HDR Bracketing: One or two stops?

[Manfred M]

Never tried HDR but am trying to study up so I can attempt it soon. I have the Nikon ViewNet 2 that came with my 7000, but didn't load it. I do not see that will handle HDR, correct? What basic program would be good for a newbie?
I have a Windows 7 with Photo Gallery editing, and IPad 3 with IMovie editing.
Nancy

None of the one you list Nancy - you need to get a specialized piece of HDR software; Photomatix is one of the most popular ones out there. There are a number of other ones as well, but all the ones I can think of off hand are more expensive that it.

http://www.hdrsoft.com/

[arith]

Choose manual Richard, two stops apart. Try to point the camera at the brightest part on spot metering and get the bottom (leftmost) needle just to the right of the middle point. Actually I go 1+2/3 overexposed the brightest point with the leftmost needle but that can lead to problems, a sort of pink cast with Picturnaut. Once you've set exposure just steady yourself, focus on manual after setting it, stability off, breath out, squeeze and keep the button squeezed until it is finished.

[Timar]

I very much agree with what _GUI_ had to say about this topic. To illustrate his point, let me give an example. I challenge those of you who believe in the advantage of using 3 or more exposures to tell me which image has been rendered with Lightroom's "Merge to HDR" function from three exposures (-2, 0, +2) and which has been rendered without the intermediate exposure (-2, +2). I'm curious!

A


B

[SebastienL]

Hi,
A is (-2,0,+2) and B is (-2,+2), right ?
B seems to loose middle informations.

[Timar]

Well, before I give the answer could you tell me in which particular area on the images you think you have observed B to show less middle tone differntiation than A? I would then enlarge that area from both DNGs to present it as images B and C and letting you again guess which is which

[Black Pearl]

This is an ancient old thread so a simple 'You're right' or 'You're wrong' might be better - unless you have more specific information that will add to the overall info already presented.

[JohnRostron]

I think Timar has a fair point. I flipped between the two images in Lytebox and could not see any difference.

John

[Manfred M]

I think Timar has a fair point. I flipped between the two images in Lytebox and could not see any difference.

John

He has a point, for that example. Unfortunately, like many things in photography, the answer depends on a lot of things.

At a high level, many modern cameras have a dynamic range (at base ISO) of around 12 stops (when shooting raw). My Panasonic GX7 has a dynamic range of over 11 stops and the D800 is over 14 stops. I suspect that the dynamic range in the example images is not particularly high and in fact, I suspect that the entire dynamic range could be captured in a single shot, so I would be hard pressed to suggest either the +2 0 -2 or +2 -2 images are even necessary.

From what I have read, human vision handles some 20 stops from the deepest shadow details to the brightest highlights, but the bias tends to be towards the shadows. This suggests that given the right camera, we should be able to capture the entire range with just two exposures.

That is only half the story, as the second part of HDRI is mapping the completed image to a dynamic range that our display can handle. Again the literature is a bit unclear, but seems to indicate that modern computer displays can handle a dynamic range of 8 or 9 stops, so those 20 stops have to be mapped to something that our displays can handle.

While I did do some HDR images in the past, I find I rarely need to any more and in fact, I can generally coax out all the dynamic range I want in an image out of a single exposure.

[Timar]

This is an ancient old thread so a simple 'You're right' or 'You're wrong' might be better - unless you have more specific information that will add to the overall info already presented.

Well, that would be too simple and unrewarding, wouldn't it? Anyone could guess the right answer with a 50% probability. Just because my post Sebastien replied to is two months old (not what I would consider ancient, though), that doesn't mean that the information it contains would have become any less relevant or valid by now

Manfred, I agree with much of what you wrote but I'm affraid you are mixing up some entirely different issues in your response.

I suspect that the dynamic range in the example images is not particularly high and in fact, I suspect that the entire dynamic range could be captured in a single shot, so I would be hard pressed to suggest either the +2 0 -2 or +2 -2 images are even necessary.

Wrong. In contrast to many people who are into "HDR" (i.e. cheesy tone mapping effects they confuse with HDR), I know exactly what I am talking about and I can assure you that I never, ever do a HDR exposure merging if the dynamic range of the scene does not exceed the capabilities of the camera within a reasonable S/N ratio. Most times when I haven taken bracketed exposures I end up deciding to simply develop the original exposure and discard the other two (+2,-2). The above image (dark, unlit interor and bright, sunlit exterior) is actually a classic example of a high dynamic range scene. In fact, even the 4 EV difference hardly sufficed in this case to capture the entire dynamic range: the +2 EV shot still shows some noise if I would have decided to further lift the shadows (Canon, alas!) and the -2 shot still has a minor amount of clipping in the brightest areas of the sky.

I take your remark as a compliment, however, as your impression is exactly what well-done HDR processing should evoke: that of a perfectly natural looking image that doesn't scream "HDR" at you.

Anyway, all of you: please read my original post and Guillermo's post I was refereng to one more time. This was not about practical limitations and sensors' capabilites but about fundamental aspects of HDR image processing. Due to their physical design, all modern CMOS sensors have a perfectly flat tonal response curve for all three primary colors, from noise floor to saturation. This means that there is nothing(!) to be gained by using more than two different exposures, spread far enough apart to convey entire the dynamic range of the scene. With a sensor dynamic range of just 10 EV, one could theoretically capture a scene of 20 EV (in order to see all the details on the filament of a 100W tungsten lamp lightening the room for example ) by just taking two exposures +5 and -5 EV apart. The overlapping EV range between both exposures is in any case perfectly linear, so it doesn't make any sense whatsoever to take more than two exposures, under any circumstances, with any modern camera to create a perfect HDR rendering (well, of course there is an important practical benefit in taking three exposures: to have the original exposure available in case you don't want to do any HDR merging) All this talk about the purported benefits of three, four, five or more exposures spaced 1 EV or even less apart is nothing but mumbo-jumbo. Maybe people get different results from more exposures because the filters in Photomatix respond differently in some way - who knows (I know why I don't like Photomatix), but there is certainly no objective benefit in terms of additional tonal information.

This suggests that given the right camera, we should be able to capture the entire range with just two exposures.

Dito. Actually, given the right (ISO-less, high DR) camera, you can theroretically capture almost any dynamic range under the sun with a single exposure. With Sony's ISO-less sensors, you can already capture many scenes with a single exposure where you would need two exposures with other sensors to obtain a similar S/R ratio.

[Black Pearl]

Here's an idea - instead of resurrecting a thread that started seven years ago and hasn't been added to for nearly two and a half years with a challenge which (IMO) looks like it was simply placed to allow you to use this quote

I know exactly what I am talking about and I can assure you

why not share your experience and knowledge by starting a new thread.

I dabble with HDR while others here will have a varying range of knowledge and experience so we would all gain from someone like yourself creating a dedicated post with detailed in-depth explanations, superb examples of your HDR work and the reasonings behind the images you have captured and why you did so.

[Timar]

Well, that's what I do here, isn't it?

I really don't see I why I should have started a new thread to that purpose, because I specifically answered the question posed in the title of this thread. Moreover, there already is an excellent answer in this topic by Guillermo, which I thought wasn't given the appropriate attention, so I wanted to provide a practical example to illustrate and confirm his important point.

I wouldn't start a topic to present my HDR work because I actually haven't done much HDR work. I don't like the typical, oversaturated and heavily tone-mapped look of most "HDR" work, and real HDR merging is rarely necessary when shooting raw. My interest is more in the spirit of the CiC website - to understand the fundamentals of photography and explain it to others.

But let's not continue this off-topic discussion here

[Manfred M]

I wouldn't start a topic to present my HDR work because I actually haven't done much HDR work.

I suspected as much. Thank you for confirming this.

[Timar]

I suspected as much. Thank you for confirming this.

Why exactly did you suspect so if I may ask?

And Black Pearl and danK - why did you find this remark helpful but apparently not my attempt at an thorough explanation and illustration of the matter? Are you actually interested in the matter or just in arguments about it?

HDR unfortunately has become a contaminated term because most people confuse HDR with the kind of tonemapping effects Photomatix creates. In a way I have done a lot of HDR work (working with the curves and sliders in Lightroom to map the dynamic range of the RAW file into the 8 bit output), but that is probably true for most of us who shoot raw. I haven't done a lot of HDR exposure merging, because I was never satisfied with Photmatix or Photoshop's HDR tool and the only way for me to get a proper merge was to fiddle around with nerdy command line tools which I soon got tired of. Now that Lighroom offers a very decent HDR merge function I enjoy creating an occasional HDR image from two exposures. Here is another recent example:



This image contains a lot more midtones than the previous example but still, as to be expected, merging three exposures provided no benefit but in this case (handheld, living subject) a significant downside: a loss of sharpness and additional ghosting artefacts. It would have been much more laborious to get rid of these artefacts in Photoshop if I would have merged the image from three instead of two exposures. Work that would have been completely unnecessary.

By the way, it certainly doesn't take an awful lot of experience to understand the concepts behind HDR. Sometimes people even get confused because of their experience. History is full of examples of complicated theories and systems derived from experience (e.g. astrology, homoepathy, etc.) that we now know to rely on nothing but conviction and imagination. Try to tell anyone who believes in astrology that it is humbug - he or she will probably answer that you are ignorant because you lack the "experience" they have in this matter. Experience that is not founded on a solid scientific understanding can easily lead one astray, even when it comes to technical subjects such as photography. Some things purported by respected photographers working with "HDR" are simply wrong, despite all their experience - such as the notion that there would be an advantage of merging more than two exposures. I need little experience to know this to be wrong - just a basic understanding of the tonal response curve of digital camera sensors and one or two good examples to confirm the theory in practise.

[DanK]

Why exactly did you suspect so if I may ask?

And Black Pearl and danK - why did you find this remark helpful but apparently not my attempt at an thorough explanation and illustration of the matter? Are you actually interested in the matter or just in arguments about it?

HDR unfortunately has become a contaminated term because most people confuse HDR with the kind of tonemapping effects Photomatix creates

I don't confuse the two. Given that there have been extensive discussion of exactly this point on this forum, I'm guessing that a lot of folks here are not confused by this.

I haven't done a lot of HDR exposure merging, because I was never satisfied with Photmatix or Photoshop's HDR tool and the only way for me to get a proper merge was to fiddle around with nerdy command line tools which I soon got tired of.

I have to agree that Photoshop's HDR Pro is disappointing. However, your next sentence is not correct. It's in fact very easy to do. It's called exposure fusion. You can even get a plug-in for Lightroom that does it. Google it. You can also do it manually with layers.

Sometimes people even get confused because of their experience. History is full of examples of complicated theories and systems derived from experience (e.g. astrology, homoepathy, etc.) that we now know to rely on nothing but conviction and imagination.

Yes, but another term for experience is "empirical data."

I wouldn't start a topic to present my HDR work because I actually haven't done much HDR work. I don't like the typical, oversaturated and heavily tone-mapped look of most "HDR" work, .

You seem to be confusing the specifics of tone-mapping with the general issue of merging more than one exposure--by one means or another--to deal with excessive dynamic range. I haven't used HDR software for this for several years, other than a brief experiment or two (more of that experience stuff) with Lightroom's new HDR merge, precisely because I don't like tone mapping. I don't use Photomatix, but someone who does tells me it offers the option of exposure fusion.

and real HDR merging is rarely necessary when shooting raw

All depends on what you shoot. Simply false for me; I often shoot in forests, where the difference in luminance between the forest and the sky exceeds what my camera can handle. A few extra stops is usually sufficient, so I can usually get by fine by merging two images.

[Timar]

Somehow I get the impression that some poeple here feel offended by what I wrote. That is a complete mystery to me, as I certainly haven't written anything offensive to anybody in this forum

I don't confuse the two. Given that there have been extensive discussion of exactly this point on this forum, I'm guessing that a lot of folks here are not confused by this.

I didn't say that you confuse them, nor that people in this forum generally do. I said that most people (everywhere, on the internet) confuse them. People on this forum are usually much better informed that most people are.

I have to agree that Photoshop's HDR Pro is disappointing. However, your next sentence is not correct. It's in fact very easy to do. It's called exposure fusion. You can even get a plug-in for Lightroom that does it. Google it. You can also do it manually with layers.

My next sentence was "the only way for me to get a proper merge was to fiddle around with nerdy command line tools which I soon got tired of." What could possibly be not correct about that sentence? Firstly, it clearly refers to my subjective preference ("for me"), secondly, it is written in the past tense. Heck, I actually mentioned in the very next sentence that Lightroom now has a decent merge capability.

And no, it is not the same as blending layers, because just as Photomatix and HDR Pro it starts with developed images and not with the untouched RAW files. It is technically impossible to get an accurately linear tonal response when merging exposures this way. This is why Guillermo and others actually developed those command line tools that allow direct merging of RAW files.

Yes, but another term for experience is "empirical data."

In an ideal world yes. In the real world we all live in it is unfortunately more of a muddle of empirical data, convictions and self-amplifying preconceptions.

You seem to be confusing the specifics of tone-mapping with the general issue of merging more than one exposure--by one means or another--to deal with excessive dynamic range.

This is funny! Several times, I clearly differentiated between tone mapping effects and exposure merging, stating they are two entirely different things. Now you are saying I would confuse them? Oh, well... maybe I should have written it in bold and underlined.

All depends on what you shoot. Simply false for me; I often shoot in forests, where the difference in luminance between the forest and the sky exceeds what my camera can handle. A few extra stops is usually sufficient, so I can usually get by fine by merging two images.

Now you take what I stated out of context - the context was in this case my personal work. I don't encouter this situation very often (or maybe I just have a higher tolerance for noisy images, heh). Anyway, I guess I schould have added a "for me" to this sentence as well to make it perfectly clear. But even making things perfectly clear didn't seem to keep you from misinterpreting some of those other quotes, did it?

Btw. do the three of you have an agreement to always find one another posts helpful, regardless of what they contain?

[Polar01]

Timar just a quick question for now, could you possibly explain you thoughts behind working you raw files as 8-bit instead of 16 bit files. Now I know that most outside print labs want 8-bit jpeds files it that the reason you work your files as 8-bit.

Cheers: Allan

[Timar]

Timar just a quick question for now, could you possibly explain you thoughts behind working you raw files as 8-bit instead of 16 bit files. Now I know that most outside print labs want 8-bit jpeds files it that the reason you work your files as 8-bit.

I'm not sure if I understand your question correctly. If you refering to the JPEGs I have posted here, well, 8 bit is the standard bit depth for JPEG. Although Photoshop allows to export 16 bit JPEGs, those JPEGs are a "hack" and do not conform to the original standard, thus many applications may not render them correctly. Moreover, it is essentially a waste of webspace or bandwith because neither display devices nor printers will make use of the additional information.

If you want to know why it is necessary to blend raw files instead of developed JPEGs or TIFs (irrespectively whether they are 8, 16 or 32 bit files) to obtain accurate results: it has to do with how the color balance is applied over the raw files tonal curve. Essentially, the images have to be merged before any color balance is applied in order to avoid off-colors and transition artefacts.

[dem]

The file size of image B is a few Kb more, so presumably it has more information in it.

[Timar]

The file size of image B is a few Kb more, so presumably it has more information in it.

Yes, but what information? Of course you could assume that it has more tonal information and therefore the larger filesize. But it also works the other way around: the more exposures you blend, the more sharpness you tend to loose. The sharper an image, the more information it contains. This is also true for noise: the more noise, the more "information" it has for the compression algorithm.

[Polar01]

It was in your line, "In a way I have done a lot of HDR work (working with the curves and sliders in Lightroom to map the dynamic range of the RAW file into the 8 bit output)". This made it appear to me that you were processing the image in LR as 8-bit files, not 16-bit, I believe that you would only change to 8-bit for web or printing at an outside lab.
As to the question of White Balance, you will tell me if I am incorrect, I believe that it is the only thing that is applied to a raw file as it gives a reference point to start with in all raw converters. In the development Mode in LR, it gives: as shot, auto, Daylight, etc., etc. it would seem that if you merged two raw files or more without any processing as it sound to me, then the result of the merged image would have the colour balance that the camera was set to. Is that not so?

Cheers: Allan