how to shoot HDR

[Robert-Peter Westphal]

Hello,

I'm a little confused about how to exposure the single images for a resulting HDR :

In many cases / manuals  you find the information that the light image ( for holding details in the light parts ) the histogram should be exposed a little away ( round 1 or 1.5 stops) from the right border , and the dark image should be exposed a little away ( also 1 or 1.5 stops) from the left border.

So, why doesn't it make sense to use ETTR for the light image and exposure to the left for the dark one to keep the noiselevel as low as possible ?

Many thanks for future answers !

Robert

[BobFisher]

Not sure where you read that but it's not something I've ever seen as a description for HDR shooting.

In general, you want to bracket your images such that each part of the scene is 'properly' exposed in at least one of the brackets.  That is, that your darkest point is exposed to a middle value and your brightest point is exposed to a middle value.  That will give the software enough information to create a blended exposure that has the potential for a full tonal scale and low noise in darker/shadow areas. 

I've got a tutorial on my website that may be of interest.  It uses Photomatix for the HDR merging/tonemapping but the same principles apply to any HDR software.

[mcbroomf]

Yes, ETTR, ETTL and ETTM (Middle ... if you like or if the DR of the scene is vast).

But don't sweat it, take as many as you like then play with them all.

[Robert-Peter Westphal]

Hello,

thanks for the help !

Bob, your page is very interesting and a great help !

I think I will take a few pictures and will play with them to find the right way. I thought that there is a serious reason for not using ETTR, ETTL and ETTM, but it looks as if there is no.

Best wishes

Robert

[BartvanderWolf]

So, why doesn't it make sense to use ETTR for the light image and exposure to the left for the dark one to keep the noiselevel as low as possible ?

Hi Robert,

For both extremes of the dynamic range, it make sense to "ETTR" (to improve the S/R ratio, which allows for some serious postprocessing). Then bracket with 1 EV steps to cover the range. That will satisfy most (if not all) possible fusion/compositing techniques.

Cheers,
Bart

[BobFisher]

Don't agree.  If you're bracketing sufficiently you'll have ample signal and noise will be reduced as a result.  Where there can be problems is in cases of insufficient bracketing where there's still noise in darker areas.  If you follow the suggestion I made earlier, that is to bracket such that your darkest area is exposed to a middle value it shouldn't be an issue.  In that case there should be no need to ETTR your 0 exposure.

[BannerB]

Renee Besta wrote (is writing) so far the only article ive been able to follow on the subject. Talk about detail.

3 part release... parts 1 and 2 are out now.

HDR Tutorial LINK

[torger]

It depends on what the goal of your HDR picture is. Do you want that cartoonish photomatix grunge HDR look, or do you want to use HDR to overcome dynamic range limitations of the camera in difficult scenes, but still keep a natural look?

If the latter, my experience is that you most often need only two exposures, one normal with some clipped highlights, and then one extra exposure made darker to get the highlights section, you might still want to clip some in that to get a suitable natural-looking "glow" around the highlight. A typical example when I use HDR is an indoor shot when I want to keep some of the outdoor detail visible through a window, or a landscape scene with the moon when I want to keep some structure on the moon. I then don't use HDR software, but mask and blend the different exposures manually. The look of such photographs will be very natural, and the advantage of having to take very few exposures is that you get less problem of moving objects.

[Isaac]

... use HDR to overcome dynamic range limitations of the camera in difficult scenes, but still keep a natural look?

Even in the deep deep shade of old growth coastal redwood stands, sometimes that bright California sunshine splashes the bark and sometimes a patch of blue sky is visible above the canopy - then yes we can expose for the deep shade and then mask/blend just the burnt-out patches that need fix-up.

Or along a woodland margin where a "normal" exposure gives a fine balance of blue sky and fall foliage, except where that prominent standing dead tree reflects bright white - again, the tonal relationship is immediately apparent and we can use mask/blend to fix-up just those small burnt-out patches.

What hasn't always been obvious to me, and what's only becoming really obvious to me from comparing the result of in-camera HDR to the scene before me, is when that fix-up won't be effective.

Given an ordinary town scene with some sun and moderate cloud, and some trees, and some buildings; I might think to make a "normal" exposure and a second exposure for the sky. But if that "normal" exposure is metered for the whole scene then the brightness of the sky will likely leave other parts of the scene underexposed - it isn't just that the sky will be overexposed, so it isn't enough to blend in just the sky from that second exposure.

In those situations, rather than a "normal" exposure and one extra, we want at-least an underexposure and an overexposure. (I'm just reminding myself - y'all already understand all this.)

[Guillermo Luijk]

That is, that your darkest point is exposed to a middle value and your brightest point is exposed to a middle value.

I was a bit surprised to read this Bob, and I see you also say that in your tutorial. But having the scene's brightest point exposed to a middle value is unnecesary, it has zero contribution to capturing the scene's DR.

In an optimum bracketing for HDR, the least exposed shot should be a strictly ETTR'ed shot (max exposure right before RAW clipping). Anything less exposed than this doesn't add anything to the final image. This is basics about how a digital sensor (and common sense) works.

In fact, a good starting point to determine the ideal bracketing series is to achieve perfect ETTR, and from that exposure do additional higher exposure shots (at 2 or 3 stops intervals, less is unnecesary, more is risky) until the deepest shadows of interest get a sufficiently high exposure. For most real world situations, 3 shots cleverly spaced will suffice.

Regards

[JonathanRimmel]

I too am a bit lost when it comes to bracketing and exposing for HDR. It seems no matter how many or how few images I take, the highlights nearly always are blown out to some degree. I am not really sure how to avoid this other than to make sure the brightest in the series doesn't blow the highlights, but then I don't get the shadow detail. Perhaps my problem is post processing? I have tried both Photoshop CS5 HDR Pro and Photomatix. Both tend to blow out the highlights.

Any thoughts?

[hjulenissen]

I occasionally have "inversion rings" around the sun when shooting hand-held and using Photoshop to do the synthesis (using one of the "natural" presets and deghosting). Any advice on this? (see attachement. First is processed image, 2nd is middle exposure)

[digitaldog]

In fact, a good starting point to determine the ideal bracketing series is to achieve perfect ETTR, and from that exposure do additional higher exposure shots (at 2 or 3 stops intervals, less is unnecesary, more is risky) until the deepest shadows of interest get a sufficiently high exposure.

Seems totally reasonable if the goal is the best possible captured data for HDR or otherwise (why ETTR is useful).

[torger]

I occasionally have "inversion rings" around the sun when shooting hand-held and using Photoshop to do the synthesis (using one of the "natural" presets and deghosting). Any advice on this? (see attachement. First is processed image, 2nd is middle exposure)

Is the RAWs developed to TIF before HDR in photoshop? If so my guess is that the problem is that around such light source which is spread out due to clouds/haze there will be quite large areas where one or more but not all channels are clipped, which means that highlight reconstruction will take place in raw development. Highlight reconstruction aims to produce a good-looking result, but may not be that "true", for example you could get a darker highlight than it should be. When the HDR software then should combine these processed files it may not be able to match together the results. It is better to let the HDR software process the RAW file directly, then it can see exactly where clipping occurs and not take that into account when merging together the files.

If photoshop works directly on the raw files then I don't have a guess what the problem is though...

[BobFisher]

You know, we may actually agree on something GL.  Imagine that?   

I agree that in an perfect world what you're advocating would be correct.  However what I've done is build in a bit of an 'error factor'.  Figuring out that precise ETTR point can be difficult for some who aren't as familiar with the concept or haven't done the necessary testing with their cameras to figure out where that point is.  So my goal was to give folks enough room for error that they could still end up with a good result (i.e., no blown highlights, no blocked shadows).  You can never shoot too wide a bracket set but you can shoot to narrow a bracket set.  And while I realise that LuLa is the internet home of the photographic measurebators, it's entirely likely that there are people who read this thread who may not fit into that group; hence the repeating of the same philosophy here.  It's the difference between writing for a more general audience and writing for the hardcore measurebators.

[jeremypayne]

while I realise that LuLa is the internet home of the photographic measurebators

I think that's an aggressive and insulting word.  This is a community and you should show your fellow forum members a little more respect.

As someone who seems to be a stickler for semantics, you should mind your words more carefully.

[bjanes]

I think that's an aggressive and insulting word.  This is a community and you should show your fellow forum members a little more respect.

As someone who seems to be a stickler for semantics, you should mind your words more carefully.

The term measurebators is usually used by those who lack technical knowledge and expertise to mock those who do know about such things. It is a manifestation of Luddism. Those who use the term measurebation seem to be proud of their ignorance, while denying that scientific analysis is how progress is achieved in digital photography and other endeavors.

One of my favorite quotes from Lord Kelvin:

"In physical science the first essential step in the direction of learning any subject is to find principles of numerical reckoning and practicable methods for measuring some quality connected with it. I often say that when you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meagre and unsatisfactory kind; it may be the beginning of knowledge, but you have scarcely in your thoughts advanced to the state of Science, whatever the matter may be." [PLA, vol. 1, "Electrical Units of Measurement", 1883-05-03]"

Regards,

Bill 

[BobFisher]

I have a sufficient understanding of what the term connotes, Jeremy & Bill.  I use it the way I use it with that knowledge in mind. 

It's not that I don't appreciate the science. I'm intelligent enough that were I to take the time I could understand it with a great depth of understanding.  I do understand it well enough to be able to apply it in practical terms. 

The issue is, and it's particularly prevalent here on LuLa, that there is a not insignificant number of people who feel that in order to have any success as a photographer in the digital realm one has to have an absolute and complete knowledge of the science.  I reject that notion.  Just as I reject Kelvin's statement that everything has to be measured or measureable.  A balanced approach is a more practical approach.  Balancing the theory of the science with the practical needs of real world shooting conditions and the 'art' of photography.  I recognise, and have stated in other discussions, that there are times when being more pedantic about the science and the need for 'accuracy' (whatever definition one wants to put on that term) is important.  But there are times when it's less important too.  I also reject the notion that an absolute and complete knowledge of the science of imaging, of human vision, of colorimetry and such is required for one to be a highly competent and qualified photographer.  Further I'd suggest that an over-reliance on those things and getting too caught up in the pursuit of 'accuracy' or 'perfection' can be detrimental to one's development as a photographer. 

[jeremypayne]

I have a sufficient understanding of what the term connotes, Jeremy & Bill.  I use it the way I use it with that knowledge in mind. 

Then I guess you are just an asshole?

[BobFisher]

Then I guess you are just an asshole?

Think what you will.