How to calibrate Gamma on a Z2

Discussion in 'Projectors, Screens & Video Processors' started by KraGorn, Dec 12, 2003.

  1. KraGorn

    KraGorn
    Active Member

    Joined:
    Aug 30, 2003
    Messages:
    4,745
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    68
    Location:
    Warrington
    Ratings:
    +27
    After running through AVIA's tests on both my Sony 730 and HCPC, the picture on the PC is a lot brighter but I know that it's not down to the white levels as they're pretty much the same.

    I'm presuming this is a gamma issue but I've not come across a test of the AVIA DVD to set that .. TheaterTek has a Gamma slider which I'm assuming is affecting the PC's picture whereas the Sony doesn't, er so doesn't. :)

    How does one calibrate gamma?
     
  2. buns

    buns
    Banned

    Joined:
    Apr 29, 2001
    Messages:
    6,100
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    58
    Location:
    Belfast
    Ratings:
    +1
    with difficulty.....
     
  3. Mr.D

    Mr.D
    Well-known Member

    Joined:
    Jul 14, 2000
    Messages:
    11,167
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,209
    Ah gamma! ( I have workmates who start crying when I start banging on about this!)

    Imagine a linear mapping on graph.
    Black at the origin , peak white at 1.
    If you grab the middle point and pull it up or down so that the linear straight line bends into a curve . Thats a gamma ( its an easily expressed quotient: the expression of which escapes me at the moment). This is what differentiates a gamma from a curve or a LUT ( look up table) a curve is arbitrary (wiggly no easy way to describe it mathematically) a gamma is a lot more simple. Its wrong (evil even!) to call a gamma a curve and vice versa. ( curves are things of great beauty and power yet dangerous and mysterious to the unwary whilst gammas will quite happily accompany you home after a couple of vodkas)

    Basically video material has a gamma of about 0.4 -0.5 ( historically to do witth the response characteristics of tubes in the first cameras and techniques to adequtely capture a referentially pleasing intensity range)
    Displays historically have the inverse of this materials gammas ie 2.2.

    When you show video material on a display with gamma of 2.2 ( or there abouts) the gammas effectively cancel each other out and you get a straight linear response ( in practice its more like 1.1-1.2 as people just seem to prefer this to a totally flat response of 1 and video actually has a curve to it rather than a gamma anyway but its designed this way to look nicer ( sexy curves remember) anyway and that just confuses all the gamma stuff even more but hey thats the broadcast industry for you!)

    The most confusing thing about gamma is what it actually refers to! When someone says aim for a gamma of 2.2 when you calibrate your display what are they refering to?

    The display itself? The material being shown ? The end result of the material being shown on the display?

    For example on Avia there is a non dithered gamma chart pattern ( next to useless in the real world for evaluating gamma but I digress)(works fine on 50/60Hz TVs but thats it)

    The reason its not dithered is so its not effected by gamma itself. Its made up of alternating black and white lines so as to give boxes of differing overall brightness to match the mid gray solid surround to.

    These boxes have numbers in them that relate to a gamma where the gray surround will match the intensity of the box.

    However the one to aim for is according to Guy Keo ( sorry for sp Guy) who makes Avia is 2.2 ( for a CRT type display but ...

    {very important] this is a given on any display if you want to look at video on it :

    .. the actual display gamma is immaterial as long as you can get video material to look correct on it) However what that block actually represents is an overall gamma of 1 ( ie if you had a display with a gamma of 3 and corrected your video material to look correct on it you would still aim for the 2.2 patch). So you see a figure for gamma is really irrelevant unless you know what its referring to ie : display (hardware) , material, end result (sometimes called end to end or system).
    ( again I repeat this particular pattern is difficult to use with any accuracy so I don't recommend you use it in the real world I just refer to it as in theory it illustrates some things about gamma quite well)

    Standalone players are normally set-up to look correct with video on a display with a gamma of 2.2 ( ie a CRT)

    Most of the recent digital panels I've seen tend to have been designed to show video correctly ( or nearly) albeit with a fudged curve ( most digital panels have a native gamma of 1 by the way so if it has to show video correctly it has to be corrected ( or the incoming video has to be corrected ...same thing really )
    ( I have a suspicion this is why a lot of plasmas posterise as the correction is a little bit severe or clumsy or badly implimented..or all three!)

    What happens when you stick a PC with a graphics card in there?
    Normally graphics cards have a linear output but its not a hard and fast rule ( don't even bring Macs into the equation).

    So you get a linear output of material with a gamma of 0.45 into a display with a gamma of 2.2 ( hardware or corrected)....should be fine you see! ( material 0.45, graphics card gamma 1 :no effect , passes straight to display with a gamma of 2.2 ( hopefully) end result gamma 1 ( or thereabouts).

    As long as .... nothing weird is going on with the software of you playback package or the gamma of the card . Radeons actually have a hardware gamma on them ( preferable to a software correction as that can/will crush your image...posterisation anyone/ oops back we go to plasma screens!) However it can be ambiguous as to what it refers to ( gamma on the overlay) especially from package to package if they aren't sensible.

    It should refer to the actual output of the graphics card and it should be marked up correctly ie 1 is effectively a linear response ( ie no gamma) up to 3 and down to 0.4 : probably not much call for anything beyond that : none that I can think of anyway)


    So to cut a long story short .

    Assume your display is around 2.2 or will mimic 2.2.
    Plug in your dvd player and see what the display does .
    Plug in your PC ( don't mess with the overlay gamma unless you know the default isn't 1). See what your display looks like.*

    *If its a bit milky looking you have a display with a gamma greater than 2.2 ( I suspect the gamma of the ae100 is nearer 3 for example) Then think about tweaking the gamma on your card ( overlay) . Try dropping it ( I don't think the radeons go below 0.8).
    eg 0.45 ( material) x 0.8 (graphics card) x 3 ( display) =1.08 I reckon for my ae100.

    Final note. Gamma is not even worth looking at until you ensure your black point and white point ( brightness , contrast) are correct as they have a big role in how gamma will behave( never correct an image that looks too dark or too bright with gamma without first checking your contrast and brightness). If you follow what I've said you should see that gamma shouldn't affect the overall whitepoint and black point of your image ( remember that linear response curve with the middle point pulled up and down)
     
  4. Mr.D

    Mr.D
    Well-known Member

    Joined:
    Jul 14, 2000
    Messages:
    11,167
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,209
    This says to me that the video inputs on the Z2 get corrected to appear correct on the panel which has a gamma of around 3 ( most likely) but he PC input ( VGA?? DVI??) doesn't get corrected.
     
  5. Quatermass

    Quatermass
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    Does that mean that the picture from the 730, presumably through component, will be the more accurate of the two (vs. that from the PC)?
     
  6. KraGorn

    KraGorn
    Active Member

    Joined:
    Aug 30, 2003
    Messages:
    4,745
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    68
    Location:
    Warrington
    Ratings:
    +27
    I'm feeding the PC via DVI, the Z2 tells me it's seeing 720p. As I said, I 've calibrated Brightness and Contrast (blacks and whites) for both inputs and in fact the settings are nigh-on identical, so the difference seems to me is in the source material.

    The reason gamma occurred to me as the cause is because of the fact I've calibrated whites.

    To my eyes the PC's image is a LOT clearer and more interesting than the DVD, which is darker and less detailed. I haven't changed the Radeon driver's setting of 1 for gamma values, nor messed with TheaterTek's default of 171 (whatever THAT means?) ... I suppose I could try playing a DVD using PowerDVD which doesn't AFAIK have screen-adjustments like gamma.
     
  7. Mr.D

    Mr.D
    Well-known Member

    Joined:
    Jul 14, 2000
    Messages:
    11,167
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,209
    Try 7.5 for brightness
    100 for contrast
    82 for gamma ( try dropping it below this point if youo think your projector has a higher than 2.2 gamma)

    in your TT settings.

    These should ensure your card doesn't clip crush or apply additional gamma correction.

    Its been that long since I saw the TT default settings I no longer know if this is what they default to.
     
  8. KraGorn

    KraGorn
    Active Member

    Joined:
    Aug 30, 2003
    Messages:
    4,745
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    68
    Location:
    Warrington
    Ratings:
    +27
    Until recently I've not known how I should consider changing them. :)

    Last night I set Brightness to 7, it was a toss-up between 7 and 7.5, I think I took Contrast up to about 105 .. both using Avia .. it was when I came to gamma that I realised I didn't know what to do and caused my to post here .. I'll try the value you suggest and come back.

    Thanks for the info and suggestions. :)
     
  9. Quantum

    Quantum
    Standard Member

    Joined:
    Dec 14, 2003
    Messages:
    238
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    18
    Location:
    Smethwick Nr. B'ham,UK
    Ratings:
    +1
    I have been thinking about how to gamma correct the desktop on a projector. I have made the following 2 files:

    Gamma127

    This one assumed that a grey level of 127 should be half way between 0 and 255 so if you displayed this on a 1:1 pixel mapped projector then the white+black areas should be the same as the grey area in brightness once gamma corrected. By defocusing the lens the white+black squares become grey. I used right click then set as background.. then right click on desktop, select desktop and tile image. However this dosn't seem right at all...the central grey square is way too dark

    Then I thought it might be a square function thus 1/sqr2*256-1=180. so I made this:

    Gamma180

    But before I go changing the powerstrip gamma settings to equalise the greys can someone give me some idea if I am on the right lines.... Mr.D perhaps?

    Thanks
    Paul
     
  10. Mr.D

    Mr.D
    Well-known Member

    Joined:
    Jul 14, 2000
    Messages:
    11,167
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,209
    I'll have a shufty mate. I can tell you that the 127 version matches if you view it with a gamma of 1.5 on a CRT in a very accurate enviroment.
    I'm just doing a little bit of looking about to see where mid gray maps to on a video code value scale.
     
  11. Mr.D

    Mr.D
    Well-known Member

    Joined:
    Jul 14, 2000
    Messages:
    11,167
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,209
    Ok.

    I could make a case for 103 being the correct mapping for mid gray( based on a video transform function to get 18% grey from film to the mid gray point on 8bit video (16-235...rather than 0-255). However that doesn't explain the problem we see. I think it can only be down to the black and white dither display not being sufficient to disclose correctly on most monitors)

    I found this :

    http://www.photoscientia.co.uk/Gamma.htm

    the guy on here effectively says the same thing about the black and white single pixel dithered patterns not being sufficient.

    He also has a very good test pattern that seems to be right on the button looking at it on my calibrated monitor with a gamma of 1. ( the 2.2 pattern on his site is the one you want ...ignore all his other speil about 1.8 gamma being best as he's not talking about video)
     
  12. Quantum

    Quantum
    Standard Member

    Joined:
    Dec 14, 2003
    Messages:
    238
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    18
    Location:
    Smethwick Nr. B'ham,UK
    Ratings:
    +1
    I'll have to give it a try. How useful for calibrating hcpc do you think this is?
     
  13. Mr.D

    Mr.D
    Well-known Member

    Joined:
    Jul 14, 2000
    Messages:
    11,167
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,209
    Well its not going to do your overlay any good unless you find some way of displaying it so that the overlay controls will affect it ( to be honest gamma on the overlay should be left at 1 at least initially anyway for video)

    It will give you some idea of what sort of gamma your display device has . It does appear to be a very accurate pattern for discerning gamma.

    Ideally you are only after a few different bits of information for gamma. We know that the video material off dvd has its own gamma of around about .44 . This is designed to look correct when feeding a display device of 2.2 ( or that gets to 2.2 ).
    The overlay should have no additional gamma correction applied ( unless ...get to that later). So if the pattern still looks off then either there is gamma correction going on at the display or the display isn't around 2.2 ( same thing for all intents and purposes)

    The way I look at that pattern is to set an overall hardware gamma of 1 coming out of my machine and let the natural 2.2 response of my monitor correct it for correct display . This is on my machine at work which is calibrated to a very high standard and also has a custom hardware gamma control which affects the entire enviroment (desktop and overlay: its effectively adjusting the voltage at the graphics card itself rather than in software so no crushing). Now there is no reason my you can't set an off the shelf HTPC to display that pattern with a gamma of 1 ( most will probably do it at default settings). One thing I'm a little unsure of is whether the desktop gamma control or the overlay control will be the one to control it especialy if you display it from a web browser. I would suspect the desktop controls.

    However bear in mind all we really want to ascertain is the gamma of the display ( as we can safely get the HTPC outputting with a gamma of 1 or rather no correction). So if you know you are displaying the pattern correctly at the PC end of things if its off on display its because the display device likely doesn't have a gamma of 2.2 in which case you have two options.

    Correct the signal at the HTPC end by adjusting the gamma control.
    Correct the display to give an end to end gamma response of 2.2.

    The only complication here is if your display has a non-linear response ie a curve rather than a simple gamma in which case its very dependant on your display device as to whether or not this is completely correctable.
    You could also correct at the HTPC end if you have the ability to generate LUTs ( look up tables) that are effectively the inverse of the curve(s) on your display.

    Whilst this is straight forward in principle in practice its rare and expensive to find these sort of profiling systems on home cinema PCs ( maybe if demand increases who knows)

    Getting the display calibrated by a professional with the right equipment may allow you to totally correct the curve but it may also be an inherent limitation of the display itself.
     
  14. lillabas

    lillabas
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    Mr. D touched at a significant point: that video sources should, and often do, emit video RGB levels, i.e., values between 16 and 235, whilst PC's emit values between 0 and 255. :cool:

    It seems like Z2 assumes PC values (0-255) no matter what input is chosen. Is this correct (I have asked this elsewhere, but it is related to visible "gamma" problems people experience with Z2)?

    If so, is there any way to fix that, so that Z2 maps 16 to black and 235 to white, as it should according to video standards? :lease:

    Has anybody else suffered from this (note: HTPC users have most likely not experienced this problem)? Mr.D?

    As far as I can tell, this is a bug in the Z2 firmware. :lesson:
     

Share This Page

Loading...
  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice