Awesome! That’s great - if you have any more closeup shots like this specifically showing darker colours, pure white (if this afterburner text isn’t pure white already), pure red, pure green, pure blue that would be fantastic. This is a great start though. Do you know what iso, aperture speed etc were used or is this iPhone auto mode?
I have a Samsung A7 2018 phone, not sure about the iso settings or anything its just the auto preset.
Thats a picture i took about a year ago or so, since then i moved the cabinet into my brother’s tattoo studio and also, recently, theres a problem with the screen flickering a bit. But if i get the chance ill take some pics.
1 Like
I have a Sony BVM-20F1U, simply calibrated with a i1Display 3. Is there need for pictures from this particular display? I have a Canon 6D with a 100mm F2.8 Macro lens and a tripod. I have taken pictures before and adjusting white balance for my particular unit is accurate at ~7250K. Is there a particular reason why white balance should be set at 5000K? WB between most CRT displays will be quite different, even among the same models (variance in factory defaults, user settings, age, etc.).
2 Likes
Yes! Would love to see pictures of that display! We can certainly accomodate WB 7250K - it was more from the perspective of the camera using a neutral WB but I should probably update this. The most important thing is that we know the WB value the picture was taken with - same for the rest of the settings - we just need to know the values.
1 Like
I’ll upload some pictures from actual games sometime over the next few days, but here is a single pixel from the SNES 240p Test Suite convergence test pattern. The 256x224 image takes up the entire 4:3 screen with zero overscan or underscan. This was taken at 1/13s, f6.3, ISO100. Before I took this, I took a photo of the IRE100 test screen in RAW, fully defocused to prevent moire, in order to get a full white frame. Checking the RAW data with lightroom, I get a WB value of 7100K with +20 tint. Therefore, I ensured this picture was taken with those exact settings. Pictures I’ll take in the future will use these white balance settings unless otherwise noted.
I tend to take the images of a static screen at slow shutter speeds from a tripod. This ensures no rolling bars appear. With my camera I can never get 1/60s to appear without some partial darkening of the full screen.
I also took a quick measurement of the display before taking the pictures, and on the 100 IRE pattern from the test suite I got a reading of 100 nits. The monitor can get a lot brighter, but this is at a comfortable brightness. I mention this because the scaliness decrease significantly in size at full light output. I doubt anyone would want to use the monitor that way, however, as it’s uncomfortably bright.
9 Likes
Amazing picture! Really great work. You can easily see the vertical deconvergence, fall off and length of the phosphors. Also you can start to see the horizontal fall off.
Interesting you got a reading of 100nits - I always find it amazing how low measured brightness of CRTs are when quite patently they’re not or at least perceived not. I guess a mixture of most of the screen being off and it being a pulse display leads to these low figures whilst needing sunglasses to view them comfortably.
It would be good to get a gist of how the scanlines change shape as the brightness goes up if that’s at all possible? Also an image of something ‘a bit more interesting’ as in colourful so we can see the interplay between various colours would be good - ‘standard’ ones seem to be based on images of Link from the SNES but whatever your choice is I’m sure we can work with.
4 Likes
Thank you very much! Yes, 100 nits is not a lot, but for this display it’s just about right for evening use in a dim room.
Please find an image of Link here. The settings I used to take this image are the same as the prior image except for a shutter speed of 1/4 sec.
To preserve phosphor highlight detail, the image is rather dark. Shadow detail is present in the image along the phosphor strips, however. It is best viewed at 1:1. I can take photos farther away from the screen, but considering I only have 20 megapixels to work with, this will come at the cost of phosphor detail. That being said, I should be able to capture a slightly larger portion of the screen while still retaining detail.
I hope this is at all useful. When I have some downtime I’ll provide exposures with nit readings of the display at different contrast settings to demonstrate the effect it has on the scanlines as the phosphors expand vertically.
4 Likes
Wow! Your pics are immaculate! You really seem to know what you’re doing! Thanks for sharing! Looking forward to seeing more of your stuff!
Perhaps some video captures off the screen from that same 6D might be interesting as well.
3 Likes
I can’t find the source now, but I once read somewhere that the peak is like 10,000 Nits, but this is for just a few nanoseconds, followed by an abrupt drop off.
3 Likes
Thank you. It’s a fun project since it’s a combination of some of my hobbies: games, display technology, and photography.
(Looks like a small convergence adjustment is needed here. Not noticeable from normal viewing distance.)
I will give this a try in the future. Unfortunately, the 6D is a pretty old camera at this point and it was released right before video became a major focus of higher end cameras. This means it’s not as pleasant to use in video mode (focusing is less than ideal) and recording is limited to 1080p/30. I’ll see what I can do.
4 Likes
Alas! I understand, well perhaps you might want to stick with a camera that does 60fps then in that case.
2 Likes
Really great pictures!!! These are fantastic and will really give us a way to recreate these screens in ‘digital’ form (probably more accurately discrete) for the future.
I need to make a Sony Megatron preset with of this display with these shots. Vertical deconvergence is pretty standard I wouldn’t worry too much about it - horizontal is a lot more damaging to image quality.
3 Likes
I decided to perform a very tedious task and count the individual aperture grill lines (each line being a triplet consisting of 1 red, 1 blue, and 1 green colored phosphor strip). I counted 1265 lines on this model. I checked the specs after. From the manual:
BVM-20F1U
- Aperture grille pitch: 0.3mm
- Effective picture size: 387 x 291 mm
387 / 0.3 = 1290 lines
I suppose I’m off by 25 lines in my count, or their aperture grille pitch figure is slightly optimistic. Regardless, it’s not a huge difference.
I took a picture with the display turned off to show detail of the individual lines:
The display is rated at 900 TVL.
The nice thing about these BVMs is that convergence adjustments are easily made. I just didn’t catch this slight deconvergence until I looked at it magnified through a macro lens. I have made the corrections since I took that photo.
4 Likes
Maybe, maybe not though as there were always parts of the screen hidden on all edges of every TV. Maybe that doesn’t really happen on a BVM though?
2 Likes
It’s interesting the phosphor lines aren’t evenly spaced - I wonder if that is on purpose or because of manufacturing limitations.
1 Like
It’s probably on purpose in order to reduce color bleed (or in other words, produce a sharper image.)
3 Likes
I tried to take photos of my setup with your suggested values of 1/60 shutter speed at iso 100 and 5000k white point, but it’s impossible. With shutter speed at 1/60 I need at least iso 600 - 1000 to represent a realistic image. Also 5000k white point is not the best, my TV is calibrated at 6500k with a brightness of 120 cd.
Nevertheless, here are some shots of my calibrated Bang & Olufsen MX 8000 TV. The game runs on a Linux PC with RetroArch and a DVI to RGB Scart Cable.
9 Likes
The entire screen is visible with the bezel attached. I had to remove it recently to check if something was rendered outside of the overscan area and found there were no additional lines visible.
The above shows how I performed the count. It’s not the most scientific or elegant method 
but I believe the resulting count is accurate. I drew a small red line after each group of ten lines on the screen and marked at increments of 100. You can inspect this image at 1:1 or greater to see the lines. The 20 megapixel count (5472 width) makes it impossible to see individual phosphor lines from this distance, but the uneven spacing (or rather, the larger gap between triplets) makes the lines clearly visible. If there is ever a reason to do a recount, I’ll make sure to do this in better lighting conditions and perhaps at greater magnification.
This sounds plausible to me. Greater spacing should reduce color bleed between each set of RGB phosphors.
4 Likes
Fantastic shots! Thanks for these really show off the MX8000 up close!
The white point is more for the camera than your TV to be honest - it’s to give a standardised neutral camera setting so we can compare and contrast between photos safe in the knowledge the white point is the same on the camera. You can certainly argue what a neutral white point value is of course.
Looking at your pictures what I would say is that they look very warm for 6500K. The MX8000 tends to do that in my experience so that’s quite normal for this set - it’s not a professional display and much more about what appeals to people.
Lovely CRT and great pictures thank you!
3 Likes
At first I have to say, I love your project. So please don’t get me wrong if I don’t agree with your camera settings. Oh, and English isn’t my native language, so if I sound rude, I don’t mean it this way.
I measured the MX8000 with a colorimeter. The native white point was way to cold at over 9000k when I got it and the gamma curve was way off too. I own a MX7000 and MX6000 too and they both have a relatively warm tube with a native white point of around 6800 - 7000K. Side by side, there was a huge difference. A fried of mine, owns a MX8000 with a warm tube. So it seems like they were using different tubes in the same sets.
That’s why I decided to calibrate all of my tubes to the same standard.
I calibrated the MX 8000 with professional equipment to a white point of 6500k, because that’s close to the native NTSC US and PAL white point of D65. The gamma curve and the color representation is calibrated almost perfectly. Took me some hours and wasn’t easy. I had to open the TV and tune the voltages on the neck board to get there. On the MX6000 and 7000 it was way easier because of the warm native white point of the chassis.
That’s why I took my photos with 6500k. It’s the white point almost every console and PC is designed for. I don’t know why you decided to use 5000k as a standard. 6500k also is the standard for sRGB on the PC, where we’re comparing these pictures.
Now to the brightness. I tried to calibrate to 120 CD and it wasn’t possible to go this bright without losing details in bright areas and without fucking up the black point and gamma curve. Most CRT TV’s have their brightness at around 80 - 100 CD max. That’s what I measured with my colorimeter after calibration.
I get why you chose 1/60, it’s the only way to not have horizontal stripes of the beam on the picture. But taking a photo at ISO 100 with a shutter speed of 1/60 will result in an almost black picture. My camera couldn’t even focus because the picture is way to dark.
So if you ask me, it would make more sense to take the photos at 6500k white point and with free iso value. Not every tube has the same brightness, I needed ISO 600 to 1000 to make the photos look like what you would see on the TV with your eyes.
2 Likes