February 2, 2018 at 1:13 pm #46671
I didn’t make a post after confirming the C32F391 purchase worked. You might have been confused by the fact the forum may have re-arranged things due to the change of time zone. So it put one of my posts in the wrong place, but it has moved back to the chronologically correct position now.
Regarding DisplayPort (not Data Port) and HDMI, as long as the signal is corrected (easy to do) then they are very much equivalent. The only exception is if the monitor makes use of technologies specific to one (FreeSync, G-SYNC, HDR capabilities, refresh rate at native resolution etc.) That doesn’t apply for the monitors you’re considering here.February 2, 2018 at 3:41 pm #46672
PCM2: Another question. We all know from what you’ve written before that a so-called screen shot (and by inference, sampling of a captured portion of a screen) is not actual data from the monitor itself. My question is a simple one: Are the data sampled (e.g., by using the eyedropper tool in Photoshop) from an image (1) input to the GPU, or (2) output from the GPU?
The reason I ask is because the calibrations on my monitor are uniformly bad, whether I pick a white point of 5800 K or 6500 K, and luminance of 120, 140, 160, or 180 cd/m^2. I want to know if it’s the GPU that’s screwing things up or the monitor (or possibly both, if the answer is “input to the GPU”).February 2, 2018 at 4:35 pm #46673
That’s a good question. Could it be both? It’s difficult to define what is ‘input’ or ‘output’ when talking about something like a GPU. It isn’t something you see or actively connect with as a user and is constantly communicating bi-directionally with other components. How are you deciding what constitutes a ‘bad calibration’ exactly?February 3, 2018 at 4:36 pm #46678
PCM2: Correct me if I’m wrong–please–but my impression is that (1) the browser reports a set of colors from, e.g., the Lagom test pages; (2) the operating system then transmits these colors to the graphics card (aka GPU); (3) the graphics card sends a set of colors–hopefully the same ones as in (2) to the monitor, which then (4) displays the colors on the screen. My question was intended to determine from which of the first 3 steps above the color data come from that are sampled by , e.g., Photoshop’s eyedropper tool (we know they don’t come from step 4–you’ve made that clear and I can see why that is so). I suspect that the colors are sampled from video memory (by Windows 10), but that’s just an educated (?) guess.
What do I mean by a “bad calibration”? Let me count the ways. I’m looking at the forum screen as I write this, and the posts are in a light green-gray, almost white but not quite. The dates and times of each post are in a slightly darker shade of this greenish light gray color. The background is white. If I snip these three colors and copy them into Photoshop as a new “Image from Clipboard” and sample each color with the eyedropper tool, I get these results: all three are perfect gray scale colors–the R, G, and B values are all the same–and they range from 255 (the white background) to 250 (background of the posts) to 244 (background to the dates & times). There is no way I would describe the latter two colors as any kind of gray. They are pale green.
If I copy the Gradient test from Lagom into Photoshop, what do I see? First, I’m struck by the banding. Lots of it. Some wide, some narrow. To my erstwhile artist’s eye very few of these colors look like a neutral gray. Most have a color cast, varying from greenish through purplish all the way to pinkish. The darker colors deteriorate into a blackish mess for RGB values (all the same still) of 16 or less. At the other end of the scale, from RGB=242 upward, there is a wide band of almost white, but it has a distinct greenish cast until it hits the white end of the strip. Mind you, this calibration was done with Datacolor’s “Better” Gray Balance Calibration rather than the recommended “Fast” setting. The results are even worse if I use the “Fast” setting. On the Lagom White saturation test, all of the checkered patterns (except the first one) are a pale green-gray, but the last three are invisible (I can see the 252 pattern faintly if I use averted vision). All are reported as pure gray by Photoshop.
I have a hard time believing that these results represent a decent gamma = 2.20 curve, although that is the calibrated value reported by the Datacolor software. In fact, the Lagom gamma test yields these results: at 48%, R=2.8, G=2.6, B=2.7, gray=2.5; 15%, G=2.3, gray=2.4; 10%, G=2.6, gray=2.2. Only one 2.2 hit out of 8 readings. Rather putrid, I’d say.February 3, 2018 at 4:56 pm #46679
As I’m on vacation and because this is verging towards off-topic, I will keep this brief and, as far as possible, specific to the monitor side of things. It’s also going to be the last comment I make on this particular topic for a bit. Three issues here:
1) VA monitors are not consistent in their gamma or colour temperature performance. Try viewing that same shade you’re observing at a different section of the screen (if you snip it into photoshop so it takes up a small section of the screen then shrink the window down, it will be easiest). It may well appear completely different. I go to great lengths to explain this in any reviews on VA monitors and summarise it in the relevant article on panel types. It’s why I never recommend them for colour-critical work. So you can spend as long as you like trying to get the central point perfect, but the problem is your eyes are very different to a small device that is looking directly at the centre of the screen.
2) Calibration is not a perfect process, especially when it’s done on a software level. True colour professionals will much prefer monitors with hardware calibration where the LUT is directly addressed vs. making changes on the GPU level for this reason. If the profiling process needs to make significant changes vs. the native values of the monitor then issues with grey neutrality and banding are par for the course.
3) Some browsers have odd colour management. Try viewing the forum in a different browser as well to see any inconsistencies that might be caused by that.February 14, 2018 at 2:07 am #46719
Sorry if I messed up your vacation. Bummer!
I just wanted to let you know two things: (1) I recalibrated my C32F391 one more time. This time I got a fairly decent calibration for a change. The date and time bands in this forum are now a nice neutral gray, but the posts themselves are still a rather greenish color. The Gradient test in Lagom looks much better than before: less banding. On the Contrast test, I can see the #1 block easily for yellow and gray, faintly for green and cyan, but not at all for blue, red, or magenta. In fact #2 hardly shows for blue. The high end is best for the red and magenta, and fairly good for the blue. The others have the last two shades too close together. On the “Black level” test I can easily see #7 and up; down to #4 or #5 if I turn down the lights. In the “White saturation” test I can see all but the last set of little squares. They blend into a greenish gray and become distinctly green from 247 on up. All in all, it’s not too bad. (2) Amazon just dropped the price on the C32H711 to $420 (less a penny). I jumped on it and it should arrive tomorrow (Wednesday, Feb. 14 here). I’ll let you know what I discover once I get it hooked up and running.
BTW, in the Lagom Sharpness test, most of the little squares appear to “shimmer” even when the image is not being scrolled (colors appear when scrolling). If I click on the “Toggle lines only” (which is underlined in blue below the test image), the little black and white rectangles in the middle of the inner circle disappear and the rest of the test display becomes rock steady. (I can’t fix the sharpness, because that setting is now grayed out in my OSD menu.) Is this behavior an indication that the C32F391 uses PWM? I’m suspicious; but if it means PWM, then I’d have to say that it doesn’t really bother me in everyday use.February 14, 2018 at 6:19 am #46720
You hardly messed up my vacation, haha. Don’t worry. 😉
Although you did just mess up my day slightly as no orders for that model were credited. Never mind.
That behaviour could simply be an inversion artifact (imperfection in voltage regulation). Quite common on many monitors. A better test for PWM is this – https://www.testufo.com/blurtrail. Observe the line at ‘100’ brightness vs. lower brightness and see if you see it separate into multiple lines ‘per line’ at lower brightness. Most smartphone cameras would show ‘flickering’ at lower brightness vs. not as well. Be sure to compare to a much lower brightness as well as this monitor could use hybrid-dimming (so PWM below ’30’ or so brightness, perhaps).February 14, 2018 at 3:27 pm #46727
PCM2: Weird! Something Windows did last night canceled the calibration. Yet clicking on Display Properties shows that the profile is still the same (#4—I numbered the profiles after a while so they would show up as separate choices, i.e., different profile names). Settings (such as NVIDIA card allowing “other apps” to control the color, etc.) are all still the same. What is causing the profile to disappear is a mystery to me. There must be someplace else that Windows 10 is getting its profile information, and it has reverted to the original (uncalibrated) profile.
The brightness was set at 75 (with the recent calibration, that setting seemed to work well), which is now too bright, so I whacked it back to 55. Then I looked at your “Blurtrail” test. If I do not track the moving line(s) I see about 3 or 4 separate lines going by. If I track it (as instructed on the website), I see a vertical band of gray. It stays the same width for pixel settings of 1 through about 8. Then if I increase the number of pixels, the band begins to get wider, and blurry along the edges. Lowering the brightness to 20 did not change anything (except everything is dimmer at that setting). I still see 3 or 4 vertical lines looking at a fixed point on the screen, and the same vertical gray band as before—just dimmer.
The lines I see while looking at a fixed point may be partially an artifact of my eyesight. I put the cursor in the middle of the display and looked at it fixedly. I then saw two bright vertical white lines (pixels = 1) about the same distance apart as the width of the gray bar I see if I track the moving image. The extra lines, maybe just one or two, look like echoes or something. They are not as well-defined as the two bright vertical lines. They trail the two bright lines. Does any of this make sense to you?
BTW, when I was checking on the new monitor I had to refresh the page for that model, and the new, lower price appeared. This may have mucked things up. I made up for it by donating $200.00 US by EFT (via PayPal) from my bank account to the PC Monitors website. I’m sure that’s more than you would have gotten had the sale gone through your link rather than “getting lost in the ether.” 😊
The advice and help you’ve already given me seems worth that much to me.
More: I went back to Display Properties and changed the profile from “Samsung 32CF391-4” (the latest calibration—I misnamed the monitor the first time I calibrated it, and the name stuck; no matter) to an earlier one (Samsung 32CF391-3). The screen color immediately changed. Then I changed it back to #4 (the latest profile). When I went back to this forum, the colors were back to what they were before: light neutral gray in the date & time sections and pale green in the posts themselves. Just as I suspected it would, this action restored the latest calibration profile. Evidently Windows gets its profile information from some other place than Display Properties when it resets things. Changing it manually (in Display Properties) restores the calibrated setting.
This post as it’s being composed has a white background. The pale green background is added when I click on “Submit.”
The profile now in effect (Samsung 32CF391-4) has now disappeared from the choices in SpyderProof! The latest profile in the drop-down list is now “Samsung 32CF391-3.” And this despite what the page says: “This dropdown list contains any monitor profile you’ve ever created with the Spyder5 calibration.” Yeah, sure it does! I suppose the “Samsung 32CF391-4” profile was just my imagination? Go figure!
All of the calibrated profiles appear in the drop-down list in Display Properties. There is also a new one I never saw before. It reads: “sRGB display profile with display hardware configuration data derived from calibration.” Well, it wasn’t derived from the Sypder5 calibration, because if I select that profile the screen goes back to the way it was this morning. Using that profile, all the checkered patterns disappear from the Lagom White saturation test, and the black crush is unbelievable: only the square at value 40 is visible, and it’s a very dark purple color. I’d say that profile is about as useful as you-know-whats on a boar.
As it appeared this morning, all of the fields in the posts, etc., looked white. Evidently the “raw” calibration has excessive “white crush” that washes out the subtle differences in the various fields in the discussion forum (background = 255, posts = 250, date and time = 244, all neutral gray, or R=G=B). In other words, a gray with value 244 appeared white in the original profile that came with the monitor. All of this just gets curiouser and curiouser!February 14, 2018 at 3:59 pm #46728
Firstly, I’d just like to say a big thank you for your very generous donation. That does indeed more than make up for any little technical issue that occured when you were on Amazon. And made my day a lot brighter, so I truly appreciate it!
I have to admit that, sometimes, the calibration process in Windows can confuse even me when used in conjunction with third-party utilities. That includes the Spyder software. It seems that Spyder can sometimes overwrite things or for whatever reason the data from the ICC profile isn’t properly loaded. If you’ve followed all of the steps in our ICC profile article and have made sure ‘Calibration On’ is selected when you right click the Spyder icon in the system tray, things should work correctly. In theory…
I actually like to approach things differently myself. I like to set ‘Calibration Off’ in the Spyder utility to stop it interfering with things. I then use the ‘DisplayProfile’ utility mentioned at the bottom of the ICC profiles article to manage the active profile. If you select the profile with this, it should work properly and the last used profile should also activate when you first start Windows. You need to also make sure you aren’t running any utility that would interfere with ICC profiles (it’s not just the Spyder utility that can do this). For example F.lux. Sorry if you’ve already told me this, but could you also confirm which version of Windows you’re using?
Regarding the PWM test from TestUFO, it is indeed normal to see multiple vertical bands if you keep your eyes still rather than tracking the band. If a monitor uses PWM at the sort of frequency Samsung uses, you’d see the band break up or appear differently whilst tracking if PWM is being used. If it appears as a fairly solid grey thing as you describe and doesn’t appear different at various brightness settings, it’s very likely that the monitor is PWM-free and instead uses DC dimming. PWM artifacts can appear on moving edge tests like this at surprisingly extreme PWM cycling frequencies as well. They can be harder to spot and the spacing qite close, but they are there if you know what to look for. You can also use a more old-fashioned method of wobbling a pencil infront of the screen (white background) or running a fan in front of the screen. And seeing if you can see multiple instances of the pencil or fan blades rather than just a continuous ‘blur’.February 14, 2018 at 5:15 pm #46729
Ah! I see that aliasing is a test (like the wheels on cars that run backwards sometimes—an artifact of aliasing between the 24 fps of video and the 30 fps—60 fps if you count the interlacing as separate—of TV). No, I get a continuous blur, no multiple images. So the C32F391 does not use PWM. But the default profile that comes with the monitor is godawful. I hope the C32H711 is better in that regard. (Hope springs eternal.) I will give you a full report on that beast. It’s supposed to come before 8:00 pm EST (it’s now 12:40 pm EST here).
The system had got back to its old trick of resetting the profile at seemingly random intervals (probably having to do with refresh timing or something like that). I had to go through the ICC stuff again, and sure enough—the wrong profile was listed as the default for Windows. Wouldn’t it be great if we only had to set things once, like say in Display settings? After going through that advanced color management series—again!—the “Samsung 32CF391-4” profile now “sticks” instead of reverting to the sRGB default every so often.
I’m going to check out that DisplayProfile app that is mentioned in the ICC article. Got my curiosity aroused (not a hard thing to do). And yes, I have mentioned the Windows I’m using, but no harm in repeating—it’s Windows 10 Home (nothing in the Pro version was of interest or value to me, contrary to XP where it made a big difference).February 14, 2018 at 5:22 pm #46730
BTW, what does DisplayProfile mean by “Contains gamma table tag” for some of the profiles?
What would be really great (for those of us who like to go where angels fear to tread) would be a LUT editor, where you could tweak the table to get the results you wanted. Obviously, there would be a lot of trial and error to that sort of thing. But, nothing ventured, nothing gained.February 14, 2018 at 5:28 pm #46731
The gamma table tag just means they’re more ‘complete’ profiles. Likely the ones created by your colorimeter. Note that, more often than not, the profiles that come with monitors are no different to Windows defaults. They’re just there to try to overwrite any other profiles/settings you might be using without realising and reset things to the default state. This will become clearer if you compare them to the Windows default ‘sRGB IEC91966-2. 1’ profile using DisplayProfile and see that they’re the same.February 15, 2018 at 11:41 pm #46734
PCM2: Your last post begs this question: On the Advanced tab of the Windows 10 Color Management window there is a section at the top labeled “Windows System Color Defaults.” It contains two lines of information. The first line, labeled “Device profile,” lists the same profile you mentioned in your last post: “sRGB “IEC91966-2.1.” Shouldn’t this be replaced by a calibrated ICM or ICC profile? The line below that is labeled “Viewing conditions profile.” It contains this line: “System default (WCS profile for sRGB viewing conditions).” It’s from a drop-down menu. The first choice is “WCS profile for ICC viewing conditions”; the second choice is “WCS profile for sRGB viewing conditions.” Somehow, it seems to me that the first choice would be better if we want Windows to use our ICC (or ICM) profile for “viewing.” After all, viewing is what a display is for, right? Your thoughts?
Also, I may be missing something about the DisplayProfile app. I thought it just showed what profile was in use. Can it also compare profiles? Or actually display their contents?
BTW, sorry about that slip of the tongue when I wrote “Data Port” instead of DisplayPort. I knew perfectly well I should have written DisplayPort. Just some kind of typographical error. Probably too many data on my mind. 😁February 16, 2018 at 6:18 am #46736
What you need to understand about ‘Colour Management’ on Windows is that it is full of confusing and unnecessary duplication. You can ignore that section, it isn’t mentioned in the article for a reason. You’ll notice that it doesn’t change when you apply a profile using the DisplayProfile utility either.February 16, 2018 at 11:05 pm #46738
“Full of confusing and unnecessary duplication”? Boy, you’ve got that right! Thanks for the elucidation.
You must be logged in to reply to this topic.