HDR postprocessing

nd750_a4076-a4079.tif.pp3 (11.4 KB)
Well, here’s the pp3, but I didn’t upload ealier because it cannot be used with regular RT binaries, one has to compile it from the “newwavelet” branch to get the “Retinex in Wavelet” filter.

I used it because the regular “Retinex” filter in RT works only on raw images.

Er… that is the one who needs/wants GTK2?

Indeed, it’s based on Master (Gtk2) branch.

Many thanks for all your feedback, and the educative pp3 files.

Sadly I still find that the nik example has some clear advantages. One thing (which imho is quite important) which none of the RT-based examples thus far could reproduce, is adding depth to the features inside the gorge:

As you can see below:

the features inside the gorge throw shadows which are nice and black, so that the elements get a bit of depth. In the examples you have posted (as well as in mine), these areas still look very hazy and low in contrast. Is there any setting that would address this particular issue?

@HIRAM’s image seems to recreate this effect in a few places, but only on the edges of the cliffs, not on their faces. What setting does that? Is there any way to extend its reach towards the darker, lower-contrast areas?

I hate to tell you, but what you want takes more than a few clicks with one program. Here’s my shot at it. I put it through Luminance HDR using the Durand operator, then into Gimp using a variety of Luminance masks, level adjustments, curves adjustments, saturation adjustments, light painting layers, shadow painting layers…a lot of stuff. But this is what I came up with in 30 minutes. I could have spent more time, but the wife is already wondering when I’m going to be done “playing with pictures” :wink:

You could use RawTherapee to create the layers you want, then blend them together. Things like local contrast, saturation layers, etc. Just look up PatDavid’s blog on that. Another good one is Serge Ramelli on youtube. And don’t feel intimidated to try. I know the first time I saw how some of this was done I thought, “I give up!” But I didn’t, and I’m still trying.




Hmmm. Does RT have something like the “local contrast” tool in DarkTable? I find that local contrast adjustment is key to getting details to pop and thus adding depth. If you can’t try Darktable for some reason, perhaps someone here more versed in RawTherapee can advise how to achieve similar enhancement of microcontrast…

In the Shadows/Highlights area of the exposure tab there is a local contrast and radius adjustment.

1 Like

The TIFF does not appear to be 16-bit. Am I missing something?

A test with the new HDR merging algorithm in “newwavelet” branch:

I used 2 copies of the original tiff: one slightly darker, one much lighter, and merged the 2 in RT. It’s still experimental, and being heavy on my slow slow laptop I didn’t try too many settings. I also added some wavelet edge sharpening. Sorry I lost the pp3, but anyways it works only with the newwavelet branch.

how did you check the bit depth?

The highlights are still a bit burned, but in the shadows I’d say it’s starting to look better than what my friend did with NIK.

How do you obtain different versions of the same TIFF - and how do you merge them in RT?

I just opened the original tiff, and changed the exposure for highlights or shadows, and saved the 2 as jpegs. Then I used a version of RT I compiled yesterday from the experimental “newwavelet” branch. The author of the algorithm, Jacques Desmis, wrote a documentation on Rawpedia in the French section.
It’s still very experimental, and of course it would work much better if merging 2 raws ewposed for highlights and shadows, respectively.

Quick attempt using darktable. Shadows/highlights to compress the dynamic range. Local Contrast + Equalizer to bring back the detail. Velvia, Vibrance & Color Zones to get the colors a bit more vibrant. I prefer it over the nick variant, but given that I edited it I guess thats no surprise. :smiley:

nd750_a4076-a4079.tif.xmp (9.8 KB)


I couldn’t help giving this a quick go as well. Maybe 15 minutes spent almost entirely in wavelets (not newwavelets, just release 4.2.880 on win7). I’m not a fan of the super-local-contrasty look personally, but it seems you can get reasonably close? (Yes, there’s some fringing in the sky, but I was more concerned about the rocks and local contrast).

Oh, and the pp3: nd750_a4076-a4079.tif.pp3 (9.5 KB)

As you can see from the filename, this was applied against your .tif file. Also - I checked with imagemagick and it appears your output tif is 8bit, not 16bit? Is that right or did I mess up somewhere?

1 Like

It is 3 x 8 bit + alpha, not 16-bit per channel.

1 Like

Played around with it for about 15 minutes, but used additional programs.

Most image viewers, editors etc. can show this info, for instance:
ExifTool - BitsPerSample 8 8 8 8
IrfanView - 32 BitsPerPixel
XnView - # of bits 32
DigiKam - Bit depth 8 bpp (properties), Bits per Sample 8 8 8 8 (exif metadata)

Most obvious, the image size is too small for a 16-bit TIFF photo of this resolution.

Also, the (truncated) output from imagemagick identify:

Image: nd750_a4076-a4079.tif
Format: TIFF (Tagged Image File Format)
Mime type: image/tiff
Class: DirectClass
Geometry: 3399x6687+0+0
Resolution: 150x150
Print size: 22.66x44.58
Units: PixelsPerInch
Type: TrueColorAlpha
Base type: TrueColor
Endianess: MSB
Colorspace: sRGB
Depth: 8-bit
Channel depth:
red: 8-bit
green: 8-bit
blue: 8-bit
alpha: 1-bit
Channel statistics:
min: 0 (0)
max: 255 (1)
mean: 61.3699 (0.240666)
standard deviation: 77.5054 (0.303943)
kurtosis: 1.34831
skewness: 1.6937
min: 0 (0)
max: 255 (1)
mean: 61.044 (0.239388)
standard deviation: 79.5146 (0.311822)
kurtosis: 1.27884
skewness: 1.68432
min: 0 (0)
max: 255 (1)
mean: 56.9013 (0.223142)
standard deviation: 80.6685 (0.316347)
kurtosis: 1.55485
skewness: 1.79006
min: 0 (0)
max: 255 (1)
mean: 255 (1)
standard deviation: 0.0756422 (0.000296636)
kurtosis: 1.13646e+007
skewness: 3371.14

Version: ImageMagick 6.8.6-10 2013-09-17 Q16 http://www.imagemagick.org

1 Like

DenB, I like your rendition and would be interested in a brief overview of methods and tools used.
Thanks, James

Nice! I like this. Basically mirrors the approach I would have taken with Darktable too, although I tend to use the tone curve tool to mess with the tonal range vs the shadows/highlights tool…