0

Let's say I want to take a picture from inside an office where both the room and the view outside the window are equally bright (e.g. like this). The easiest way to pull off that shot is to shoot a series with exposure bracketing and stitch the result later in Lightroom or Photoshop. It's a straightforward processing especially when dealing with perfectly static subjects.

But how do I figure out the optimal number of shots and the exposure bracketing distance between them? Should I take 3, as my camera would by default in bracketing mode? 1 stop difference between each shot or aim for less? Is there a set of rules I can follow to achieve perfect HDR shots?

  • That's not so much 'stitching' as it is 'layering and masking'. – Michael C Dec 28 '17 at 23:32
1

There is no real "formula" that I know of that is going to give you a perfect HDR everytime. Fortunately there are some things you can do to increase your chances of a good HDR.

  1. Make sure your darkest shot has little to no clipping in the highlights.

  2. Make sure your brightest shot has no clipping in the shadows.

  3. Try to avoid blowing out the highlights completely (washed out levels of clipping can cause problems at the border between light and dark subjects in the scene)

  4. Try to avoid more than one stop between each bracketed shot. I've found that colors in the final result are better if you don't use large steps between bracketed shots.

In some instances 3 shots will work fine, in very wide dynamic range scenes (especially in real-estate photography) I'll sometimes take up to 7 bracketed photos. Usually I spend a few seconds at the beginning of a shoot getting an idea of how many shots I'll need per bracket. During post-processing I'll delete extra shots that got too dark or too bright before merging the remaining images

1

You need to:

  • Determine the overall range needed for your series of shots. What is the exposure value needed for the dimmest parts with the shadows exposed bright enough? What is the exposure value needed for the brightest highlights to not be blown out? What is the difference, in stops, between those two exposure values?
  • Decide how fine the gradations you need between the brightest and dimmest exposure. Some folks like to use a one-stop gradation. Others go as high as 2-3 stops between frames. A lot depends upon your camera's dynamic range (at the ISO you are using - for the same camera higher ISO settings have lower DR).

It is then a simple process of dividing the difference in exposure stops by the size of your gradations to determine the number of exposures needed.

An alternative method is to use one or two exposures for the scene outside the window, leave a larger gap between the window and the interior, and then take a few exposures to capture the full range of the interior. How many you need of each depends on the total brightness range of the scene out the window and the total brightness range of the interior.

Having said all of that, your linked example doesn't really look like "HDR", as narrowly defined by taking multiple images at different exposure values combined into a 32-bit floating point light map and then tone-mapped to a useable bit-depth (usually 8-bit or 10-bit). It looks more like separate images were combined using multiple layers and masks. This is also a form of 'High Dynamic Range Imaging', but it is not what a lot of people mean when they use the term "HDR."

1

Two shots are sufficient for most situations. Three shots will cover pretty much any scene that does not have the sun and a dark shadow in the same frame.

Recall that many modern cameras have above 14 EV dynamic-range. In theory two shots could cove 28 stops which is a contrast ratio of 268,000,000:1. In practice though you need some overlap since the process to match features depends on it, particularly with JPEG source images which have a non linear response. Given a 10 EV difference between frames, just two would cover 20 EV o a 1000000:1 contrast ratio.

All you therefore need is a an image captured with no clipped shadows and one with no clipped bright spots. They they are less than 20 EV apart, you can produce a good HDR image. Otherwise, add an image in the middle get something which overlaps with both the dark and bright image.

For simplicity, most automated HDR system (other than Nikon which only uses 2 images), capture 3. One is the standard exposure and one above an below it is captured with the expectation that contrast will fall within the combined range of all three (accounting for some overlap).

  • this answer photo.stackexchange.com/a/81794/32110 to a very related question seems to contradict your assertion. – ths Dec 29 '17 at 11:33
  • That answer is wrong. Most cameras take 3 to get a normal exposure and one for highlights and one for shadows. The HDR process gets more complex and with more images and has a greater change of introducing errors, so there is no reason to shoot more than you need, except when unsure. Myself, I take three, mostly because I like having the optimal single-exposure image in case movement makes the set of images impossible to merge. Even a little wind can make leaves on trees not mergeable. Adding more images to the problem just makes it worse since there will be leaves in more different positions. – Itai Dec 29 '17 at 14:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.