Sunset over the marsh

Happy Friday!

A few months ago I was on a family vacation to our favorite beach, and the surrounding area has some beautiful marshland. I had some time one evening to get out and catch the sun as it set. As I was shooting I had my camera set to fire off 5-shot brackets, giving me flexibility to do HDR in post if desired.

I have attached the middle exposure from a 5 exposure bracket sequence, as well as a .dng created from the full 5 exposure sequence via HDRMerge. I did my edits on this HDRMerge file and came up with the attached jpeg (I may go back and add an exposure layer to brighten the foreground a bit, but I’m mostly happy with the results as attached).

I always struggle with extremely wide dynamic range scenes like this, and would love to see what other people come up with, but more importantly, HOW they come up with it. For a long time enfuse has been my go-to for handling situations like this, but I found the results a bit to unnatural in this scene for my liking.

I look forward to seeing what you all can do with this.

HDRMerge and edits (darktable):
_MG_4354-4358.dng (11.5 MB)
_MG_4354-4358.dng.xmp (9.1 KB)

Original file(s):
_MG_4354.CR2 (24.7 MB)
PS I would be happy to upload the additional 4 .CR2 files from the 5 shot sequence, but didn’t want to overload PIXLS.US with excessive data. Not sure what the best practice is for that?

These files are licensed Creative Commons, By-Attribution, Share-Alike
(Creative Commons, By-Attribution, Share-Alike)


I think the unnatural feeling has to do with the saturation of the grass in the foreground. Our eyes are less sensitive to colors in low light situations.

Here is my take:
_MG_4354-4358.dng.xmp (5.3 KB)

The raw you posted has definite saturation clipping in the sun, which is expected if the center shot of the sequence is “exposed normally”. Here’s what that looks like with minimal processing:

I posted a screenshot to include the histogram. All that was done here can be seen in the Commands pane, top-left: 1) camera colorspace assignment, 2) subtract camera black of 2047, 3) multiply each channel by the camera white balance numbers, 1.965820,1.000000,1.701172, 4) demosaic, and then 5) set the black and white points at the limits of the data. You can readily see the pileup of pixels in the histogram spikes; each of the channel pileups has bee separated by the whitebalance multpilication. This causes the magenta cast in the image. For these areas, highlight reconstruction would attempt to make some tone out of the places where not all three channels are blown, but for the pixels occupying those spikes, all you can hope for is white. For these cases, I put a checkbox in my “data” blackwhitepoint option to scale to the minimum of the channel maximums; if I selected that for this image, the white point would then be set at the lowest (green) spike, the magenta would go white, and incidentally, the rest of the image would be a little brighter.

I’d be curious to see the two under-exposed raws, to determine if/where the camera stopped piling up data at its saturation point.

My take with your .dng file

darktable (2.7-git): _MG_4354-4358.dng.xmp (11.6 KB)


Using Darktable:

IMG_8994.CR2.xmp (4.4 KB)

1 Like

GIMP 2.10.10


Ma try using the DNG file.

_MG_4354-4358.dng.xmp (9.1 KB)


… and from the CR2 file:

_MG_4354.CR2.xmp (11.2 KB)

1 Like

That’s a really good point, thanks for the insight @CMOS!

Love seeing what everyone has done here! Looking forward to downloading the .xmp’s and seeing how you all accomplished your results!