- This topic has 47 replies, 6 voices, and was last updated 1 year, 7 months ago by
.
-
Posts
-
How many fields have you used in uniformity test? For 32″ display there should be 7×5 at least, 9×7 for good error representation. Switch central selector in the test page to delta C (chroma), it is more important, than lightness (L), C difference up to 1.2 is acceptable for neighbor fields, but wish to have 0.9 or less, maximal delta C (center filed vs. any other one) should not exceed 2.3. Lightness fall up to 10% may be acceptable. Border fields are not valuable, you don’t place image there. Panels with no UC usually exceed these values. You can simply set light, than dark background and check whole screen field by your eyes, looking from 3m/10ft in orthogonal direction. 500:1 contrast is too low for editing images that will be displayed on computer/smartphone, cause typical IPS contrast ratio is 900…1300:1, but it could be usable for prepress, cause gloss paper does not exeed 280:1.
How many fields have you used in uniformity test? For 32″ display there should be 7×5 at least, 9×7 for good error representation. Switch central selector in the test page to delta C (chroma), it is more important, than lightness (L), C difference up to 1.2 is acceptable for neighbor fields, but wish to have 0.9 or less, maximal delta C (center filed vs. any other one) should not exceed 2.3. Lightness fall up to 10% may be acceptable. Border fields are not valuable, you don’t place image there. Panels with no UC usually exceed these values. You can simply set light, than dark background and check whole screen field by your eyes, looking from 3m/10ft in orthogonal direction. 500:1 contrast is too low for editing images that will be displayed on computer/smartphone, cause typical IPS contrast ratio is 900…1300:1, but it could be usable for prepress, cause gloss paper does not exeed 280:1.
I just used 3×3 (it’s a 27″ display).
I have the central selector on “average luminance & deltaC*00”
There is a C difference of up to 4 (maximum) and 2.9 (average) on the worst square, the others aren’t so bad. I have attached the report.
But maybe I need to repeat the test with 7 x 5?
Thank you for the fast reply, much appreciated.
To check i understand what you mean about using the community corrections, is it right that I should be looking at the one on the left of my attached screenshot, and not the one on the right (because it looks to be measured in Rec. 709 preset)?
Yes
Regarding the Spyder bundled corrections, in Instrument mode I am offered LCD CCFL Wide Gamut, or LCD RGB LED, that seem best to match, would it be one of those?
RGBLED, not equal to.. but it’s the closest one.
Regarding enabling UC it depends on
-how bad it is monitor without UC and with UC active regarding uniformity
-contrast with & without, specially if you want to use it fro web or othe rmultimedia puproses
-how far native white is from your target. 6dE for example can be attained by GPU channel limitation without issue (excluding banding on some GPUs)Most often i will be using it to edit content for web (photos, illustrations).
A small amount of video stuff, again for web.
And sometimes I print stuff, but mostly this is just on my own, not very special printer, and it doesn’t need to be dead accurate. I’m more interested in colour accuracy from the point of view of stuff that will be seen online on a screen.
I think this means that I want to retain a reasonable level of contrast, and contrast appears to be significantly better with UC off. I think I will do another uniformity measurement but with a 5×5 or 5×7 grid this time.
I need to do some reading-up to better understand the significance of white point and where it matters most.
C difference up to 1.2 is acceptable for neighbor fields, but wish to have 0.9 or less, maximal delta C (center filed vs. any other one) should not exceed 2.3. Lightness fall up to 10% may be acceptable.
A question … are the values you suggest here the values that would be given by the DisplayCAL measurements as the “average” ones – or the “maximum” ones?
Coming back to this after some months and having now swapped my Spyder5 for an i1 Display Pro (aka i1d3?) …
I am once again trying to work out the best “correction” to use.
From reading through this and other threads, of the standard corrections offered to me, I think the closest is likely to be “Spectral: LCD PFS Phosphor WLED IPS, 98%Adobe RGB/96% P3 (HP DreamColor Z24x G2” – is that right?
Or am I better just to use “Spectral: LCD PFS Phosphor WLED family”? This one sets the instrument mode to “LCD generic” while the other sets it to “refresh generic” which doesn’t seem right.
If I use a “community” correction then I should check
- does it look like it’s been run on the native gamut (ie. the RGB points on the graph don’t match up exactly with the profiles for AdobeRGB/Rec 709 etc
- it should be “spectral” type
- is the resolution 3 or 10nm (is 3 better?)
Calibrite Display Pro HL on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.From reading through this and other threads, of the standard corrections offered to me, I think the closest is likely to be “Spectral: LCD PFS Phosphor WLED IPS, 98%Adobe RGB/96% P3 (HP DreamColor Z24x G2” – is that right?
Or am I better just to use “Spectral: LCD PFS Phosphor WLED family”? This one sets the instrument mode to “LCD generic” while the other sets it to “refresh generic” which doesn’t seem right.
It uses a LED hybrid so none of them. If you want the best. If you want aclose one, use WLED PFS family and it’s done.
If I use a “community” correction then I should check
- does it look like it’s been run on the native gamut (ie. the RGB points on the graph don’t match up exactly with the profiles for AdobeRGB/Rec 709 etc
- it should be “spectral” type
- is the resolution 3 or 10nm (is 3 better?)
yes. If you plot it (specplot-exe or Excel) you’ll see it’s an WLED PFS hybrid with certain amount of other phosphor close to the ones used in a GB-LED that raises floor in SPD for red channel
Thanks. If I’m trying to choose between the “community” ones, does it make any difference whether they are described as 3 or 10nm?
As as said many times, it is an WLED PFS hybrid, hence 10nm is not enough to capture the spectral power distribution. Plot them and you’ll see.
As as said many times, it is an WLED PFS hybrid, hence 10nm is not enough to capture the spectral power distribution. Plot them and you’ll see.
Sorry if you are having to repeat yourself to people like me who do not have a full understanding of all this. I unfortunately don’t know how to plot them (either using excel or specplot.exe).
I know that if I look at the available corrections via the DisplayCAL interface, it lets me select one and choose “colorimeter correction information” which lets me see a graph. I’m assuming that the Y axis is “spectral power”? And I can see that on some of them, the plotted curves are smoother, which I would assume suggests a higher resolution. But then I get confused because some that are labelled as 3 or 3.3nm (eg the one on the left of the attached screenshot) seem to have more jagged curves than some that are labelled 10nm (eg the one on the right of the attached screenshot).
Perhaps it is safest for me just to use the generic “LCD PFS Phosphor WLED family” which I understand is not really correct but maybe “close enough” for someone like me who doesn’t entirely know what they are doing.
The issues are mostly in red, dual narrow spikes from PFS backlight, but raised floor due to being an hybrid.
Ok… but I am still confused about this:
I can see that on some of them, the plotted curves are smoother, which I would assume suggests a higher resolution. But then I get confused because some that are labelled as 3 or 3.3nm (eg the one on the left of the attached screenshot) seem to have more jagged curves than some that are labelled 10nm (eg the one on the right of the attached screenshot).
Ok… but I am still confused about this:
I can see that on some of them, the plotted curves are smoother, which I would assume suggests a higher resolution. But then I get confused because some that are labelled as 3 or 3.3nm (eg the one on the left of the attached screenshot) seem to have more jagged curves than some that are labelled 10nm (eg the one on the right of the attached screenshot).
Who cares about those screnshots in green channel? Red channel can be captured (more or less 1nm wil be ideal) at 3nm, it can’t ant 10nm , period. Plot red channel.
There are other threads, maybe UP2516D, maybe Benq SW271C with the same old and now abandoned PFS hybrid backlight, with SPD plots. Search them.
Yes I can see that in the red channel, more detail seems to be captured in the 3nm one (three separate peaks).
What was confusing me is that the graph in the 3nm looks more jagged than in the 10nm one. I can see that the 3nm is capturing data at 3nm intervals. Maybe the 10nm has some kind of smoothing applied to the basic data.
