22 replies to this topic

[Joejr14]

Breaking this off into a separate thread since the Monpod one was turning into something different.

In that thread a discussion came about on histograms, exposure, and some other technical mumbo jumbo. I know many are not at all concerned about any of that, but I'm sure a few of you are.

So, let's start with exposure. Exposure is nothing more than the amount of light that falls onto the photographic medium (either film or a sensor) during a shutter cycle. What's correct exposure then, well, that depends on the photographer and what the intent is with the photograph.

Since this is a bird photography website, we'll go with the assumption that the intent is to have a correctly exposed bird that shows detail in the whites and blacks (ie has dynamic range). Overexposure is when the whites are 'hot, clipped, or blown', same with underexposure except it obviously related to the blacks in an image.

So, time for some examples. Tell me what each of these pictures are and why, and remember we're focusing on the bird, not the background. Btw, for these shots I specifically used pictures of tough subjects (all white birds and dynamic birds).

[Joejr14]

A couple more....

[canon eos]

Getting too technical for me, Joe

I studied and studied photography some 35 years ago (and haven't stopped!) but am happy to back off from technical discussions wherever I am. It's the same with retirement; after a career in engineering it's the last thing I want to think about now........LOL

But this could be a good thread. There are many who could gain from this, especially as you say as applicable to bird photography, one of the greater challenges for exposure control.

[Joejr14]

Or it could be a thread that nobody participates in! LoL.

Nobody wants to take a crack at the exposure examples? I promise I won't be too harsh on the answers!

[David Case]

OK, I'll bite ...

#1 - the exposure is about right. There is slight blowout on the back of the head in the whites but on the other hand the black wing tips and tail tip show no detail. So you will lose detail either way you adjust the exposure.

#2 - overexposed. Lots of detail loss in the whites, none in the grays.

#3 - overexposed. Egrets are tough and easy to overexpose. Lots of detail loss on the neck and down the back of the bird. There is detail in the shadows though.

#4 - overexposed. Detail loss on the right wing and legs and top of head in the whites. Lots of room to bring the shadows down.

#5 - underexposed. No blownout whites, but some loss of detail in the dark shadow areas of the tail, wingtips, and back of the head.

All of this depends on monitor calibration, etc. But that is what I am seeing on the monitor I am working on.

[Joejr14]

I figured you and Michael would be the two to respond to this thread.

I'll withhold my thoughts to see if anyone else wants to respond.

[JimB]

I agree with David's response to the images as presented. However, the histograms are not showing any blown highlights (except for a few hot pixels on the neck of the egret) so if you were to judge the exposure solely on the histogram they would appear to be OK. But, through jpeg compression and other digital influences detail has been lost in the brighter areas resulting in the first four images appearing to be overexposed. If you shot in RAW I'm sure there is detail in those areas that can be brought out with some more post-processing. So, I would say the capture exposures were OK for the first four shots it's just that the image quality suffered some in processing for the web. The bottom image does appear to be underexposed, but again, the histogram doesn't show many blocked areas...No real easy answer to this

[David Case]

I agree with David's response to the images as presented. However, the histograms are not showing any blown highlights (except for a few hot pixels on the neck of the egret) so if you were to judge the exposure solely on the histogram they would appear to be OK. But, through jpeg compression and other digital influences detail has been lost in the brighter areas resulting in the first four images appearing to be overexposed. If you shot in RAW I'm sure there is detail in those areas that can be brought out with some more post-processing. So, I would say the capture exposures were OK for the first four shots it's just that the image quality suffered some in processing for the web. The bottom image does appear to be underexposed, but again, the histogram doesn't show many blocked areas...No real easy answer to this

Yes, looking at the histograms is the way to go. I was at work getting ready to head home and just did a quick judgment based on what I could see. Regarding histograms and exposure, the advice I have heard most often is to expose as bright as you can without blowing highlights. That makes sense to me. After all, the more light you have the better chance the sensor has of picking up color and detail and eliminating noise. But I am usually not quite that picky and am happy if I don't blow highlights or shadows even if I am not as right as possible.

[lonestranger]

To my eye, I have to agree that 2, 3, and 4 look overexposed, and 5 looks underexposed. As far as the histograms go, I haven't seen them myself, but aren't histograms potentially misleading? It may not apply to these particular photos, but sometimes it's necessary to blow the highlights or lose the shadows to get the correct exposure on your subject. I review my histograms all the time but use them only as a guide, I don't rely on them 100%.

[David Case]

To my eye, I have to agree that 2, 3, and 4 look overexposed, and 5 looks underexposed. As far as the histograms go, I haven't seen them myself, but aren't histograms potentially misleading? It may not apply to these particular photos, but sometimes it's necessary to blow the highlights or lose the shadows to get the correct exposure on your subject. I review my histograms all the time but use them only as a guide, I don't rely on them 100%.

Yes, it is sometimes necessary to lose something on either the high end or the low. The problem is that the eye can handle a wider dynamic range of light than a camera can, at least in one exposure. So some scenes are just too much for the camera. HDR is great in these cases provided you don't have moving subjects, and birds are moving subjects. I did manage to get some HDR shots of a fishing Great Blue Heron once, but off the top of my head I can't think of many birds that will sit still long enough to make that possible.

But if I do have to lose dynamic range I usually prefer not to blow out the highs since shadows are dark and show less detail anyway. But that's just usually. Sometimes I prefer to blow out the highs for a special effect. This shot of a quartet of egrets is an example. Since I was shooting into the sun the birds were strongly backlit and to get any detail in them I had to increase the exposure to the point where the water in the foreground was completely blown out to white. But I like that effect as you only get the sense of the water in the shadows of the egrets. Overall I would not think of this image as a 'photograph' in the realistic sense, but something more akin to an impressionistic painting. In the end my goal is to make interesting and beautiful images and not photographs, per se. So I agree that we have to sometimes lose highlights or shadows, but I don't think that there is a simple formula for the 'correct' exposure. It all depends on what you are trying to accomplish.

BTW, most of the creative work on this image was done in post processing. I didn't meant to imply that I thought all of this through while I was taking the shot because I didn't. I was trying to stay hidden so as not to flush the birds and get as many interesting poses as I could. The rest of the creative work came later. I have a hard time tossing shots for this reason. Every once in a while I will find some possibilities in a shot that at first glance I would just trash. This was one of those.

[Joejr14]

Okay, so for my answers....

1) I think this is actually a bit underexposed. In lightroom I have details in the whites and blacks, and I think a little extra light would help in the shadows under the wing. Some selective fill lighting would also work wonders here.

2) The posted image was pushed 1/2 stop in Lightroom for the purposes of the quiz. Although I'm not sure that it's actually overexposed even pushed the 1/2 stop. If it is, it would appear in my raw file it's only the back of the head by the white/grey line.

3) Totally blown highlights along the back of the head all the way down the neck. The image really isn't salvageable through post-processing. Good thing that this was my first shot at the egret and I quickly fixed the exposure and nailed it for the rest of the pics.

4) Obviously over-exposed. Right wing and left thigh.

5) Underexposed. Any increase in exposure would have blown the whites on the head. Not much you can do about crappy lighting.

[Joejr14]

Yes, looking at the histograms is the way to go. I was at work getting ready to head home and just did a quick judgment based on what I could see. Regarding histograms and exposure, the advice I have heard most often is to expose as bright as you can without blowing highlights. That makes sense to me. After all, the more light you have the better chance the sensor has of picking up color and detail and eliminating noise. But I am usually not quite that picky and am happy if I don't blow highlights or shadows even if I am not as right as possible.

For ha ha's, take a static picture of something indoors in constant lighting. Change your exposure from what the camera suggests to brighter without clipping anything. Upload your pictures and look at the size difference between the 'properly exposed' one and the picture that's 'exposed to the right'. Shots exposed to the right are always larger in terms of size---they hold more details.

[David Case]

For ha ha's, take a static picture of something indoors in constant lighting. Change your exposure from what the camera suggests to brighter without clipping anything. Upload your pictures and look at the size difference between the 'properly exposed' one and the picture that's 'exposed to the right'. Shots exposed to the right are always larger in terms of size---they hold more details.

I have been meaning to test out your suggestion for some time and finally got around to it today. I shot a series of 34 exposures of my guitar music collection, all at the same aperture, ISO speed, and focus. My camera was set up to change the exposure in 1/3 stop increments. All files were shot in RAW and clipping was tested using Digital Photo Professional, Canon's RAW editing software that comes bundled with their cameras. Here is a table of the results ...

Camera - Canon EOS 5D Mark III
Lens - Canon EF50mm f/1.8 II
Aperture - 2.8
ISO Speed - 100


FILENAME ........................ SHUTTER ..... CLIPPING ..... EXPOSURE ..... SIZE

Exposure_Test_01.CR2 ... 1/125 ............ shadows ....... under ............... 23.4MB ........ darkest exposure
Exposure_Test_02.CR2 ..... 1/100 ............ shadows ........ under ............... 23.5MB
Exposure_Test_03.CR2 ..... 1/80 .............. shadows ........ under ............... 23.6MB
Exposure_Test_04.CR2 ..... 1/60 .............. shadows ........ under ............... 23.7MB
Exposure_Test_05.CR2 ..... 1/50 .............. shadows ........ under ............... 23.9MB
Exposure_Test_06.CR2 ..... 1/40 .............. shadows ........ under ............... 24.1MB
Exposure_Test_07.CR2 ..... 1/30 .............. shadows ........ under ............... 24.4MB
Exposure_Test_08.CR2 ..... 1/25 .............. shadows ........ under ............... 24.7MB
Exposure_Test_09.CR2 ..... 1/20 .............. shadows ........ under ............... 25.1MB
Exposure_Test_10.CR2 ..... 1/15 .............. shadows ........ under ............... 25.5MB
Exposure_Test_11.CR2 ..... 1/13 .............. shadows ........ under ............... 25.9MB
Exposure_Test_12.CR2 ..... 1/10 .............. shadows ........ under ............... 26.3MB
Exposure_Test_13.CR2 ..... 1/8 ................ shadows ........ under ............... 26.9MB
Exposure_Test_14.CR2 ... 1/6 ................ none .............. correct ............ 27.4MB ........ correct exposure (closest to actual scene)
Exposure_Test_15.CR2 ..... 1/5 ................ none .............. over ................. 28.1MB
Exposure_Test_16.CR2 ..... 1/4 ................ none .............. over ................. 28.8MB
Exposure_Test_17.CR2 ..... 0.3 ................ none .............. over ................. 29.6MB
Exposure_Test_18.CR2 ..... 0.4 ................ none .............. over ................. 30.4MB
Exposure_Test_19.CR2 ... 0.5 ................ none ............. over ................. 31.3MB ........ brightest exposure with no clipping
Exposure_Test_20.CR2 ..... 0.6 ................ highlights ...... over ................. 32.1MB
Exposure_Test_21.CR2 ..... 0.8 ................ highlights ...... over ................. 33.1MB
Exposure_Test_22.CR2 ... 1.0 ................ highlights ..... over ................ 33.1MB ........ exposure with the most detail (using file size as a proxy)
Exposure_Test_23.CR2 ..... 1.3 ................ highlights ...... over ................. 32.8MB
Exposure_Test_24.CR2 ..... 1.6 ................ highlights ...... over ................. 32.4MB
Exposure_Test_25.CR2 ..... 2.0 ................ highlights ...... over ................. 31.9MB
Exposure_Test_26.CR2 ..... 2.5 ................ highlights ...... over ................. 30.7MB
Exposure_Test_27.CR2 ..... 3.2 ................ highlights ...... over ................. 29.5MB
Exposure_Test_28.CR2 ..... 4.0 ................ highlights ...... over ................. 28.4MB
Exposure_Test_29.CR2 ..... 5.0 ................ highlights ...... over ................. 27.6MB
Exposure_Test_30.CR2 ..... 6.0 ................ highlights ...... over ................. 26.7MB
Exposure_Test_31.CR2 ..... 8.0 ................ highlights ...... over ................. 26.0MB
Exposure_Test_32.CR2 .... 10.0 ............... highlights ...... over ................. 25.2MB
Exposure_Test_33.CR2 .... 13.0 ............... highlights ...... over ................. 24.5MB
Exposure_Test_34.CR2 .. 15.0 ............... highlights ..... over ................ 23.4MB ........ brightest exposure

The exposure with the most detail (using file size as a proxy for detail) was number 22 which came in at 33.1 MB. File sizes increased up to this point from a minimum of 23.4 MB and then began to decrease again until the same minimum was reached. However, the exposure with the most detail was not the best exposure, at least judging by the closest overall match to the brightness of the scene. That was number 14 which was also the lowest exposure without clipping. My guess is that the scene overall had more shadow areas than bright areas so that the camera found more detail in the shadows than it lost in the highlights as the exposure increased. So, Joejr14, your thesis was proved correct and then some in that detail continued to increase even as the highlights began to clip.

Finally, here are the five images in bold for reference ...

Exposure_Test_01.CR2 1/125 shadows under 23.4MB ........ darkest exposure


Exposure_Test_14.CR2 1/6 none correct 27.4MB ........ correct exposure (closest to actual scene)


Exposure_Test_19.CR2 0.5 none over 31.3MB ........ brightest exposure with no clipping


Exposure_Test_22.CR2 1.0 highlights over 33.1MB ........ exposure with the most detail (using file size as a proxy)


Exposure_Test_34.CR2 15.0 highlights over 23.4MB ........ brightest exposure


Well, I hope that was interesting and helpful. The take away message, expose as bright as you can without clipping the highlights. You can always tone things down later if the overall effect looks too bright.

[lonestranger]

Very interesting experiment, David Case. I was wondering if you'd be kind enough, and/or interested enough, to process the photo with the highest exposure with no clipping(test #19), and the photo with the most detail(test #22), so we can compare the differences after they have been toned back down to the correct exposure of test #14?

[David Case]

Very interesting experiment, David Case. I was wondering if you'd be kind enough, and/or interested enough, to process the photo with the highest exposure with no clipping(test #19), and the photo with the most detail(test #22), so we can compare the differences after they have been toned back down to the correct exposure of test #14?

Yes, and now that you mention it I should have done that in the first place for the sake of thoroughness. Digital Photo Professional only allows two stops of adjustment up or down though so I was unable to bring #22 all the way down to #14 as you requested. So instead I used image #16 as a reference and brought the others up or down to match it. Here are the images with the stop adjustments marked for reference ....

Exposure_Test_14 ....... exposure adjustment = +0.67 stops


Exposure_Test_16 ....... exposure adjustment = 0 stops


Exposure_Test_19 ....... exposure adjustment = -1.00 stop


Exposure_Test_22 ....... exposure adjustment = -2.00 stop


To my eye #14 and #16 are indistinguishable. #22 shows strong clipping in the highlights on the wall and on the white titles of the loose-leaf notebooks on the book shelf. The interesting one is #19. Even though DPP says there is no highlight clipping, there is still loss of detail in the bright areas. They look brighter and flatter than in #14 and #16. So maybe you have to be a bit careful about exposing too bright after all, especially if the highlight areas contain detail that you want to preserve. If we assume that the algorithm for detecting clipping in the software is essentially the same as the one used in the camera, there may be flattening in the highlights even if your camera says there is no clipping. Interesting.

[lonestranger]

Interesting results, David Case. It appears to me that shooting slightly to the right might actually be better than shooting as far to the right as possible, at least in these examples. Just out of curiousity, how do the file sizes compare after processing? Is the detail comparable to the test shots with the processed #22 still having the most detail based on file size?

[David Case]

Interesting results, David Case. It appears to me that shooting slightly to the right might actually be better than shooting as far to the right as possible, at least in these examples. Just out of curiousity, how do the file sizes compare after processing? Is the detail comparable to the test shots with the processed #22 still having the most detail based on file size?

lonestranger, I can't see offhand a meaningful way to make the comparisons you want. Editing RAW files is non-destructive and so will not have any effect on the file size. Converting from RAW to JPEG involves moving from a 14 bit to an 8 bit format so there will be loss of detail for all images as a result of the conversion. But I will give it some more thought as I have time.

[Joejr14]

I think Lonestranger was getting to whether or not the file size changed for ETTR images when they were then pulled back a stop or two via post-processing. And to answer that, no.

And yes, you're always better exposing to the right to get that extra detail. I think exposing how far depends on exactly what you're trying to photograph (ie, bald eagle vs deer....one a high dynamic range subject the other not).

[lonestranger]

Actually, it was the size of the processed jpgs that I was curious about. In the raw images, #22 had the most detail based on file size. After processing, does #22 still have the most detail based on file size compared to the other processed jpg images? I know there is a loss in detail when converting from raw to jpg, and theoretically the largest raw file should convert to the largest jpg file, does that theory hold true in this test?

[Joejr14]

Actually, it was the size of the processed jpgs that I was curious about. In the raw images, #22 had the most detail based on file size. After processing, does #22 still have the most detail based on file size compared to the other processed jpg images? I know there is a loss in detail when converting from raw to jpg, and theoretically the largest raw file should convert to the largest jpg file, does that theory hold true in this test?

Right....isn't that what I said? Or maybe just what I meant? Depending on the level of post-processing, yes, in theory the largest RAW file converted will make a larger JPEG.