HDR Question

[Paul in OKC]

Why take several photos in varying f-stops with the camera on a tripod and then combine them with Photomatix? Why not just take an existing photo, adjust the photo to different exposures, rename and save them, then combine the renamed files with Photomatix? What is the advantage (if any) of combining multiple ORIGINAL photos versus combining multiple VARIATIONS of the same photo? I addition, when Photoshop layers with different exposures are combined doesn't that do the same thing?

[OldClicker]

Each RAW image has an exposure range - let's say 11 stops we call -5 through +5. Your file for your output (monitor, print, etc.) has less range - let's say 7 stops - -3 to +3. When you compress the dynamic range of one image by making the 3 non-RAW images (-5 to +1, -3 to +3 and -1 to +5), you are only compressing the 11 stops that you have captured. That's fine if the scene you are photographing only has an 11 stop range. But if you are shooting sunlit snow through a window in a dark room, you may have a much wider dynamic range. Now if you take 3 separate 11-stop RAW images, say -10 to 0, -5 to +5 and 0 to +10, you have 21 stops for the HDR program to work with.

[Paul in OKC]

Not understanding. If I am on a tripod, shooting in RAW, and I shoot three photos (comp -10 and comp 0 and comp +10) this will provide a 21 stop range? What is the definition of an "11-stop RAW image"? I have been using compensation plus and minus for my bracketing. Am I using the wrong method? Thanks for your input.

[OldClicker]

The compensation bracket for the ranges I used above would be -5, 0 and +5.

There is an exposure range that your camera sensor/file can capture. A P&S shooting jpg is about 7-8 stops. New DSLRs are claiming 14+ stops for their RAW files. (The 11 stops I used was just an example.) This is the information that HDR software has to work with if you do a single image HDR. If the scene you are shooting has a dynamic range wider than the range of the sensor/file, then there will be bright parts and/or dark parts of the scene that you have not captured. You will then need to use multiple brighter/darker images to provide that information to the HDR software.

For each EV/stop compensation you add in a 3-image bracket (i.e. -1, 0, +1), you add 1 stop to the bright end and 1 stop to the dark end when the 3 files are combined.

Be sure what is doing the 'compensating' when you use exposure compensation to bracket. Shutter speed is probably the best thing to change if you have a solid support for the camera. If it uses aperture, each shot will have a different depth of field.

[Paul in OKC]

Thanks again. Excellent point about NOT using aperture to bracket. I must remember that.

[wfooshee]

Continuing, late to the party, I know, but your original question comes down to:

HDR is basically compression. The dynamic range of the scene is too large for the camera to capture, so why not manually compress a single image? Use the dodge and burn tools to reduce overexposed areas and bring up dark areas, right?

Well, the whole premise is based on the fact that the dynamic range of the scene is too large for the camera to capture! Those overly bright and overly dark areas of your single image have no detail to recover. White is white, period. Black is black, period. No dodging or burning will get anything out of there.

With multiple bracketed images you get the detail for those parts of the scene from the other images. The frame that is overall underexposed will have useful detail in the bright areas that is lost in the "correct" image. The frame that is overexposed will yield detail from the dark areas that are simply black in the "correct" image.

You end up with a "fake" image, where the dark areas are not as dark as they should be, and the bright areas are not as bright as they should be. You have compressed the dynamic range. That's what HDR is about.

Here's a shot I took of a friend's cat in a window. The cat is shadowed, and the background outside is blown. There is nothing to do for it but accept it, or plan for it by bracketing.


Here's what I could come up with by dodging and burning that image. The cat looks awful, I got some background, but the color shifted in some of it, especially that very bright area to the left, which is still blown white.


Since I shot it with HDR in mind, I bracketed, here are the 2-stops-under and 2-stops-over frames


Here's the final version. This is what Photoshop's Merge to HDR process could do, I don't even have a separate HDR program.


So your question, why can't I boost the darks and limit the brights from a single image basically shows a misunderstanding of dynamic range. The brights are gone because there's nothing there to recover. The darks are gone, again because there's nothing there to recover. The camera can only capture a certain level of light below the "middle" tone level of the image. Anything darker than that is simply.... dark. Similarly, it can only capture a certain amount of light above that "middle" level. Anything brighter than that is simply white.

By bracketing, you're shifting that "middle" tonal level, thus capturing more of the bright detail on the underexposed image, and more of the dark detail on the overexposed image. That stuff is simply not available from the correctly exposed image. It just ain't.

[Paul in OKC]

Thanks wfooshee. A special thanks for the effort you put into your reply. I really appreciate that.

[keroroslayer]

yeah it does paul

[adamagarthar]

Generally f8,f11 are the sharpest overall areas within a lens. F22 or F32 will give you more depth of field, but the center f-stops are generally the sharpest overall. I use f8 or f11 for almost all of my HDRs.

However, the larger the aperture, the larger the lens flare you will have also have them in your shot. Therefore, closing down to a smaller f-stop can help minimize this problem.