[bossanova808]

I wouldn't immediately say so, no.

I just looked at dispcalGUI's docs, there are some quite advanced options for black point and gamma there. And some mildly controversial statements too....but basically, I'd need to read that several times over just to fully grock what they are saying, and I can see that there are options there that could cause quote non-standard behaviour near blacks.

I should really try it out, basically.

(also - I haven't really inspected monitor profiles in detail for a while - I am much more focussed on the print side of things and all the new really nice monitors have direct hardware calibration now, so there are no vgct tags in the profiles anymore. I am embarrassed to say right this moment I even can't fully recall the implications of the white and black pints - e.g. white is always set to the theoretical L* 100 even though of course this is not an accurate measured white, but I can't fully recall precisely what happens with black point and BPC and monitor profiles right now....there is theoretical black (o input = total absense of light output = 0, and then there is reality 0 input = some minor light output...which is where BPC/scaling comes into play...gets complex fast).

I'd definitely just take the profile out of play on the Pana just to see if it is the root cause...and if so, re-do at least that one (and the air if you see the same) - and pay careful attention to your black point settings...

[jingai]

I just looked at dispcalGUI's docs, there are some quite advanced options for black point and gamma there. And some mildly controversial statements too....but basically, I'd need to read that several times over just to fully grock what they are saying, and I can see that there are options there that could cause quote non-standard behaviour near blacks.

I should really try it out, basically.

If you do, I'd really appreciate your take on what you think I should use for settings.

there is theoretical black (o input = total absense of light output = 0, and then there is reality 0 input = some minor light output...which is where BPC/scaling comes into play...gets complex fast).

Yes, it does get complex.. and I don't think I understand it enough to make choices past the defaults for the white and black points.

But basically, my understanding is this: dispcal goes through the cal process first, where it is taking measurements every second from the colorimeter. You adjust the color, contrast, brightness, etc on the display's controls while it's measuring it and try to compromise between color accuracy and maximizing contrast. At the end of the cal process, it uses the measured white and black points for the profiling process, which is totally hands-off (just measuring patches).

You can alternatively set the white and black points in dispcal, but from what I had understood, you'd be sacrificing contrast by doing so. Does this sound vaguely right?

I'd definitely just take the profile out of play on the Pana just to see if it is the root cause...and if so, re-do at least that one (and the air if you see the same) - and pay careful attention to your black point settings...

I don't really understand what you want me to try here. What I was suggesting is that XBMC shows a brighter image on the Panasonic (and my MBA) than what I see in Aperture on the HP. XBMC doesn't use the profile at all, so even if I discarded it (which I can't really, on OSX -- I'd just be picking the Apple-supplied profile for the display, or generic sRGB, etc, but it won't let you run with no profile at all) it shouldn't change the output.

I think I may be misunderstanding what you want me to try though?

[bossanova808]

Yep that does sound vaguely right and how the other systems do it too. But of course you would always leave things at max contrast (in this domain, in print definitely not).

If your profile is the thing raising the black levels, then switching to say GenericRGB on the mac would solve the black bump...that's all I am saying. Then you know it's the profile. So it's just a very simple test to work out why XBMC on the pana is doing this. As you say, it's not happening on your HP right? So I'd try without profiles on both the others.

Sorry if I got confused/confusing - I didn't realise so much initially there were 3 screens involved!!

[jingai]

If your profile is the thing raising the black levels, then switching to say GenericRGB on the mac would solve the black bump...that's all I am saying. Then you know it's the profile. So it's just a very simple test to work out why XBMC on the pana is doing this. As you say, it's not happening on your HP right? So I'd try without profiles on both the others.

But if the profile is the cause, I'd think I'd have to disable it on the display on which I edit and re-edit the photo to see a difference, yes? Since XBMC doesn't use the profile for output at all.

[bossanova808]

The profile has vgct tags - these are tags (LUT tables, really) which manipulate the video cards output *at all times* - the profile transform (or absence thereof) is a layer on top of this. It's these tags that are funky in your air profile at least.

Basically, I have a tool here which lets me poke any profile into a video card and see what happens and as soon as I put your air one in, weird things happen, and to a lesser extent with your TV one too. Those adjustment are happening to all output through that card...so even if the numbers transform isn't happening, the profile is inducing big changes in your output. Simply switching between then should make this immediately obvious, I am pretty sure on the Mac the profile LUT loading is instant when you do this in ColorSync...

[jingai]

The profile has vgct tags - these are tags (LUT tables, really) which manipulate the video cards output *at all times* - the profile transform (or absence thereof) is a layer on top of this.

Ahh, I didn't realize. I guess I should have though.

Ok, I switched to the Generic RGB profile on the MBA. Aside from the now-horrendous blue cast to everything (why is this standard?), the brightness is pretty much the same. BUT... re-doing the test I think the viewing angle is what prompted me to suggest it was 2 stops brighter. I tried harder this time to look at it head-on, and I'd suggest it's more like half a stop. For low-key photos like this though, it's still enough to bother me, but maybe I've just been staring at it too long.

I still think it'd be a good idea to make XBMC profile-aware, ultimately. I'll redo the calibrations for these two displays, but I'm going to guess it's my colorimeter that's at fault for the oddities you see in the profiles, since you say the HP one looks normal (the only display that isn't LED-backlit).

Would it help if I took a photo of the two displays showing the same photo? I promise you the MBA doesn't look as wonky as it sounds like you think it does lol.

[bossanova808]

Yeah I would like to have a swing at it myself but I am not sure I will have time...it's very hard for me to get longer than a few minutes at a time unless I cheat and do it at work. I suppose this is vaguely work related though The real man to convince would by Jonathan - he could probably knock something up in a few horus that would take me weeks and many tears to do a half-arsed job at!!

LED backlights and Spyders (3 and below) are a notorious issue. I have never yet seen a Spyder do a good job with an LED except the Spyder4 which is...tolerable...but not great.

Yep take a photo - can't hurt and might at least give me a more concrete idea of what you're seeing if you can get the photo to reflect it properly...

[jingai]

LED backlights and Spyders (3 and below) are a notorious issue. I have never yet seen a Spyder do a good job with an LED except the Spyder4 which is...tolerable...but not great.

You suck. It's going to bug me now, and getting my wife to let me buy a $950 camera this year was hard enough lol..

..still it's not horrid.. and would be mostly ignored if XBMC were profile-aware. But I realize also that I'm probably the first one to bitch about this, so I doubt (and totally understand) it'll ever be a high priority

BTW, I'll google it tomorrow, but this direct hardware calibration thing sounds interesting.. . how does it work?

Yep take a photo - can't hurt and might at least give me a more concrete idea of what you're seeing if you can get the photo to reflect it properly...

Will do. Tomorrow is my wedding anniversary, so it will probably be Saturday or Sunday before I do it. But I would like to know what you think.

[bossanova808]

DHC is not dissimilar to normal calibration, the main difference is in the profile but. There is again calibration (although through software doing direct manipulation of the monitors hardware by say, SpectraView 2 (NEC) or ColourNavigator (Eizo). Followed by a profiling stage that uses the (up to) 16 bit 3D LUTs in the monitors. Same same, but different - it just bypasses the relatively crude adjustments do-able in 8 bits and single LUTs and does it in all the LUTs in 16 bit, basically. The result is more accuracy and better smoothness (and particularly noticeable with wide gamut monitors).

DHC is also fast (especially so with the i1Display Pro) - and fully automatic - no manual fiddling with RGB levels or whatever. Once you go DHC you can never go back Some of the Eizos have calibrators built in and can wake themselves up at night, calibrate, and turn themselves off again - very cool!

Ideally this goes with 10 bit output at the video card end of things, but that's PC only so far, and hit and miss even there - you need workstation level graphics cards and (weirdly) you lose things like transparency in Aero.

[jingai]

That sounds... awesome. Hopefully one day all displays (TVs too!) do this.

[jingai]

I re-did the calibration and profile on my MBA today. It's here, but it looks pretty similar to the last one. I did it at 120cd/m2 again (mostly to compare with the previous one) and I did notice that during the calibration, it measured 119.92cd/m2, but after profiling it reported it was 110cd/m2. So I bumped the backlight .2% and it reported 120cd/m2 during the cal, but 119% after profiling.

Do you know which I should pay attention to? Maybe this is due to the LED backlight? I ended up trusting the first measurement because it appeared off otherwise (relative to the HP).

If you're curious btw, the display characteristics as measured were:

* 60.4% sRGB coverage,
* 6550K color temp,
* 119cd/m2 white point,
* 0.30 black point,
* 388:1 contrast ratio

So as you can see, it's not a great display. But really I bought this thing because it's tiny and lightweight, not for the screen. So long as it's roughly close to the HP (appears to be to my eye) I'm OK.

I'm redoing the HP now as well. I'll post it when it's complete.

[jingai]

Here is the new profile for the HP.

[bossanova808]

From what I can see here at home, those look much the same.

At a guess - probably yhe variability in the figures is from backlight variability - low end screens vary quite a bit in backlight over time. Given you're probably using the air just as a temporary judging thing before you do proper work on your HP, I'd just go with the best visual match and not worry to much, which it sounds like you've done.

[jingai]

Yeah, the Air is just for on-the-go stuff. I always follow up at home on the HP.

Thanks for all your help btw. Even just confirmation helps my peace of mind

[bossanova808]

No worries at all!