Discussion in 'Lighting and Hardware' started by smoke665, Dec 13, 2019.
The best way to determine exposure accuracy is shoot a control target that has a full scale of known densities such as a colour checker, grey card with reference data or my preference Sekonic Exposure Profile Target II. That way you can check the RGB values in post to known values on the target. Back in the film days this was a common process followed up by diligent processing and final densitometry readings for both negative and reversal emulsions.
For digital it is a wee bit different as in shooting in a particular profile can influence readings but a known single grey value should give you very close exposure guidance. The caveats is, it should not be a WB card since it’s only purpose is to set WB rather than exposure, so use an exposure target for best results.
Also, if you flash exposures are suspect, likely the incident ambient readings are too. A sunny day, exposure target and filling the frame with your target with the lens at infinity should confirm both camera and handheld meter to the sunny 16 rule.
Sorry I may be saying it wrong. In this example my incident meter is reading f/8 with the flash, but if I set my camera to f/8 the image is underexposed. Its an older Seconics meter that I've had for awhile, I've always had to open up about a third stop from what the meter reads, but of late it seems to be getting worse. I'm having to correct the exposure more and more in post.
As to any curve adjustment applied to the JPEG I'm not seeing it reflected when viewing in Lr?
As I rarely, if ever, use the meter for reflective readings, relying instead on the camera metering, you could be right. I've been looking at exposure targets for some time now, but kept postponing the purchase. Might be time. I have an Expo Disc that was part of a box of stuff I bought, but never used. Supposedly it can be used to dial in an 18% gray exposure by centering the spike in the histogram of a test image.
Another thought I've had is how much could different lenses have in the equation?
Redid the test this morning with fresh battery in the meter. Then did the following side to side composite with no adjustments other then cropping to fit. Flash was adjusted to read f8 on the incident meter. From left to right - Camera set per meter f8 ISO 100, 1/3 stop adjustment to f7.1, 2/3 stop adjustment to f6.3, and finally a full stop adjustment setting the aperture back to f8 and raising the ISO to 200.
Not shown above, but when I let Lr do an Auto adjust, it bumped the exposure on the original .86, on the second it raised it .52, on the third .20, and surprisingly on the last .15.
I need to repeat the test with all the studio lights contributing, to test the theory on other lenses, and skin tones, but it appears that the simplest for me might be to just bump the ISO, rather then deal with 1/3 aperture stops. All of the images show data from left to right, and none are blown, so my feeling is to opt for the exposure that gives the most data in the file????
Good to see you did another test.
The main issue as I see it is, you don't have a known control such as an 18% grey card. For example, the inclusion of an 18% grey card will provide you with a known value when correctly exposed. In post an eyedropper exposure value of a correctly exposed 18% grey card should be R:124, G:124, B:124. You don't have a known control in your test photos so what is actually influencing the histogram is quite frankly unknown.
One more issue that can affect exposure is lens extension, as you rack the focus ring to achieve an in focus subject that in your examples is quite close, you affect the distance the lens elements are from the sensor. For example, a micro lens(or some would say macro, a technical term intermixed) when at its closest 1:2 or 1:1 lens magnification will drop the exposure reaching the sensor by as much as 2 stops. That is why when actually calibrating a camera meter the lens should be set to infinity and your 18% grey card should fill the frame, an out of focus target is irrelevant when seeking exposure accuracy. One can then use the camera meter to determine if the hand held meter gives the same reading or visa versa. With flash, the lens should still be set to infinity and two stops down from wide open(if possible), one should choose a lens that produces less vignetting or at least a solid portion of the centre at which one can test with an eyedropper in post.
For the most part, your hand held meter if properly calibrated and the camera and lens are functioning correctly they should produce bang on exposures. Camera light meters on the other hand can be subject to manufacture bias but that's another topic.
@JBPhotog I was well out of the minimum focus distance on the lens, and cropped but I see your point. I shot these with my FA 100 mm f2.8, would a 50mm f1,4 be a better choice?
I was not aware that shooting the grey card OOF would not have a bearing on the exposure value. That sounds somewhat like the approach that the Expo Disc uses. For incident exposure you can take a test shot and look at the histogram, if the spike is centered in the middle, the exposure is correct. If not you adjust either way to bring it to the center. I found it in a drawer, but unfortunately it's not in a diameter that matches any of my lenses, and I don't have any compatible step down rings. Once I get through the holidays, I'll order a target and try again in studio, profiling all the glass.
Rather than a solid grey target, have you used one of these??? GEP Store: Exposure Target - Large they look fairly simple.
Have you consulted a thorough testing agency like dxomark to see if your camera manufacturer offers you real, true, dead-on accurate ISO speeds? It is not uncommon for ISO to be slightly off from true. If the normal ISO setting of 100 is really 80, or 64, then a particular camera model will have better noise scores in strict Laboratory Testing. If ISO 400 is really ISO 250 or 320, then the same thing is true, better noise results at a well-known intermediate ISO setting. ISO settings are not really dead-on accurate in most cameras, as far as I know. This subject has been bandied about in photo forums for multiple years now.
The need to add a bit of exposure from your meter reading has been common for a long time. Fine-tuning your working methodology is normal, and it helps account for all sorts of variables such as misadjusted lens diaphragms, light loss due to hi element numbers, differences in f/stop versus T-stop, etc..
It would be a simple adjustment to lower your light meter ISO setting one third or one half or even 2/3 of a value. Shoot your camera at ISO 100, but set the light meter to ISO 50, or 64, 80, and you have easily perfectly adjusted your light meter to your working methods and your equipment.
In your first post you noted concern about the accuracy of your meter. None of the tests you're doing appropriately reflect or provide info about your meter's accuracy. You're looking at raw files rendered in LR. Derrel was correct to note that Adobe's processing is a factor here and not a control variable. You can't make a judgement about the accuracy of your meter based on a camera raw file and Adobe's interpretation of that file.
So what you're doing is useful for you to sync your meter and camera to get a best result but you don't have any know constant in the loop that you can use that will allow you to judge the accuracy of the meter. If anything the meter is probably the most reliable player in the crowd.
If you want the exposure that gives you the most data don't bump up the ISO as that reduces data -- use an EC adjustment that changes f/stop or shutter speed.
What you really should be trying to arrive at is an exposure index commonly abbreviated as EI. I would guess that your EI should be roughly 1/2 stop lower than the iso level that your camera is set to, at least when the camera is set to ISO 100.
Not so much about the accuracy, but rather matching the two, so that if the meter says f** and the camera is set to the same, I should be reasonably sure that my exposure will be correct for a flash shot. I'm assuming that I should be somewhere in the +/- 1/3rd stop, but it's more like just over 2/3rds stops. I realize the problems with histograms, but that's not the only thing I'm judging the image on.
So would there really be that much data loss between ISO 100 and 200?? While I can adjust the aperture to account for the difference but then I run into a declining DOF. Shutter adjustment is pretty much a non issue on flash photograph, unless you're dragging the shutter to include ambient light.
Well duh, I hate it when I get tunnel vision on a problem. Since the L308 doesn't offer calibration I'd completely eliminated it as a solution only viewing it as a problem. I followed your advise setting the meter ISO to 64, to get the 2/3rds of a stop I needed, adjusted the light to read f/8 on the meter, set the camera to f/8, ISO 100, problem solved. That gets me close enough that only a tiny bit of adjustment will be needed post. Thanks Derrel!
I need to set it up in studio and test on a subject, but for now I think I have a baseline that I can adjust to. That still doesn't explain why the difference seems to be getting worse. Eventually I'm afraid I see new meter in my future.
Oh right flash -- no shutter speed. So exposure in your case is flash + f/stop. ISO is not going to change exposure then and will not effect how much data you record unless you cause some ISO clipping in which case you would lose data off the top. If raising the ISO then makes the raw image easier to work with in LR no harm; you're just under-utilizing the sensor some.
FWIW, I always used a 50mm when doing film densitometery calibrations. The main issue is to not cast a shadow on the card. With digital I am not as fussy and use the lens I am seeking numbers on.
I have never used the card you linked to, my exposure card is the Sekonic Exposure Profile Target II but I use it with my L-478DR-U to build DR profiles for my cameras. As long as you have a known target that has some references for quality and testing then it should get you real close.
Separate names with a comma.