Hard data on 5D focus screen brightness

Discussion in 'Canon EOS' started by fourfa, Jun 7, 2006.

  1. A little back story first. I bought a 5D, got the Ee-S screen for better manual
    focus with fast primes. Noticed that the viewfinder was considerably darker.
    Wondered if the Bill Maxwell Hi-Lux optical treatment that people rave about in
    other fora would do anything. So I bought a second Ee-S, send it to Bill, and
    got it treated. Got the treated screen today and thought I would run a
    comparison with the three screens.
    <p>
    5D was mounted on a tripod, pointed at a featureless, evenly-lit whiteboard.
    Nikon 85mm f1.4 AFD lens mounted (for ease of changing f-stop without having to
    hold the DOF preview, not enough hands), focused at infinity. Nikon Coolpix
    4500 was held lens flat against the 5D viewfinder, manual exposure (1/8" f2.6),
    fixed focal length, fixed 100 ISO, fixed white balance, minimal adjustments,
    macro focus, fine JPG, no NR. The 5D focus brackets were centered in the 4500
    finder for each shot. In photoshop, each shot was desaturated and three color
    picker points were selected (inside the center bracket, and just next to the
    left- and rightmost brackets), as close as possible to the same in each photo.
    The greyscale values measured from those points were averaged and plotted below
    versus the f-stops tested (f1.4, 2.0, 2.8, 4.0, 5.6, 8, 11, 16).
    <p>
    <img src="http://andyradin.com/focus%20screen%20comparison.jpg">
    <p>
    Looks like the stock Ee-A screen is about 2-2.5 stops brighter than the Ee-S,
    and the Maxwell Hi-Lux treatment is worth about 2/3-1 stop.
    <p>
    Comments? Particularly on formatting the chart to make the comparison more clear?
     
  2. P.S. Bill Maxwell's contact info may be found here and the treatment cost $83 shipped, in addition to the screen itself (which I had drop-shipped from B&H, $38.65 shipped). I didn't test the Brightscreen product. I believe KatzEye also offers a brightening treatment of some kind. I haven't had a chance to shoot outdoors with the new screen yet so I can't offer any opinions about its overall utility yet.
     
  3. After the treatment, is the Es-S still easier to MF than the stock Ee-A screen? With other
    screens I noticed as the screen brightens, some contrast is lost.
     
  4. Andy, that's a splendid piece of experimenting. It would have been interesting to me in days gone by to have the same comparison between the Ec-CIII and Ec-N screens. When I had an EOS-3 I actually replaced the standard EC-N screen with the slightly less bright Ec-CIII standard for the -1V, and found it much better for MF in close-up work without any problem of not being bright enough. Later I had a -1V, and was very happy with the Ec-CIII on that as well.

    I think it might be worth giving a bit of thought to the analysis and interpretation of the results you have obtained. First, it might be better to plot the f-stops at uniform spacing on the horizontal axis - in different language, change to a log scale. Second question is how best to present the vertical scale. I think there are two steps needed to get the best value out of your data. The first is to calibrate your test camera, the Coolpix, since it is not a given that the relationship between the light level falling on the sensor and the grayscale level is linear. This could be done by pointing the Coolpix at a uniformly illuminated grey card, putting it in manual mode, and looking at the relationship between exposure time and grayscale output level at a constant aperture. That would give you the functional relationship you need, and you could then apply this to convert the current grayscale vertical axis to a scale of illumination level, which is what you want. Then, because of the way perception works, you probably want to take logs. On the rescaled plot, your curves will probably be much closer to straight lines, which may make interpretation easier.

    Next, how should the plot be interpreted. I have the impression, and correct me if I am wrong, that this has been done by looking at where the curves cut a horizontal line. I think the question that usually arises is how the screens compare for brightness at constant aperture, not what apertures give constant brightness, and that means looking at intercepts on a vertical line. That tells a somewhat less dramatic story. Ignoring the possible illumination-to-grayscale nonlinearity for the moment (because we don't yet know the correct transformation), but remembering that the vertical scale needs to be interpreted in terms of ratios, the answer would seem to me to be as follows. There is no significant difference between the screens at f/1.4 - if this does not tally with your assessment of perceived brightness, there may be a need for further investigation. At f/4, however, the stock Ee-S is giving a reading of just under 100 whereas the Ee-A is just under 200, so there is about a 1-stop difference. By f/8 this has become worse than 2 stops, and beyond that the stock Ee-S is dead in the water. The Maxwell treatment is worth less than half a stop until f/8, where it is worth about 1 stop. The conclusion is that if you really must use the Ee-S screen with lenses slower than f/4, the Maxwell treatment may be worthwhile. These conclusions could easily be affected quite a bit by transforming from grayscale to illumination.

    I hope these thoughts are helpful - they are offered in a constructive spirit! - and I shall be interested to have your comments.
     
  5. It seems very counter intuitive, my understandign was that the Ee-S made manual fcusing easier because it is brighter than the Ee-A - I guess it's down to contrast instead?

    I'm seriously wanting another screen for my 5D since I have invested in some superb MF lenses, what route do people suggest following?
     
  6. Andy -- you beat me to it. I had thought of doing the same thing several months back. Since then, I've just gotten used to the EE-s screen, and am loving it. (I only own prime lenses.) I may end up doing what you did, but have you checked yet how Bill Maxwell's treatment affects camera metering? I was concerned it would throw the evaluative metering completely off. Thanks, Howard
     
  7. The Ee-S screen makes it easier to focus by reducing the depth of field in the focusing screen. It has nothing to do with brightness, though the effect is to make the screen darker as you stop down. For example, I see no difference in screen brightness between the two screens at f:1.4. But at f:2.8 there's a very strong difference, the Ee-S screen is noticeably darker.
     
  8. yep, my reason for buying was that the visible depth of field with the stock Ee-A does not change between f1.4 and f2.8, which I found problematic for manual focus with fast lenses. I found myself saying "what the hell? that sure *looked* in focus in the finder" a lot, and the resulting photo would have much less depth of field than visible in the finder below f2.8.

    With the Ee-S screen, the visible depth of field changes between 1.4 and 2 and 2.8, and the what's visible in the screen matches the result.

    I can't yet say anything definitive about the contrast or metering. Thanks for the advice on the chart, I'll work on that today.
     
  9. You're right, looking at the raw luminance makes no sense. Below is the same graph represented as loss in brightness versus the stock Ee-A in f-stops (log base 2 of the luminance ratio versus Ee-A)
    <p>
    <img src="http://d6d2h4gfvy8t8.cloudfront.net/4541267-md.jpg">
     
  10. FYI, with the custom function C.Fn-00 set to "2" for the Ee-S screen, I'm getting pretty consistent 2/3 stop underexposure with the Maxwell screen. Set to "0" for Ee-A, it's been close to right on with my lenses (fast primes, f2.8 and f4).
     
  11. Good job Andy. I can add some practical experience from the field to your hard data...
    I replaced the stock screen with the Ee-S and I've had no problems using it with lenses as slow as f/5.6, even with extension tubes (giving a true wide-open value of f/8). An eyecup would be useful to reduce glare at f/8 in bright sunlight though.
    IMO the 5D should have shipped with the Ee-S screen. Of course a brighter viewfinder is nice to look through, but it's rather pointless if you can't tell whether or not your image is in focus!
    Karl Lehmann
    Lost World Arts
     
  12. Update
    <p>
    Turns out there was good reason to be concerned about the tone curve in the P&S camera I used in the initial round of this test. I repeated the test using a bright, even light table to illuminate the lens. Light levels were measured using the built-in light meter of an EOS3 with a 100mm macro lens. Both test and measurement camera were mounted on tripods. The gap between 5D viewfinder and measurement lens was sealed with black gaffer's tape. Screens were tested without moving either rig, and the same point on the focus screen was measured with the spot meter. Data presented are averages of three full tests running through each aperture.
    <p>
    <img src="http://andyradin.com/tests/Ee-S%20screens2.JPG">
    <p>
    The results are somewhat different in magnitude, though not overall conclusion.
     

Share This Page