I wanted to share my processing steps for this photo of Manhattan from Brooklyn (in the middle of the Empire Fulton Ferry Park lawn to be precise), which I think I’ll try to get printed - it might look good in my living room.
This started out with 13 frames shot in portrait orientation with my Fujifilm X-T2 and Samyang 12mm f/2.0, set to manual mode for shooting (ISO 1250, f/11, 2 sec).
It turned out that when trying to stitch all the 13 frames, Hugin had quite some troubles correcting the images, and introduced a lot of distortions and bad stitching errors. No matter what, I couldn’t get a good panorama. I thought that, maybe there’s too much overlap between the frames, So I decided to try selecting only 1 out of 2 (you can see the selected frames in the above screenshot from Rawtherapee’s file viewer).
I did only light processing of the raw files in Rawtherapee (highlights reconstruction in blend mode, a custom tone curve, bundled dcp profile for X-T20, custom White balance, cropping and 3-pass+fast demosaicing). The center frame looks like this in Rawtherapee’s editor view:
I loaded the exported 7 files (16-bit TIFFs) in Hugin, asked CPfind to: 1) find , 2) fine-tune and 3) clean the control points for me. Oddly, while my lens is a rectilinear 12mm lens, the only way to get a satisfying panorama was to set the lens type to Circular fisheye (Hugin set the focal length to 17.75mm by itself):
Then after trying several options, the best panorama I obtained was to make Hugin adjust custom parameters instead of just positions:
I also did some masking, to include/exclude some parts that spoiled the final panorama. For example I wanted to be sure that this man in the middle frame would appear in the final panorama (it’s a personal choice, at first I hated the fact he was there, but now I quite like it, he’s a photographed photographer):
The preview, with equirectangular projection, straightened and adjusted crop window, looked like this:
The resulting panorama is a 44 MP, 13547x3283 (which would make a 40x15 inches print at 330 ppi) image that looks quite good already, but I’ve been suggested to turn it into a contrasty black and white image that would look great printed. So I loaded it up in Gimp (converted to 32-bit float, linear light internally), fixed the top corners and cloned out some distracting things here and there. I also used the “illustration look” filter (actually, I used my own script fu, which was the base of the “illustration look” filter in G’MIC) to give some more “pop” to the image, but I applied it only on the parts I wanted to highlight (the skyline, the 2 bridges, the carousel, the river and fences).
Finally, I loaded the resulting 32-bit image back in Rawtherapee, and used the Agfa APX 25 LUT in “Film Simulation” module to render the image to black and white, turned “Local contrast” and “RL Deconvolution Sharpening” on and used a curve to brighten the light tones a bit more. “Et voilà !” (here’s a downscaled version, framed with G’MIC):
What do think?
Do like the presence of the man in the middle?
What could be improved?
Do you agree it would make a nice print?
Any advice for printing (print size, on which medium, in a frame or just bare…)
