Separate names with a comma.
Discussion in 'Exposure Discussion' started by michael_r, Feb 18, 2013.
ORIGINAL POST DELETED - EDITED THIS POST TO MAKE SENSE WITHOUT IT
Flare curve is expected to taper, in the first experiment the curve didn't do what would be expected so to avoid misleading the discussion is edited.
Exactly, Bill. This was a sort of "step 1" - ie see what kind of flare factor I can generate. Because it is not a single exposure with different subject brightness levels (or reflectance values), the curves should be parallel (as you and Stephen have shown me in the past ). I was trying to prove that here. This does not show the effect on an actual scene where flare affects the lower values more.
That would be "step 2" where multiple reflectance values are in one exposure.
How was the camera lens "masked?"
Could you give a frame of reference for "3 density points," in stops perhaps?
So to counteract the flare, would you reduce exposure to make the exposure match the first test subject?
Now I'm having second thoughts about posting this curve first. It might be confusing on its own since it does not show the effect of flare on an exposure of a full scale subject. This test only shows the effect of flare on a single tone. Showing the effect of flare on a full scale subject would illustrate that you cannot "counteract" subject flare, and reducing exposure does not help (in fact it can be detrimental). This was the point I was trying to make in the other thread but now I may have confused things for people.
Regarding lens masking, I made an attachment for the front of the lens that reduces the image circle to a rectangle the size of the 35mm frame. This prevents most of the flare that might be caused by extraneous light bouncing around inside the camera.
EDITED FOR CONTINUITY
The original (deleted) test shows the beginning of a valid flare test.
The first points to the right of the origin show a possible "flare factor". The displacement on the x-axis... it's how much light (if you can deduce the mcs) the flare added. Taper would start there and within a few steps to the right it would re-join the Non-Flare curve.
A good test to determine flare could be to shoot a variety of gray patches that meter says differ by one-stop each, surrounded by a white background.
Yes exactly. I should not have shown a "curve" for flare in this way because it is misleading. The different exposures just add more data points to the sample, but it doesn't add any value to show a curve of multiple exposures of a uniform target for this purpose, and it is confusing.
I prepared a set of grey cards that measure one stop (metered) apart for the second part of the test (single exposure of grey scale) but should have posted that part first so that one could see the displaced curve that then merges with the "non-flare" curve.
Sorry about this, people. Wish there was a way to delete the thread at this point. The last thing we need is a thread that confuses people.
EDITED FOR CONTINUITY
The original post showed good scientific method, but Michael decided to delete it for clarity.
Still, I don't like that this can confuse people into thinking the effect is a parallel shift of the curve.
I'm going to delete the original post for now.
Michael, can you post the testing procedure?
Stephen - it was a bad thread. Or at least bad to start it the way I did because it would confuse people. It was the first step in a broader analysis but on its own it was not good. The idea was to take a typical ZS-type test (ie in-camera instead of wedge/tablet, shooting a target tone), refine it somewhat to remove as many variables as possible and as much camera flare as possible, and try to create/isolate a subject flare factor. Part of the purpose there was to "prove" that with a uniform target the added flare density is constant regardless of exposure, which it did. But the problem is it quickly became obvious presenting this data in graphical format (a characteristic curve) is misleading because on its own it only shows flare exists (ie a parallel shift in the curve). It does not show the effect of flare when making a single exposure of a full scale subject. As the experimenter I understood this, but it was a confusing illustration. Showing parallel curves can give the impression the effect of flare is eliminated by simply reducing exposure - the very thing I was arguing against. So I figured it was better to get rid of it. The last thing we need on APUG is more confusion.
But at least if nothing else it proved to me, with my own data, that flare density is constant in a typical ZS test, whatever the flare density is (if any), which was part of the purpose of this first step. The other interesting thing is it showed it is actually quite difficult to create a lot of flare in a test shooting a reflective target. With my limited indoor setup I was only able to get a 5 stop metered difference between white and black, and the amount of flare density was a little less than I had expected. Actually I didn't really know what to expect, but I got only 3 density points of subject flare density - although at the bottom of a H&D curve this could indeed be material.
Flare discussion seems to pop up quite often
I have read a lot of Stephen's writings and perhaps understand something. Perhaps not all.
Would the simpliest way to measure flare or get it's effect in the charts just expose two wedges. One without any optics (as contact print under the enlarger or using sensitometer) and another one by photographing wedge with normal coated lens.
Then just compare results?
If same exposure is done with couple of different lenses, it is easy to averaging flare.
Well, I was thinking of shooting a flare test outdoors today but it's raining and I don't want to get the step tablets wet...
I was looking forward to discussing subtle differences that different tests might reveal, for example...
LP Clerc describes the difference of flare indoors versus outdoors. Indoors your subject lighted by photofloods is well lit but the surroundings are dark = less flare. Outdoors your subject is lit by sun, cloud or open sky... but the background may be bright open sky = more flare.
With a lens shade and controlled lighting indoors, the flare would be "different" than it would be without a lens shade. Outdoors open sky will send many rays into the lens which aren't aimed directly at the film plane. Some of these rays will be scattered by the optics and camera internal reflections. Flare would illuminate the entire camera chamber.
It wouldn't invalidate the test done indoors. Just that different scenarios would give different results, which is exactly the problem Flare poses.
Very true. In my indoor test I tried to maximize flare as much as I could by filling the frame with a white background, throwing as much light as possible at it and making the black "target" in the center as small as I could without making it difficult to read the density of it in the resulting negatives. As noted of course, using a reflection target in this way obviously limits the maximum difference in luminance between bright and dark. I better way indoors would be to have a brightly backlit target filling the frame, with an opaque dark patch in the middle. This would create more flare. It might better approximate the situations I'm often dealing with in my photography - predominantly dark subjects and/or night shots with lights in-frame.
I have a question about your methodology. How would you fill the frame with your step tablet? This is a problem I struggle with in designing these sorts of tests.
I would only work to make the step tablet patches about twice the aperture of my densitometer on the final film.
Because I am encouraging flare, I wouldn't worry about maximizing the frame. Also, I am satisfied with only a subset of steps for a flare test. I was going to be doing a "typical scene" kind of test. (But now it's hailing).
Backlighting a transparent test pattern is one good way to design a test that includes flare and also controls the lighting.
I initially wanted to do the test I outlined (and deleted) with a step tablet on my white and black frame-filling backgrounds - which is part 2 of my test. But the problem I keep running into is how to use my Stouffer reflection target and still get something that reproduces on the negative large enough to measure accurately with the densitometer. If I move the camera close enough to get that, I have to focus significantly closer than infinity to avoid blurring the tablet too much. But then lens extension becomes a variable. I wish I had some kind of bigger step tablet.
Hey Michael R 1974!
It got sunny just long enough late afternoon for me to take some high flare and low flare shots. I added some sensitometry to the rolls for a baseline curve.
Methodology: Sekonic test target with a black patch, white patch and 7 gray patches. The gray patches are centered on 18% (yes really). Total range is 1.4 log units, the gray patches are closely spaced. 100 TMAX 35mm in an ESII with 50 f/1.4 SMCT at f/5.6 for 1/125 sun facing camera (high flare), then another shot with sun facing target at f/16 for 1/250 (low flare).
Results: Low flare so low it is not measurable. High flare is 0.26. Classic taper of flare curve to nominal curve.
Another test using M2 with 50mm Summicron but with a UV filter that has significant cleaning marks... Difficult to measure because image was smaller on negative... Some abrasion marks interfered with high flare shot readings... Exposure error forced me to do some interpolation of my readings... Results of that test: Low flare 0.28 High flare 0.74. That test is not worth showing, but I think it helps validate the methodology.
Bill, what's the reflection density on the darkest patch?
Patches are white = 90.7%, grays = 25.5%, 22.7%, 20.2%, 18%, 16%, 14.3%, 12.7% and black = 3.6%
The low flare marks are all just marked on the curve at the densities they measure. They don't magically coincide with the curve. I expected LogE displacement from nominal, and there is some, but I was surprised how closely the marks fall in line. The reason I say I can't measure flare on the low flare exposure, is that it spans more than the nominal 1.4 on the LogE axis (gives me about -.02 flare which is illogical).
I THINK the flare may be hiding. It is possible that in reality I get a higher real CI at the section of film with test targets. With measured CI 0.63 coming from a section of film near the end. If the local CI where the test target is shot is more than the measured CI 0.63, then the actual LogE span of the low flare target test COULD be less than 1.4 LogE. And THAT would be my real low flare measure. There's no missing the fact that the ESII has low flare, the density of 0.07 is not what you would get with my usually accepted flare factor of 0.4
The high flare marks aligned at the high density so their Nominal LogE would be directly above each of the low flare marks. So I merely marked their actual densities.
Bill, one of the problems with evaluating flare is the determination of the shadow exposure which is a factor in determining the flare factor. In general flare for a highlight of 100% reflectance is about 1 to 3%. For a flare factor of 2 (1 stop), where will the value of flare equal the camera exposure without flare? Your target's black isn't dark enough. Here is the breakdown of the statistically average scene for a 125 speed film exposure.
Shadow exposure 4 1/3 stops below metered exposure point is 0.6% reflectance. Your 3.6% is over two stops lighter. In this example, flare equals 0.0034 mcs. Added to the shadow exposure of 0.0034 and you've double the shadow exposure, but add it to the exposure of 0.0169 mcs at 3.2% reflectance and you get 0.0203 which is only a 20% increase.
How can I explain why my test gives me a satisfying shadow?
I hit 0.07 density 2 1/3 stop below 18%. That's a pretty satisfactory low density.
That corresponds to about 0.0079 MCS, this leaves room for a possible 0.0034 MCS flare.
I must assume that the difference between what you would expect from a 3.6% reflectance comes from effective underexposure due to a variety of possible causes; error in metering, incident meter not calibrated to 18% or maybe mechanical shutter speed problem. But it sure looks like I underexposed the test.
Which effectively gave me a good shadow placement.
I don't know. What I do know is that 0.0080 is 0.8/100 at speed point for a 100 speed film. You rated the film at 50. You've also hit this value 2 1/3 stops below meter exposure which is one stop higher than the 3 1/3 stops it should be and the shadow exposure is still one stop further below that.
But you shot the sun facing the target at f/16 at 1/250. Shouldn't that have been at 1/50 sec? So let's say the TMX has a speed of 100. That's a potential 1 1/2 stop underexposure. That's now getting closer to the shadow exposure value except that flare won't have the same effect as if the actual reflectance is lower.
Film tested as 80. Fell short of ASA parameters.
I have the meter set at EI 100.
Sun was late afternoon, fairly low in the sky following a storm (clear skies with minimal dust scattering).
But I just checked and the meter still reads same f/16 at 1/250. BUT, it's in spotmeter mode.
So I must have spotmetered the 18% background of the test chart and not made compensation!
If 3.6% isn't black enough for a flare test, I guess I will have to get the top-hat out for the next test...
These are the difficulties I run into. Designing a good test ain't easy.
Definitely a topic I've been focusing on lately. Good advice to keep in mind with speed and development tests too.