Calibration issues/sytems input required

Discussion in 'Home Cinema Buying & Building' started by Mr.D, Nov 24, 2001.

  1. Mr.D

    Mr.D
    Well-known Member

    Joined:
    Jul 14, 2000
    Messages:
    11,190
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,231
    Ok been looking into calibration issues related to using PC based systems for video playback.

    What I'd like to do is initially do a coarse calibration of my display device ( be it projector or monitor etc). This would correctly set black point white point whitebalance and simple display gamma. Basically get the device healthy on its own terms

    Then what I'd like to do is run a closed loop calibration through the PCs graphics card using a sensor attached to the PC that mapsthe specific response characteristics of the display device across the intensity scale ( ie gave me a picture of any non-linearity and an accurate mearsurement of the gamma).

    'd like the PC to then generate a look up table ( LUT) or even multiple LUTs that when loaded on starting up things like a software dvd player ( or anything really that allowed it : editing package/ compositing package) would effectively colour correct in software for non-linearity on the monitor and give you a very accurate overall system gamma/ intensity distribution : probably like it definable so I could set any gamma I liked ie 1 ( no additional correction beyond the neutralisation of the LUT) , 1.5 , 1.7. 2.2.

    So I need PC based sensor system to handle the closed loop. Software that will work with the sensor to create a map of the display's behaviour ( and probably generate relevant patches). Then another bit that writes the inverse LUT. Then another bit that reads the LUT and allows you to add further correction on top of this baseline. ( some of these parts may be part and parcel of the same bit of software).

    I've had a quick look at powerstrip which doesn't offer overlay correction on cards without hardware gamma ( ie Nvidia cards) and only offers a dumb overall linear gamma correction anyway and won't read an LUT.

    And hopefully this won't cost a packet! Any ideas anyone ?

    Just had a look at a calibration system called Dilard ( or caling LCOS projectors)and whilst it does use a sensor to effectively automate calibration I don't think its mapping the intensity response of the device
    .Its going in the right direction though...
     
  2. Gordon @ Convergent AV

    Gordon @ Convergent AV
    Distinguished Member AVForums Sponsor

    Joined:
    Jul 13, 2000
    Messages:
    14,214
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    166
    Location:
    Living in Surrey, covering UK!
    Ratings:
    +3,058
    Keith,

    You've probably gone way over the heads of most of us here....
    I thought that Dilard and Colourfacts did what you are asking.It's just that it only creates look up tables for a limited amount of projectors they have worked with. Noticeably D'Ila's and the NEC150 thingy.

    If this isn't what it does can you explain to me what it does do?


    Gordon
     
  3. Mr.D

    Mr.D
    Well-known Member

    Joined:
    Jul 14, 2000
    Messages:
    11,190
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,231
    I think the end result is effectively the same but not as flexible :if you have an LUT that effectively corrects for any non-linearity on the display internal to the image chain ie HTPC, you can apply other corrections ( LUTs) on top of your nice flat linear response to give you more than a simple single exponent gamma ie custom response curves. ( you can funk with the curves in software whilst the hardware just does its usual thing)

    ( within the ultimate tolerances of the display hardware thats why you need to get it performing optimum first)

    theory is straight forward ( can't really talk about what we do at work but effectively we have very very accurate custom LUTs that we apply after we've acheived an accurate overall system gamma/aim)

    The idea being you make the system behave itself and adhere to known overall display criteria then you can pretty much select whatever system gamma you want and then apply LUTs for looking at film material , video material , print material etc.

    The thing about Dilard is that whilst its a referential cal its about making the end display adhere to certain criteria not about getting the entire image path into a known state and then accurately colour correcting the input signal to look correct with reference to the end display.

    Basically you create a smart image path that knows pretty accurately what the monitor at the end of the chain will do to images you send it. Every so often you run the cal again and remap the display and adjust the LUT to match. Eventually the peak output of the diplay will weaken and the colour and gamma will start to freak out as well ( based on my experience of CRTs not too sure what happens with an LCOS or LCD but assume the bulb has a varying behaviour over time). But because of the up to date regenerated LUT your colour and gamma will still appear perceptually accurate only your white point may drop a bit over time and your overall dynamic range will get lower until you reach the point where you will get percievable banding because of the limited intensity scale. ( then you replace the bulb!)

    I'll have nother look at the dilard site but I think its based around correcting everything at the projector itself. ( ie sticking an LUT into a gamma preset)

    I can see exactly how you'd do this but the tricky part is getting the software talking the same language: probably all doable to someone who understands directdraw scripting or something??
     
  4. JohnAd

    JohnAd
    Member

    Joined:
    Mar 14, 2001
    Messages:
    362
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    18
    Location:
    London
    Ratings:
    +4
    Keith

    I think I understand what you want. On the playback side it is sort of possible. The API for DirectShow overlays has space to allow a custom colour by colour LUT to be assigned but as far as I know no drivers implement this. You can do all this with normal PC graphics but not with the overlays yet.

    Dscaler has the option to load up custom LUTs but for luma only at the moment.

    On the detection side I think the problem is getting a linear device to measure the output (or a clenaly non-linear one) Most previous attempts that I've seen to avoid getting a expensive meter seem to hit against this.

    John
     
  5. Gordon @ Convergent AV

    Gordon @ Convergent AV
    Distinguished Member AVForums Sponsor

    Joined:
    Jul 13, 2000
    Messages:
    14,214
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    166
    Location:
    Living in Surrey, covering UK!
    Ratings:
    +3,058
    I believe the Terranex processor is capable of doing the sort of thing you are after. I would guess they run their own custom software.

    I know that you can basically tell it the response of your display device, clour co-ordinates, black, white point gamma etc and then ask it to output using SMPTE, EBU, NTSC, PAL, whatever really. Is this what you are trying to do?

    If it is then I suggest you go to AVScience forum, do a search of the Special Guest forum for Chris Stephens (stevens). He talks a bit about this I think

    Gordon
     
  6. Mr.D

    Mr.D
    Well-known Member

    Joined:
    Jul 14, 2000
    Messages:
    11,190
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,231
    Hmm I know there are some cheapish PC based meters out there : under 500quid. Seem to remember one called a spider or something. You are right though you are effectively limited by how accurate the initial mapping of the display is ( I'd have thought measuring a selection of numerically generated patches would have got round the problems of linear measurement of a non-linear device cause your effectively measuring the deviation from a known curve rather than the real curve itself: gets you the same result though)

    Even if you only made samples based around say 20 patches from 0-100IRE say ( 0-255 code values for 8bits per channel: have to query the actual values of known video for this which I think is 16-235 code values not sure this matters actually) The map you have would still increase the overall accuracy of the system beyond standard hardware calibration and any innaccuracy would be limited to the space between any two 5 IRE increments so itself woud be small.

    Like you say though I wonder if the cheaper sensors can actually perform to that level of accuracy??
     
  7. Mr.D

    Mr.D
    Well-known Member

    Joined:
    Jul 14, 2000
    Messages:
    11,190
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,231
    Kind of but what I'm really getting at is the initial stage where you measure and map the intensity response of the display device and generate an LUT that effectively linearises the whole perceptual display ( gets rid or rather counteracts any bumps kinks or non-linearity in the hardware response curve of the display)

    Once you've got the system into this state when you have a known accurate overall system gamma you can then apply additional LUTs which are just colour corrections in software to get you into any colourspace you desire . (within the limitations of the hardware tolerances obviously)

    The actual application of the LUTs is probably just a question of formating and the capabilities of individual bits of software. As john mentions whether you have software/OS which allows RGB LUTs as opposed to just overall intensity and the biggie whether anything at all will allow you to apply LUTs to the 2d overlay on any PC based systems. ( I know SGIs do and I know Macs do but PCs seem to be a but of a 3 legged horse).
     
  8. JohnAd

    JohnAd
    Member

    Joined:
    Mar 14, 2001
    Messages:
    362
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    18
    Location:
    London
    Ratings:
    +4
  9. Alaric

    Alaric
    Active Member

    Joined:
    Mar 12, 2001
    Messages:
    1,804
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    66
    Location:
    Peterborough, Cambs, UK
    Ratings:
    +174
    Hi there,

    Does anyone know anyone whoes tried this ????

    I'm right in thinking the same sort of principles will work on any projector (eg my Barco Data 800) ???

    Cya,
    Lee (off to do some reading and maybe some making eventualy)
     

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