Detail in bright whites

[Andre van As]

Hi Roger
This science section is welcome as it allows focus on specific issues. As I am learning about the vagaries of the digital medium it seems that to avoid prolonged trial and error understanding what you are doing helps. I have read the full version of your article on ISO at least a dozen times and I note that you challenge the concept of ETTR. I tried ETTR last week while trying to capture shots of a giant egret in bright sunlight and despite some aggressive PP in ACR (6.1) a very burned out patch remains on the birds' shoulder area. The image was captured with the HIIID2 50 MP CCD sensor. ISO 800, EV + 0.3, SS 1/800, f14, lens HC 300mm X1.7 TC. In addition to not recovering detail in the burned out area I am disappointed in the detail on the rest of the bird. Seems under these conditions I need perhaps to ETTL or at least remain neutral. I would welcome any advice that you can give me. An ISO of 800 results in a very noisy picture but I decided to try it to give me some latitude with short exposure times. The lower ISO's of < 400 seem to give better results which is commensurate with what you suggest in your article.
The image posted below

Regards

Andre

119658

[John Chardine]

Hi Andre- I don't think Roger has challenged the concept of ETTR in previous threads, but I'll of course let him answer that. The specific (frequently read) idea that somehow more bits of data are used to record the right side of the histogram than the left, and that's why ETTR works, has been deemed bogus in at least one previous BPN thread on the subject. ETTR works because by doing it you increase your signal to noise ratio. You do this by increasing your exposure and thus letting more photons in, not by upping ISO. Perhaps this is what you were referring to in the ISO thread.

In the egret image you posted, the burned patch has nothing to do with ETTR. The exposure was simply pushed too far to the right to the point where all three colour channels were clipped, thus not allowing Recovery to reconstruct based on at least one non-clipped channel.

[Roger Clark]

Hi Andre, John,

Yes, I have been saying ETTR becomes irrelevant at high ISOs, but that is for underexposure. Once ISO (post sensor gain) is high enough, sensor read noise dominates over camera electronics noise so it does matter if one "underexposes" as one can boost signal in post processing with no loss. But I did not mean that one can clip the highlights. Andre, your image had too much exposure for the ISO which saturated the highlights in all channels. The sun is to your right (phase angle about 90 degrees), which means a very high contrast scene, making the dynamic range difficult to cover. The bird's back and top of the head are facing the sun, so maximum brightness. If you move to the right so the sun was behind you or over your shoulder, the dynamic range would be reduced. Then, whatever ISO you choose, the exposure should not clip whites that you do not want clipped. If you still want the sun to your right, then the exposure must be reduced to not clip the whites, or reduce ISO to digitize the larger range (keeping exposure time and aperture constant).

Roger

[Andre van As]

Hi Roger and John
Thanks for your rapid responses. From the practical aspect I am limited (with my 8 lb rig!) to a max ISO of 800 and max SS speed of 1/800th. So if I want to capture fast moving images those are my rate limiting variables that I have to work within. So BIF and charging lions, in harsh lighting conditions where the sun is well above me, will be off my choices of photography with my current system. It seems that in the conditions that I encountered an EV of + 0.3 was what pushed the the ETTR too far and thus clipped the whites. I do envy the owners of CMOS based systems that seem to be able to push the limits of SS and ISO to give quite good results in marginal lighting conditions. I presume they will have the same issues as I have with harsh lighting?
Roger, If I then choose to work at lower ISOs (<400), that largely eliminate the sensor noise issue, I presume that I will still have to be very aware of the problems associated with harsh lighting conditions and tend toward under exposure? So it seems that the best time for outdoor nature photography is still before 10am and after 4pm - as in the old days.

I appreciate your input and will continue to experiment.

Best regards

Andre

[John Chardine]

Andre- Regarding detail, your sensor has 6 micron pixels which puts it somewhere in between a Canon 1DIV and 5DIII. So you are laying a good number of pixels down over your subjects. Could you provide a 100% unsharpened crop of a focused part of the image with lots of (potential) detail (the eye looks pretty good). It's virtually impossible to assess detail in a full-crop, down-sized, optimised image.

[Andre van As]

Hi John
I took the tiff file and converted it to a jpeg to meet BPN posting standards. The tiff file was created from the RAW fff Hasselblad file and I did some adjusting prior to exporting it as a tiff. This consisted of adjusting the histogram grey spaces, sharpening as recommended by H - 235/3.5. Denoise to remove obvious noise because of the 800 ISO. Lens correction is automatically applied. So this image has no PS/ACR PP.

I hope I have interpreted your request correctly

Regards

Andre119692

[John Chardine]

Not quite Andre. Sorry I didn't elaborate on what a 100% crop means. You need to crop the image to 1024 x 800 px with no resampling of the pixels. That way, a pixel in the image will be rendered by a pixel on the display- hence "100%". In Photoshop CS6, just set the units to pixels and use the crop tool to drag across your image. The readout will be in pixels and you can stop dragging when you reach 1024 x 800. Then position the image within the crop marks so that you are keeping a detailed, focused part. Then crop. This method of "pixel-peeping" is very unforgiving but is a great way to assess IQ.

[Tom Graham]

"....no resampling of the pixels....."
I've been wondering about such. If I start with the camera RAW file and "bring" it into PhotoShop, will not PS resample or otherwise re-interpret the RAW pixels into JPG? Does PS have settings as to how it changes the RAW into JPG?
Tom

[John Chardine]

Hi Tom- Resampling means either combining pixels to produce a lower resolution image or interpolating pixels to make a higher resolution image.

I don't know a lot about the technical aspects of RAW development but I can say that when you open a RAW file in Photoshop, Adobe Camera Raw intercepts the RAW file and allows you to change a huge variety of settings before the image is brought into Photoshop. At that stage it is not a jpeg file. Rather the data are internalised in a format used by Photoshop. Once you save the file, the data can be written out in one of several file formats such as tif, psd (Photoshop native) or jpg etc.

[Tom Graham]

Oh, yeah, John, thanks, guess I have jpg on the brain. Experts here like Roger I think convert the RAW to 16 bit TIFF (?) and use it as base for further processing.
Tom

[Michael Gerald-Yamasaki]

Greetings. I've long been unhappy with the concept of ETTR but ever more so with ETTR and whites. Here is a grayscale image I concocted:

119771

Here you have three grayscales differing in resolution 256, 64, 32 shades of gray (maybe one less ;-). In any event, I think it useful in showing that one's visual impression differs from white to black depending on where one is in the range and how much difference there is between bands. Here the 3 scales differ in contrast, that is, the bands differ in gray values by 1, 4, and 8 respectively (in the grayscale range of 0 to 255).

Take a close look at this image (download it, zoom in, scroll around). The banding may not be obvious in the left part of the image but it is there and if you zoom in you will see it (particularly if you move the image around a bit). The thing to notice is how hard it is to see the banding in the whiter part of the image. This is true for all three grayscales... detail in the whitest whites is hardest to see (and the blackest part as well).

The other thing you might notice is in the rightmost scale the bands don't look a constant gray in each band. A each band looks darker against the lighter neighbor and lighter against the darker neighbor. I haven't studied this edge effect all that much but it seems to me that the wider the band the greater the edge effect. I have noticed that sometimes I perceive a halo that isn't there (not there by the color values) because of this edge effect.

For white feather detail, it seems to me that:
- contrast, that is very localized contrast, sometimes refered to as "micro-contrast", rules much more than exposing to the right
- exposing to the right actually pushes the whites into a part of the grayscale where differences are harder to perceive
- feather spacing, light, interstitial shadows play a greater role than ettr

Anyway some food for thought.

One other thing, color (off white) helps by adding color contrast to grayscale contrast in promoting perceived detail in "whites".

Cheers,

-Michael-

[Roger Clark]

Here you have three grayscales differing in resolution 256, 64, 32 shades of gray.....


Hi Michael,
Very interesting post and interesting diagram. But I feel it has drifted from the original post, and it is quite interesting on its own. So please resubmit it as a new thread, and I'll ask that others also do not respond to your post here, but respond to your new thread. Tom, you would resubmit your response to Michael in his new thread; that would be appreciated. I tried to find a way to move the posts, but it seems I only have power to move whole threads. As a moderator, I can delete posts, but did not find a way to move one.

Roger

[John Chardine]

I moved Tom's post to the new thread but it's coming out before Michael's post. When you move a post it's inserted in chronological order into the thread and for some reason, Tom's post has an earlier time than Michael's. I think we have a time warp here!!!

Here's the moved thread:

http://www.birdphotographers.net/forums/showthread.php/104075-ETTR-amp-Perception-of-Whites

OK I see what happened. Michael's post has a time stamp when it was moved by Roger. Tom's seems to have retained its time stamp from posting. I'll see what the web guys can do about this.

OK, solved it but don't know how I did it!

[Roger Clark]

Hi John,
I didn't move Michael's post. He apparently reposted it as I asked. I finally found the moderator tools at the bottom of the thread (only mods see that). But I still see Tom's thread as the main and first thread in the new thread.

Tom, would you please repost your response to Michael in the new thread, then I'll delete the moved post and everything will look fine. Once all is complete, I'll delete this discussion here as it is confusing on not relevant to the topic.
Thanks,
Roger

[John Chardine]

Roger- I managed to sort it out and was able to move Tom's post over and get it in the right order. Your browser must be caching the page and giving you the old one.

[Doug Brown]

I can't say that I'm a big fan of ETTR, unless there are dark areas in the frame that I'm concerned about extracting detail from. In your photo of the Egret, pushing the exposure to the right on a white bird doesn't offer any appreciable visual benefit, but does significantly increase the risk of overexposure. About the only thing that ETTR will do in your photo is help with the shadows in the grass, but that can be done quite nicely in post if you're using a quality camera body. If I had been taking this photo, my sole focus would have been to get the exposure correct on the bird; underexposure to protect the whites would have been my likely choice. ETTR also extracts a penalty; slower shutter speed, less DOF, and/or higher ISO for a given amount of ambient light.

[Doug Brown]

Here's a flight shot of a Barred Owl that is grossly underexposed. I was using manual exposure, and the bird flew from an area of sun to an area of shade.

120040

[Doug Brown]

Here's the same frame with the exposure slider moved up 1.72 stops in Lightroom. This is obviously an extreme example of shadow recovery, but the fact that it can look so good makes me wonder why so many folks advocate ETTR.

120041

[John Chardine]

Yes but Doug, I think we are talking about optimisation not just improvement. No one would dispute that the (beautiful) post-processed owl image is not an improvement over the original, but is it optimal? - no I don't think it is.

In all but the flattest of images there are going to be bright parts to the right and dark parts to the left so ETTR will work and be of value in most cases. The minuscule risk of clipping the highlights (if you take care to expose properly), is an insignificant price to pay for an image with an overall higher S/N ratio.

[Roger Clark]

I agree with John. Doug, recovering an underexposure is certainly preferable to trying to recover a clipped overexposure, but your recovered image does have a fair amount of noise. It would have been a better image with a stop or so more (real) exposure, assuming that more exposure time would not have resulted in blur of course.

Roger

[Doug Brown]

I agree with John. Doug, recovering an underexposure is certainly preferable to trying to recover a clipped overexposure, but your recovered image does have a fair amount of noise. It would have been a better image with a stop or so more (real) exposure, assuming that more exposure time would not have resulted in blur of course.


To John and Roger, I'm not advocating underexposure by 1 2/3 stops; I had metered the bird for sun, but it flew through a shaded area while I was tracking it. In spite of my incorrect settings, the salvaged image is entirely usable (the 100% crop that you see has no NR applied). Overshooting the ETTR mark as the OP did results in an unusable image. For me, ETTR has to have a demonstrable benefit before I'll consider using it on a given shot (for example a white and black bird). I fail to see the benefit of ETTR in the OP's image of an all white bird, other than brightening the shadows in the grass. But I do see one of the downsides of ETTR (overexposure). Here's a 100% crop of that same Owl flying in sunlight with no exposure adjustment in LR. It's a correct exposure give or take 1/3 stop, which is generally what I strive for in camera.

120049

[Greg Basco]

Doug, I think your main point advocating use of ETTR strategically in terms of concentrating on the important tones in the image rather than just using it as a matter of course in every image is a valid one.

Cheers,
Greg

[Julian Mole]

Hi Everyone,

Just found this interesting thread and thought I would add my perspective on Exposing To The Right as someone who would class themselves as relative novice.

I think that ETTR is a valid technique when used by experienced photographers in the appropriate circumstances but can lead inexperienced photographers into overexposing unnecessarily because they neither have a solid understanding of exposure theory, judging an exposure on a histogram or of ETTR itself.

With the recent DSLR models people like myself are better off getting a solid understanding of exposure and applying it in camera (especially estimating EC for the first exposure) before trying to do advanced techniques like ETTR, ie. It's better to learn to walk before trying to run. ;-)

Anyway that's my take on ETTR,

Julian.

[David Stephens]

Many people seem to be losing sight of what ETTR is not. It is NOT exposing so far to the right that you blow out highlights. If you use -EV on a white bird in full sun, you may still have to lower EV further to the left to avoid blowing out important highlights. Those of us that routinely ETTR know that a white bird in direct sun will require a -EV adjustment. For brown and black subjects you almost always can apply significant +EV, but white subjects or black and white subjects need to be exposed in favor of not over exposing important highlights.

[John Chardine]

I will add to Dave's NOT list and say that ETTR is not rocket science either. Say for example you have a nice sparrow (not a White-throat!), evenly lit in overcast conditions against a BG of more or less similar tonality (i.e., not hugely brighter or darker than the subject). Take an evaluative meter reading with no exposure compensation and see that the histogram is a fairly well centred and not extending to the right hand box. Add some exposure (slower shutter speed, bigger aperture) to get some pixels in the right hand box. Bob's your uncle as they say. You have just EedTTR. In case you have a White-throated Sparrow, then the pixels on the throat will likely extend some into the righthand box of the histogram with no EC, so take the exposure no further. If the throat is blown, bring the exposure back an amount to keep some pixels in the righthand box.

[Tom Graham]

[QUOTE=Greg Basco;851141]Doug, I think your main point advocating use of ETTR strategically in terms of concentrating on the important tones in the image rather than just using it as a matter of course in every image is a valid one.

Immediately for me flashed Ansel Adams Zone System and his book "The Negative". A very enjoyable/readable book (as I find all of A. Adam's), and although about B&W it stills gives a wonderful explanation of light and film(sensor) and the photographers visualization of the final image. Book still in print, available from amazon for less than $20.
Tom

[Roger Clark]

I agree with John and David above. Of course in an action situation with changing light and working in manual, one may not be able to respond fast enough to get the best exposure. ETTR is about getting the best exposure that maximizes signal-to-noise ratios without clipping any highlights one didn't want clipped. Certainly on a good camera, one can recover a decent image with underexposure at a decent ISO for the given camera, but that it not optimum.

Roger

[Andre van As]

Hi to all
I was somewhat reticent to initiate this discussion as I clearly have a lot to learn about digital photography and PP. However I thank all of the participants for chiming in and I have learned a lot! Seems that by observing some basics there is room for a common sense approach.

Regards

Andre