Aardenburg......argh!

Discussion in 'Wet and Dry Hybrid prints' started by Kirk Gittings, Feb 25, 2011.

  1. Kirk Gittings

    Kirk Gittings Member

    Messages:
    84
    Joined:
    Mar 21, 2006
    Location:
    New Mexico
    Shooter:
    4x5 Format
    Can some one explain to me in simple terms the key stats in an Aadenburg longevity test? I've read through the whole site and just can't grasp it.

    Light Fade Test Results
     
  2. gmikol

    gmikol Member

    Messages:
    618
    Joined:
    Mar 2, 2009
    Location:
    Vancouver, W
    Shooter:
    35mm
    In the "Documents" section, have you read the following 2 papers:


    An Overview of the AaI&A Conservation Display Ratings

    and

    An Introduction to the I* Metric


    Basically, as a result of research collaboration with Wilhelm, he developed the I* metric for characterizing color image accuracy. Instead of simply measuring color differences (a la "delta E"), I* takes into account color fidelity, tonal fidelity, and the ability (of the print, in this case) to still convey information.

    An I* of 100 is perfect fidelity, i.e. the sample is identical to the control. I*=0 means all color fidelity has been lost, and negative I* values indicate the appearance of false colors.

    The image samples are exposed to light (from fluorescent bulbs, to provide both visible light and UV) for extended periods of time, and the color shifts are measured at regular intervals, with I* scores computed at each interval.

    The conservation display ratings have an upper limit and a lower limit. The upper (longer, in principle) limit is reached when the I* average of all colors falls below 90. The lower (shorter, in principle) limit is reached when the worst 10% of the colors has an average I* below 80. I say in principle, because there may exist some system out there which fades very uniformly, in which the I* overall average falls below 90 without any I* falling below 80. This would reverse the upper and lower limits, but would still define an acceptable range.

    The time limits are expressed in megalux-hours. 1 megalux hour = 1,000,000 lux illumination for 1 hour or 1 lux illumination for 1,000,000 hours, or any combination in between. In each report, there is a table which relates "display years" to megalux hours based on different illumination levels.

    I don't know if MHMG participates on this board, but he (or anyone else) can feel free to correct any factual errors or omissions I may have made.

    Hope this helps....

    --Greg
     
  3. MHMG

    MHMG Member

    Messages:
    4
    Joined:
    Apr 10, 2010
    Shooter:
    Medium Format
    I think Greg did a fine job explaining many of the nuances of the AaI&A testing methodology. I would just add a couple of points.

    1). AaI&A conservation display ratings (CDR) differ significantly from other testing labs. Simply put, the AaI&A CDR scores quantify the fading performance of the system at the early stages of deterioration where the viewer would observe "little or no noticeable fade" and thus the print would still appear in excellent condition. Other industry-sponsored fade tests generally use a more consumer-oriented criterion for allowable fade, subjectively described as "easily noticeable fade". Both rating schemes have merit but the AaI&A ratings are intended for a more discriminating end-user (i.e., serious collector, museum curator, etc.) whereas consumers generally will tolerate considerable fading in prints before becoming concerned. Indeed, to know the whole story for a given printing process, one needs to evaluate the whole fading curve from beginning to middle to late stage deterioration where only the faintest of images may remain. AaI&A tries to provide the full curve response by continuing the testing well beyond the exposure dose where the conservation display rating has been reached.

    2). The AaI&A reports provide a lot of numerical data for more technically savvy AaI&A members who understand LAB color space, delta E, and as such will probably be interested in the mathematics of I* as well. That said, the reports also use the colorimetric measurements to post the actual target colors as they fade with high reproduction accuracy in the test reports. On a reasonably calibrated montor, the target images in the reports enable the reader to visualize the nature and magnitude of the fading as the testing progresses....color fade testing for dummies:smile: Just kidding. Actually helpful for novices and experts alike.

    3). 24 of the 30 colors in the AaI&A color target are the same ones used in the widely known Macbeth Colorchecker chart. I chose this test pattern because so many photographers know this color pattern well, and it does a great job exercising the various printer driver blend points of printers with multi-colorant ink sets.

    Greg, I would also add that in the situation you keenly noted where the upper limit of the CDR is reached before the lower limit, AaI&A then assigns the Upper value to both limits on the range (e.g. 37-37 megalux hours). This is a fairly rare occurrence but it does happen on occasion. What it means is that the fade rates of all the colors are fairly close (i.e., the statistic distribution of color fade is very tight) so the weakest ones are not much worse than the best ones. Because the onset of noticeable fading is dependent not just on the weakest colors in the system but on the image content as well (i.e. whether those weak colors are even used in the printed image), a fair and balanced rating must report not only the weakest link in the system but also the overall average performance of the system. That's the rationale behind the AaI&A Conservation Display rating being expressed as a range not just as a single limit factor. Two different systems could share the same limiting factor (for example a weakness in the media white point stability or a shared weak colorant such as yellow, magenta, etc) but behave quite differently on average due to one having a better overall set of image forming colorants compared to the other.

    I hope these additional remarks are helpful
    kind regards,
    Mark
    Home (index) page
     
  4. MHMG

    MHMG Member

    Messages:
    4
    Joined:
    Apr 10, 2010
    Shooter:
    Medium Format
    Oh, and one other thing

    For Kirk: a relatively nice "rule of thumb" to convert megalux hours of exposure to Wilhelm "display years" or Kodak "display years" is as follows.

    To compute WIR display years divide Megalux hours by 2. To compute Kodak "display years" multiply by 2. Want to split the difference in the assumed average light level between Kodak and WIR? Then it's really easy. Megalux hours = years on display.

    WIR assumes an average light level in the print display location of 450 lux for 12 hours per day whereas Kodak assumes 120 lux for 12 hours per day. At an approximately even compromise between the two companies' assumptions for print illumination levels in the real world, we could choose an assumed light level of 228 lux for 12 hours per day, and in that special case, megalux hours of exposure extrapolates directly to the number of years on display needed to reach the rated exposure dose.

    Example: AaI&A tested five samples of Fuji Crystal Archive II paper and rated this traditional color print process at approximately 15-30 megalux hours. If you display according to WIR assumed light levels, the AaI&A rating says the print will remain in excellent condition for approximately 7.5-15 years. If you used Kodak assumed light levels in the display area, the same print would remain free of noticeable light fading issues for 30-60 years. Note that WIR rates Fuji Crystal Archive at approximately 40 years but uses a more liberal allowance for fade. If you pull up one of the AaI&A test reports and look at the 80 megalux hour exposure results (40 WIR years), you will get an idea of what the more liberal "easily noticeable fade" criteria used by industry allows for in testing.

    A little exercise anyone can do: take a lux meter (can be bought for as little as $30 on Amazon.com) around to various levels in your home where you have prints on display. You may be surprised at the spot readings you measure and then average over multiple measurements during the day. 10 lux will be common, 100 lux, quite common, and 1000+ lux not out of the question in bright areas like near picture windows, sun porches, etc. Real world lighting conditions don't obey the law of averages very well. There is even a seasonal component. This is why AaI&A makes no "standardized assumption" about your real world light levels on display and reports accumulated exposure dose instead, leaving it to the end-user to estimate average light levels where the print will be displayed over long periods of time.

    cheers,
    Mark
    Home (index) page
     
  5. OP
    OP
    Kirk Gittings

    Kirk Gittings Member

    Messages:
    84
    Joined:
    Mar 21, 2006
    Location:
    New Mexico
    Shooter:
    4x5 Format
    Thanks for the excellent replies guys. Mark, FWIW though I have trouble understanding your testing process, I respect what you are about and have been a subscriber for the last couple of years.

    My main interest is b&w, in particular, comparing various Piezography ink sets to each other and b&w results from Canon and Epson color ink sets. I guess I need to sit down and really get into your numbers at some point.
     
  6. gmikol

    gmikol Member

    Messages:
    618
    Joined:
    Mar 2, 2009
    Location:
    Vancouver, W
    Shooter:
    35mm
    Thanks for chiming in here, Mark...it looks like you've been a member for a while, but these are your first posts. Welcome!

    One follow-up question. Are the I* scores (and the CDR ratings derived from them) computed based only on the 30 patches shown in the report, or is it for all of the (800?) patches on the user-submitted test target?

    --Greg
     
  7. MHMG

    MHMG Member

    Messages:
    4
    Joined:
    Apr 10, 2010
    Shooter:
    Medium Format
    The lightfastness tests use the 30 patch target and I* is calculated using these thirty colors plus two virtual rows and columns of inserted media white color values surrounding the 30 patch array (needed to compute I* tone because it looks at near neighbors). The bigger 800+ target on the color sheet that members print for me is used to ensure that the whole printer/ink/paper system is meeting minimum quality standards. If the overall image reproduction is not satisfactory, i.e., shows evidence of nozzle problems, banding, or poor overall color and tone reproduction not representative of reasonable system performance, I reject the sample from testing. It happens very rarely because my members generally submit very high quality samples, but when it does happen, I usually work with the submitting member to get the situation corrected and the sample resubmitted. It should probably come as no surprise that the biggest challenges tend to come from members using 3rd party inks for which no custom profile has been made:sad:

    For the monochrome target, the 30 patch gradient represents L* value incrementing in 3-4 L* increments which is a fairly demanding test on I* tone and more than adequate to see reproduction issues when they occur. Interestingly, the members submitting B&W samples are typically right on top of their game and print quality is usually outstanding:smile:

    Greg, thanks again for your nice explanation of the AaI&A testing.

    kind regards,
    Mark
    Home (index) page
     
  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. If you have a Photrio account, please log in (and select 'stay logged in') to prevent recurrence of this notice.