abluesky: Re:ETTR in Hi DR scenes
I thought that sever pulling of the shadows was to be avoided precisely because of the increase in noise that it creates?
This question is answered fully in your thread:
edhannon: I seem to remember (harder as I pass 70) that ETTR was originally designed to overcome the problem of an exponential tone curve wrt f-stops. Half of the bit depth is in the first stop below saturation - half of the remaining is in the nest stop, etc.
If you underexpose and then brighten in PP won't you end up on the shallow slope of the tone curve and loose most of the bit depth? So don't you have to balance the gain in highlight headroom with the lose of effective bit depth?
The issue in your first paragraph is dealt with in the "technical underpinnings" addendum on page 2, Ed.
As to the second paragraph: everything is a balancing act. If you want to retain specific highlights and do it all in one shot, then you've got to put up with what you've got. If you're willing to give up the highlights (I rarely am), then you can blow things to your heart's content to gain more in the shadows. But I would call that falling over the edge rather than balancing.
If you want the best of both worlds, and you have the opportunity and/or patience to do so, then multiple shots or other HDR techniques would be called for. But that's not ETTR, and I am dealing here with ETTR.
The Sage Knows: Thanks for taking the time to post this article. Like others, I think the example of the two gentlemen in the room may be more suited as a two exposure HDR-ETTR rather than a single ETTR. By putting the window to the right of the histogram and letting the main subject go into deep shadow at the LEFT of the histogram did precisely the opposite of what I believe is the main aim of ETTR, that is to reduce image noise by maximizing the recorded signal. Hence, a second exposure that disregarded the window and Exposed the interior-TTR would have been better. And if what is in the small window is still important, the two exposures could then be combined in post processing pretty easily. I'm sure you already know this.
Yes, Sage Knows, but this conventional wisdom is often more easily said than done, as is the case here. The men were moving and I was hand-holding the camera. I was at 1/6 secs, and it amazes me still how well the IBIS worked.
I had just come into this dark room from bright sunlight, and my aperture was at f/8. I needed to shoot quickly and didn't have a lot of time for adjustment. I could have gained an EV by opening up to f/5.6, but even then my shutter speed would have been only 1/12 secs. And, if I now tried to increase exposure the 2 or 3 EV needed for the "shadow shot" you suggest, I would have had to use such a long shutter speed that a decent hand-held shot would have been impossible.
And, in doing all that, I would have missed the shot.
Chris Noble: Good article Gollywop, thank you. However, in cases where the DR of the image is greater than 6-7 EV, it is often necessary to compromise between preserving highlights (by reducing exposure) vs. capturing shadow detail (by increasing exposure). In your example 1, the top 3-4 EV are "consumed" entirely by the small area of the window (50 pixels or so per exposure value), leaving most of the image exposed below -6 EV (50,000 pixels per exposure value, with no significant drop-off in pixel count even at - 12 EV). I would have added 2 to 3 EV to this exposure. You would have saturated a very small portion of the image (parts of the small window, which is not of great interest in this image), in return for pulling the majority of the image up out of the shadows. Same comment for example 3: the top 3 EV are (in my mind) "wasted" on keeping the sun below 0 EV, yet there is no tonal range there in the final image; meanwhile, the bulk of the image is below -6 EV. Just my opinion of course.
There are definitely cases where your point is well taken, Chris. However, the conventional wisdom expressed there is not always of hugely practical significance. It would be if one really required exquisite detail in the shadows. But that is often not the case.
In all the examples I included, I feel I got excellent content in those lower reaches. I love the images. What I hate, however, is blown, contentless windows -- or clouds, or white shirts, or whatever of substance is being sacrificed in the highlights. In the second image of those three high-DR examples, the portion of the scene that would have been blown by your suggestion would have been outlandish. The photograph would have been useless, whereas what I ended up with is far from so.
Blowing inconsequential highlights and specular highlights is one thing, but, to me, blobs on an image are ugly, and they distract me very much. I judge the quality of a photograph by such things.
That's a good thought as far as it goes. But if severe pushing of the shadows is the only way you can get the shot -- and if it nevertheless results in an acceptable final image -- then severe pushing of the shadows is the only game in town. The idea in these cases is to get acceptable shadows without blowing important highlights.
And, while less noise is clearly better than more noise (unless you're looking for an effect), noise is not always the end of the world. It can often be handled very well. So: either take the shot and have the image, or be content with the memory.
Thanks for posting this shot, DM. I really needn't comment because you know as well as I how wonderful a mood this carefully crafted shot sets. The composition, exposure, and processing are just what was needed. With a clear and commanding central theme, I particularly like how the cloud formation then flows into the foreground spruce (?) at the right-hand side. Lovely.
cmantx: Thank You Jimix Photo. I have nothing to add. You covered it well.
I'd briefly like to explain how I took my shot. I like to use "cloudy" WB in-camera most of the time. I shoot in RAW so I can tweak WB to my liking. Using MY monitor I was trying for an accurate rendering of the colors in the Hummingbird and the Cape Honeysuckle Flowers. The streak of yellow on the hummingbirds head is pollen. The background is a mix of pine tree foliage and a crepe mrytle's leaves that were in the process of changing to fall colors. So some red/brown is in the background. Not gollywop's cup-of-tea but I liked it.
Yes, you are no doubt correct that I was more caustic to cmantx than was appropriate. I'm sorry for that. But my comment was only partially directed at cmantx; I was more upset by what's implied about the somewhat errant voting in some of these challenges (which seems well characterized by those who mistake a yellow cast for warm afternoon light).
But, yes, my apologies to cmantx, who, to his great credit, appears to be willing to listen despite my moment of irritation.
gollywop: How is it that a photo with such egregiously incorrect WB can be awarded anything at all? How could you even think to enter it into a challenge without a little appropriate PP? And how could people be so unknowing as to vote for it? A good capture (which, by the way, it is) is not enough to make a good photo. I've lost all faith in the value of these "challenges."
I've posted my version at
I modified the jpeg WB in ACR with a Temperature adjustment of -36 and a Tint adjustment of -6. As I said above, try those values and toggle between the resulting image and the original WB and you'll see what I mean. You may prefer it a bit warmer, but, to my taste, I wouldn't go much more.
I do wonder if this issue may not have been the reason the voting on this image was so bimodal.
No, not my cup of tea. There is a decided, and, IMO, distracting yellow cast to the image, including the flower colors. The red is a somewhat unpleasant color, and is not at all "reminiscent of a warm afternoon light of a sunny day."
Of course, to each his own, but the eye during raw processing is easily tricked. The explanation is given in your saying you use a cloudy-day WB, which will indeed produce a significant yellow cast when employed in daylight lighting. After looking at such an image for a while during processing, the eye begins to adjust to make it look "right," and so leaves you with an undesirable WB. Use a WB reference card to keep you straight.
With the jpeg you've posted, a WB modification in ACR with a Temp adjustment of -36 and a Tint adjustment of -6 produces, at least to me, a very nice image.
Try starting there and toggle between that setting and the one you have. You'll see what I mean.
I love the capture, by the way, but not the color cast.
How is it that a photo with such egregiously incorrect WB can be awarded anything at all? How could you even think to enter it into a challenge without a little appropriate PP? And how could people be so unknowing as to vote for it? A good capture (which, by the way, it is) is not enough to make a good photo. I've lost all faith in the value of these "challenges."
This marvelous shot wants to be looked at in the original on a 24" wide-gamut monitor. It has its charm with the thumbnail, but when the original fills a large screen and is viewed at 2 feet or so, it becomes a wonder.
I would probably have messed it up by trying to pull the shadows up and putting a deeper blue saturation in the sky. But the more I looked at it, the more I realized that DM has it right.
Good work (as usual).gollywop
walkaround: I don't know where this term "brightening" came from. It's GAIN, since this is signal amplification we are talking about. "Brightening" implies there is no downside, but there is: noise. I would like camera buttons to say "Gain" instead of "ISO"... but current cameras and the people who use them (myself included) are still in a film mindset for many things. You probably won't change this any time soon.
This is all true, Timur, but it is also the source of unnecessary noise and confusion.
The important thing I would hope others would take away from the article is the separateness of all the kinds of things you mention from exposure.
Timur Born: Most people should just stay in sRGB space throughout their whole processing pipeline (camera, software, display, printer) and get away from the bigger-is-better mentality. Not to mention that today most images are viewed on all kinds of displays which more likely are working within the limits of sRGB than anything else.
Things can get *lot* more complicated once you want to use proper color management. Not only because of the "management" part of things, but also because you only really can make good use of it once you understand what you are doing and what the corresponding benefits and *limitation* are.
Thanks for writing this article and trying to help everyone get a better picture (pun intended) of all this.
And for those shooting raw, broader color spaces and proper color management definitely have the potential for increasing the quality of one's photographs.
Thanks for the comments, Timur. I do not, however, agree with your advice.
I do agree that those who shoot jpegs and do little, if any, post processing, might do well simply to stick with sRGB. But for anyone who does PP, including those who shoot jpeg and use sRGB, there is still a need to calibrate and profile one's monitor and printer.
These are essential parts of color management that are relevant to anyone who is beyond the P&S-send-it-out-to-be-printed category. These are the minimal aspects of color management that would almost certainly be relevant to anyone who has taken the trouble to read this article.
Fish tanks are designed and braced to withstand low-head water pressure applied from the inside. They are not designed and braced to withstand water pressure of any head applied from the outside. This idea seems to me to be particularly ill-conceived, even in "calm" water.
Thanks, Ted. However, you might read footnote 2 on page 2 regarding QE.
Also, as to the generality of the term brightening, please reread the definition given in the article:
Brightening: increasing the brightness of an image by any means (in-camera, in the computer, or in the output medium) other than those determining exposure (scene luminance, f-ratio, and shutter speed).
dcassat: GW, I read your posts that led to this article and wish to thank you for the continued effort on your part to bring the facts of digital exposure to the surface. Not only do you understand digital exposure but are willing to write about it and in a very clear, concise manner.
Your patience with me to understand digital exposure, and ETTR principles certainly engaged my technical brain and allowed me to seek out the best in the images I produce.
This article will continue to help others reach a higher level of awareness, if they wish to move beyond old school concepts/rigor/dogma as it is excellent explanation of what proper exposure in the digital realm actually is.
Thank you. - Dan Cassat
Hello Dan. Good to bump into you here. :-)
I truly appreciate your comments. Comments like those from people like you mean a lot to me.
thanks, and take care,gollywop
tony brown: I tripped over this 'ISOless' issue in some large threads recently and have found your article above to be the source material. It has given me much food for thought and practical experimentation with my Canon 6D. From sensorgen, I find the 6D to be a partial ISOless camera, ISOless from about 3,200 ISO, and have tried some experiments comparing brightened RAWs (in ACR) at 3,200 to those taken at ISO 102,800.
As you state, the noise is just the same but the gain of the ISO 102,800 setting has blown highlights when compared to the 3,200 + brightening. Note, there seems to be a considerable colour difference between the two methods, greatly reduced by also adding extra brightening to the green channel. So my tests continue with the prospect of never increasing ISO past the 3,200 setting unless the DR of the scene is low.
Thank you for the stimulating post. I will have to rethink much on the back of it. Empirical results will be my yardstick but thanks again.
Thanks for the kind words, Tony. I'm not sure what you mean by the "source material" for ISO-invariant. But the real sources for this are due to efforts from the like of Great Bustard, bob2n, Pierre Sottas, and others. The concept has been around for several years.
Meanwhile, have fun with your experiments.
WilbaW: Congratulations on your article. Maybe one day manufacturers will stop making digital cameras by following the traditions of analogue cameras... :- )
So where do we go from here with "exposure compensation"? It suffers from all the same problems, since we can change the image brightness via the ISO just as easily (often preferably) as we can via aperture and shutter duration. I find "metering compensation" works without any major problems - thoughts?
Thank you very much, WilbaW. I owe it to you that I took this route for publishing this material. I think it was a good suggestion and I'm pleased with the result.
As to EC: I shoot only raw and always with a fixed ISO, usually base. I typically shoot using A priority and use EC only as a fast means for altering shutter speed, aiming to achieve the "marginal blinkies" for ETTR and ignoring any metering implications. If that results in an SS that is too slow, I then consider whether to open the lens a bit (if possible or reasonable) or boost ISO up as far 800, which is as far as I go with my partially-ISO-invariant E-M5. By then I can usually get a shot that is capable of being rendered well during processing.
For me, EC is precisely what it says: Exposure Compensation. I do not use it as Brightening Compensation. :-)
I also use M mode with fixed ISO -- no EC.
Thanks again WilbaW. Take care,
"Gain" is a good term, and I use it in the article to augment the explanation of brightening. However, it's general use in this context is discursively ungainly (pun intended). It's one thing to say you're "brightening" an image; it's another to say you're "gaining" an image -- or that an image is "over gained" or "under gained." Something more awkward is needed, like "adding gain," or "has too much gain" or "has too little gain." It doesn't work well for normal conversation.
Moreover, "gain" doesn't convey. The intended meaning of "that image has too much gain" would be almost opaque to most people, while the meaning of "that image has been over brightened" is immediate and complete.
Gain, of course, has the same use in audio. But no one says, "that music has too much gain." They say it's "too loud" or "too soft." And when you hear someone say, "that music is too loud," you don't indignantly respond, "it's not loudness, it's GAIN."
That's where "brightening" came from. :-)