no white light. We have primaries (for humans, red, green and blue). We have secondaries (two of red, green or blue mix, and we perceive shades of cyan, magenta or yellow), and we have tertiary light, or where we mix all three and perceive many colours, including white. For lots and lots of reasons, this mix is more frequently imperfect - the electron guns in your CRT have drifted out of spec differently over time, the cells in your LCD panel require different voltages to create the same intensity of light and are out by a given quantity, or your OLED has the primaries decay at different rates. But also, because there is no white light, we can't define white light. There's no such perfect thing. So we measure instead the "white point" as a temperature value. Some common standards are 5000 Kelvin, or 5000K, or D50, D65 or 6500K, and on older PC monitors even right up to 9300K.

    The cinema, broadcast, VFX and TV industries have all more or less agreed on D65 as the agreed standard for most setups, as that is what is considered to be "midday daylight" white, as perceived on a clear day on Earth (as opposed to Mars, where day is red, and sunset is blue!). For that reason, I use D65 as my preferred standard too, as it makes life a lot easier when trying to calibrate a display.

    Relative to that, D50 will look "warm" (redder), and 9300K will look "cool" (bluer). Remember however that this warm/cool sensation will pass as your eyes adjust quickly, which is all a part of the challenge.

    3) Colour - this is the real tricky bit. Finding the full range or gamut of colour that your display supports, and making sure that at any given pixel colour setting your display is closely matching what it should technically be spitting out is not only difficult and time consuming, it can sometimes be a slow descent into madness. Even with your white point set correctly, you can different colour/gamma curves on your three primary channels, meaning that at different levels the red, green and blue colours drift apart in intensity, before coming back together again at the black and white ends of the brightness for that colour. So often for displays that are old, like our precious CRTs, getting the colour right at a given luminance/brightness point can make it look and measure wrong at another.

    Great, let's calibrate!

    For everything I talk about here, I'm going to be using the Rec.709/Rec.1886 colour standards, with a white point of D65 and a gamma curve of 2.4. These standards are largely agreed upon for home theatre setups, which are closer to what gamers want than cinema setups. More details here:
    https://partnerhelp.netflixstudios..../215669787-Display-Calibration-Best-Practices

    On to the hands-on bit. There are a near infinite number of ways to calibrate, all that come with their own level of cost and effort. If you've got the cash, shelling out for a colorimeter (remembering that a CRT is like an OLED, is wide gamut, and needs a colorimeter that can handle that), or a more expensive spectrometer is the way to go. But these aren't cheap, and the more you spend, the better the product. Cheap and nasty narrow-gamut colorimeters (only good for LCD displays) like the ColorMunki Smile can be had for as little as $130. Mid range devices like the Spyder5 Pro (now wide gamut thanks to the popularity of OLED) closer to $300. And then you've to the iDisplay and clones (like the Spectracal C6) that will start you in the $700 or more range. Upgrade to spectrometers, and you're talking multiple thousands.

    But what if you don't want to spend the cash? Let's look at some cheap options. Remember that these aren't going to be anywhere near as accurate as using proper equipment, but they're probably going to give you some sort of improvement.

    Good options for test images that can help us are:

    Consoles with flash carts, and a copy of the 240p test suite:
    * http://junkerhq.net/xrgb/index.php?title=240p_test_suite

    Blu-Ray or DVD players (including some newer consoles that can play these discs):
    * https://calman.spectracal.com/downloads-pattern-discs.html
    * https://www.avsforum.com/forum/139-...48496-avs-hd-709-blu-ray-mp4-calibration.html

    Computers (including the Raspberry Pi, which can use VLC for media, or just show static images):
    * https://www.avsforum.com/forum/139-...48496-avs-hd-709-blu-ray-mp4-calibration.html

    1) Black levels, white levels.

    On the 240p test suite, fire up the pluge pattern. Here's an example of one:

    Pluge.png

    On the left and right are three bars of varying intensity (note: if you're on a properly calibrated display or even a good quality phone OLED, you may not see the third bar). You want these set so that the darkest bar vanishes in to the black, but the second bar can be *JUST* seen and distinguished. For reasons I won't go into here, there are levels lower than "black", and the third bar is the lowest "reference black" (black-16, or an 8bpp RGB image 16,16,16).

    There are similar patterns for whites available, and here's an example video of calibrating for both black and white, again remembering that reference-white (white-235, or an 8bpp image 235,235,235) and reference-black (black-16) are the points you care about, so you should see detail on one side of that range, but not the other. Depending on your display, you may have different options available - either just brightness, maybe brightness and contrast, or if you're lucky proper black level adjustments.





    2) White point. Now we get tricky. As above, your eyes suck at this, so we need tools to help. Thankfully in the modern era we have a few bits of luck on our side:

    * Smartphones today often have luminance and colour measurement hardware built in
    * Frequently these are factory-calibrated for D65, because we use our phones outdoors to take photos

    Now, it's important to note that ALL HARDWARE DRIFTS. Your monitor will drift (calibrated it once, use it for a few years, and the colours will change over time). Your colorimeter and spectrometer will drift (even when paying hundreds/thousands for these tools, they have to be re-calibrated by the manufacturer, or thrown away and replaced after some time), and your phone's internal electronics will drift too. The newer the phone, the better your chances of things being good.

    I use an Android phone. If someone can try to do this on an iPhone and report in, I'd be greatful.

    There are a number of apps that offer light measurement, as well as colour pickers. Two I use because they have the particular information I want are:

    Light Meter (measures light in Lux - identical to cd/m^2, and also colour temperature in K):
    https://play.google.com/store/apps/details?id=com.bti.lightMeter

    Test out Light Meter before if you can. Try to find out where your phone's in-built light meter is - it's usually on the front of your phone, and not normally your camera, but rather little sensors near the camera. Simple to test by just putting your finger over it and watching the values change. When you've found it, also test to see what values it returns. Overcast days should be in the 5000K range, and full sun on a clear day at midday should be near the set standard of 6500K.

    Color Grab:
    https://play.google.com/store/apps/details?id=com.loomatix.colorgrab

    Color Grab in particular seems to do a better job of not auto-white-balancing constantly, and sticking with a fairly sensible factory setting, even when in "auto" mode. Test it out on things when you download the app.

    What I do is the following:

    * Do this in a dark room, with no other light sources. Artificial lights have their own colour temperature, and can throw off your devices.
    * If you're on a CRT, no problems. If you're on a modern TV, turn off any automatic backlight/brightness adjustment your TV might have if you can.

    * Using both apps mentioned above:

    ** Light Meter - Bring up a 100% white image. Try to point the light sensor at the screen (often on the front of your phone and it's very directional, so it's a bit tricky), and read the value. See if it's in the 6500 range. If not, play with your TV's preset colour temperature settings - often there'll two levels of warm, two levels of cool, and one neutral. These are generally meaningless, and historically Japanese TVs were all set too cool anyway. You might find "warm" gives you a value closer to 6500 than others (or maybe cool, if your TV is really old and busted).

    Light Meter will also give you "Lux" output (identical to cd/m^2). That's the amount of light coming out of the display. Netflix recommend 150-250 Lux for daylight viewing. I subjectively like 170-200 myself. But don't try to adjust too much here via brightness controls, as it'll affect your black and white levels. Only change this if you have a dedicated setting for it, like a backlight on an LCD.

    ** Colour Grab - Bring up a 50% grey image. Inside Colour Grab, set the "Camera Left Cell" to RGB, and the "Camera Right Cell" to YCbCr SDTV. Point the camera at the screen. You ideally want to see RGB values in the same range (0 being lowest, 255 being highest, so grey is ideally 127, 127, 127). If these numbers are up or down together, that's OK (your camera's automatic aperture can change the perceived intensity). If they're radically different (say, 130, 120, 120) then try different settings. The three numbers are red, green and blue respectively, so a high blue value means you're too cool, and need to go warmer. A high red value means you're too warm, and need to go cooler.

    The YCbCr values are three numbers, with the first being luminance, which you can ignore. The second two are colour values, and should be the same. You'll see these drift apart like your RGB values if things are off.

    Repeat this test for other grey values - I try to at least test 25%, 50%, 75% and 100% (i.e.: white). But it depends on your camera and whether or not it can distinguish high/low values.

    If you're on a TV or monitor that allows you to adjust red, green and blue independently, do this instead of overall colour warmth, and get it to a point where you're happy with the measured result at different grey levels. If the measurements drift too much between levels, prioritise 50% and 75% grey as your most important values to calibrate against, and ignore the rest.

    Again, if you've got a colorimeter or spectrometer, you won't bother with any of this. Those tools are orders of magnitude more accurate.

    3) Colour. With the light/dark points set (i.e.: brightness/contrast) white point set, colour should be pretty much in the ballpark. If you're using a Sony PVM, BVM or another broadcast monitor from companies like Sony, Panasonic, JVC, Eizo, etc, then you can use the SMPTE colour bars provided with the 240p test suite (these sometimes don't show on PAL consoles mind you, and will be missing detail on limited colour palette consoles like the Sega Genesis/Megadrive), and your monitor's ability to show only the Blue channel, to get the hue/saturation levels (another way of measuring colour on the HSV scale, different to the RGB scale) in check:



    Similarly, there are DVDs and BluRays that come with varying colour patterns and a plastic filter you can look through that shows only specific parts of the colour spectrum, and lets you calibrate in the same way as the pro monitor method by using the maths of colour intensities to for given primary and secondary colours. I'm in the process of ordering some different "gels" (term for the plastic colour filters) from lighting and photography places, and I'll try to report back later on the success rates I have with those and the overall price. If I can find a three-primary gel pack in the order of $10-$20 delivered that does the job, that's a win, and I'll document the process and different test patterns you can use with hue/saturation settings.

    But again, consider the first two calibration items the most important. And if you've got better tools, use them instead of any of this, as they'll be far more accurate.

    Happy calibrating!
     
    Last edited: Sep 13, 2018
    oculi likes this.
  15. OP
    OP
    elvis

    elvis Old school old fool

    Joined:
    Jun 27, 2001
    Messages:
    37,112
    Location:
    Brisbane
    Using my "Light Meter" app method above, here's an example.

    Chinese, no-name SD CRT TV. No real colour options to speak of (certainly no R/G/B or H/S/V tweaks in the main menus).

    I've connected a Raspberry Pi spitting out NTSC Composite video at 480i using a 640x480 frame buffer. I've made an image using ImageMagick to be an exact R,G,B-127,127,127 (50% grey) full screen, unscaled, and displayed it on screen.

    Cycling through my temperature options, here's what it looks like to my phone camera:



    Especially noticable when flicking between the "WARM2" and "COOL2" extreme values. "NORM" appears to be the setting in the middle I'd want, right? Let's ask Light Meter.

    I have a screenshot timer on, set the TV to a setting, put the phone's light sensor flat against the screen and wait for the screenshot to be taken.

    Left to right, with a 100% white image, the results are:

    WARM2 , WARM1 , NORM , COOL 1, COOL 2

    light_meter_white.jpg

    And with 50% grey:

    light_meter_grey.jpg

    So two things to come away with from this test. Assuming the light meter on my phone (Samsung Galaxy S6) is correct [edit - and quick tests says it is NOT accurate, but I'll go into detail in a later post]:

    1) "WARM1" is the desired white point, closest to the D65 / 6500K value of common colour standards like sRGB, Rec.601, Rec.709, Rec.2020, etc.

    2) At full white, I'm spitting out around 170-180 Lux, which is nicely in the 150-250 range that is optimised for daylight viewing.

    So that's a great starting point to test. I'll get a hold of a proper wide-gamut colorimeter and test that again to compare and see what it says.
     
    Last edited: Sep 13, 2018
  16. badmofo

    badmofo Member

    Joined:
    Oct 10, 2011
    Messages:
    1,003
    Location:
    Australia
    I have a RetroTink 2X on the way and am starting to think about what I'm going to use as a display. I'll be trying it with a VGA CRT via a HDMI -> VGA converter but it seems pretty clunky and bulky given I could just use something with HDMI in, but I don't want a wide screen sitting with my C64 for aesthetics reasons.

    Is there a 4:3 option with HDMI in and a decent picture quality? Could be a TV or monitor I suppose or am I putting my ignorance on display here for all to see? No pun intended.
     
  17. Vanne

    Vanne Member

    Joined:
    Apr 20, 2003
    Messages:
    774
    I am probably in the same boat as you Badmofo, NFI really here, just trying to learn a bit as i go along. I just spent 20 US on buying component cables for my PS2, i figure thats the cleanest signal the RetroTink is going to be able to get. Ive been running the PS2 the last couple of days via a PS2-2-Hdmi thingy, but its pretty rubbish. the picture on screen looks washed out, and the colours are rather drab. Hoping for good things from the Tink. :)

    Not sure when mine is shipping though. which batch were you in? the Pre OCT or the post?

    I am also hoping for good things for my Wii/N64 collection with the Tink. (which is also component in)

    If it goes pretty good, ill probably purchase an extra one for my DC.

    Either that or ill be on the hunt for a nice Loewie big ass crtTV.
     
  18. OP
    OP
    elvis

    elvis Old school old fool

    Joined:
    Jun 27, 2001
    Messages:
    37,112
    Location:
    Brisbane
    Not that I've seen.

    There's a number of good quality 16:9 PC monitors on the market that will do 4:3 properly (select in-menu). For example, my Dell E2311H monitor at work allows this on it's digital inputs.

    But if you don't want a 16:9 monitor (even if it does offer the correct ratios), I'm not sure what else is on offer. This is where I'd be inclined to make a case that replicated an old CRT, and put the LCD monitor inside - see this post from earlier in this thread.

    But I prefer CRT, and for an Amiga I'd be looking at a SCART solution. That could mean a real SCART TV, an RGB-modded TV, or a Shinybow SB-2840 RGB to YPbPr (more accurately, analogue YCbCr) (aka "component") input TV.
     
  19. badmofo

    badmofo Member

    Joined:
    Oct 10, 2011
    Messages:
    1,003
    Location:
    Australia
    OK thanks for the response elvis, I'll give up on retro aesthetic HDMI screen then I guess. I have a few nice VGA CRTs here so will see how that goes - I do like a CRT too.

    Vanne I'm post October I think - I have S video cables here for my C64 and PS1, I need to look into component I guess and see what I can do about getting something better that composite out of my SMD. Should be a fun project!
     
  20. WuZMoT

    WuZMoT Member

    Joined:
    Feb 2, 2002
    Messages:
    1,832
    Location:
    Brisbane
    the pre october units have also been delayed due to a lost fedex package containing parts.
     

Share This Page

Advertisement: