Playing with a 3-Exposure HDR Series

I thought about whether or not to post these in the HDR Challenge but decided to do this separately so as not to clog up the thread. In that thread I challenged folks to try different edits with a 3 shot exposure bracket series (see the challenge thread for the originals). One of the edits I'd mentioned in a tutorial was creating an exposure series from a single image, and I thought it would be interesting to see how much different the HDR created from one image would be from an HDR created from a 3-exposure set. So, I decided to do an edit in HDR Efex Pro 2 using the original images without using control points of any kind, and then saving that edit as a preset that could then be applied to any series of one or more exposure sets that I would send in. The idea being to see how much different they are. I then took it to some additional extremes, applying varying amounts of exposure compensation to one image and sending in sets of varying numbers, again, all processed with the same presets.

It's very difficult to see just how different they are in this format, and I'm going to try and come up with a way of displaying the differences. But while I'm working on that, I thought I'd let you take a look. I'll post my comments on each below in separate posts, each time comparing the photo to the original, traditional HDR in Image 1.


Image 1 - Original Three Image HDR (+/-2 EV bracket series)




Image 2 - Three Image HDR with +/-2 EV images created from virtual copies of the 0 image

With this image, I created 2 virtual copies of the center image in the series and applied a -2EV exposure adjustment to one and a +2EV exposure adjustment to the other in Lightroom. In the HDR Efex Pro 2 mapping screen I input those adjustment values in order override the metadata. Variations on this technique will be used to add'l images below, so I will not reexplain.




Image 3 - Seven Image HDR with +/-1 EV images created from virtual copies of the 0 image

As with Image 2, I created 6 virtual copies of the 0EV image to create a -3, -2, -1, 0, +1, +2, +3 image series. The rule of thumb is that you can always go +/-2EV, but can start losing details after that. I was interested to see how this compared to Image 2.




Image 4 - Single Image HDR Tone Mapping using only the 0 image

I sent only the 0 EV image to HDR Efex Pro 2 and applied the same edit.

Image 5 - Five Image HDR with +/-1 EV images created from virtual copies of the 0 image and processed with the +/-2 original images

This is essentially Image 1 with two add'l images sent in, a +/-1EV copy created from the center image in the series.




Image 6 - Seven Image HDR with +/-1 EV images created from virtual copies of the 0 image and a +1EV and a -1EV image created from the ends

Essentially Image 5 with two added images, a -1EV copy of the dark original, and a +1EV copy of the light original.




Image 7 - Nine Image HDR with +/-0.7 EV images created from virtual copies of the each image

I created 2 virtual copies of each of the original 3 and applied +/-0.7EV adjustments to each giving me 9 images in total.

Comparing Images 1-4

The first 4 images are very, very similar, with almost no discernible differences between them. Here's a side-by-side-by-side composite of the 4...




The differences are very faint and can really only be seen when you bounce back and forth between them in Lightroom. So, how can I show that physically? I decided to open them as Layers in Photoshop and use the Difference blending mode, which essentially compares the color content of each pixel by dividing one color value by the other and displaying the result. Identical layers would produce a completely black image. I chose to use Difference instead of Subtract because the differences were more apparent (when I used Subtract they were almost completely black and barely perceptible at this size). In each case the image is compared to Image 1.





As you can (hopefully) see, there is very little difference between the original and images 2 & 3, and slightly more in image 4. Images 2 & 3 are actually nearly identical, which I would almost expect since they were both multi-image HDR series sourced from the same single image. Image 4, however, starts to lose some saturation and color, which is likely due to the limitations of using a single image in a tone map. Still, it's very close and could likely be tweaked until it's nearly idential.

What this tells me is that, all other things being equal, a 3 image bracketed HDR produced from separate images doesn't necessarily give you any substantial advantage over a single, well exposed image provided that the image does not lack light information at either end of the histogram. In other words, just because my image looks like this...



...it doesn't mean that I have to shoot a bracket series and go HDR to make the photo I want. It should be noted, however, if I am blowing out one end or the other of my histogram on the metered exposure then I will need to bracket exposures if I want to capture all the available light.

When I was doing your challenge, I had more trouble with the foreground then with the sky.
When I took your single RAW file into LR, I had a much easier time with it.
I still love playing with HDR whether it needs it or not.:very_drunk:

Very interesting! Thanks for posting those examples!