From my trip to Venezia in August, I brought quite a number of images. I avoided the San Marco place as much as possible, and I found that one way to get a different view from Venezia, is to go to the island San Giorgio Maggiore, just across the Venetian lagoon. There’s a bell tower, from which one can see a great embracing view of the Serenissima. I thought it would make a great point of view for a wide panorama. But there started the difficulties.
First, from the top of the bell tower, you’re not allowed to take pictures with a tripod! The guard at the entrance told me that I could go up with my tripod, but that he was going to watch me. So, my panorama had to be taken hand held. Solution: try to be as steady as possible, a try to put two fingers about where the nodal point of the lens is to rotate the camera around.
Second, from the bell tower I could make a 360° panorama, but even if the most interesting part is the city of Venice itself, not so much the lagoon, I was still going to make a wide panorama of more than 180°:
And since I went there in the middle of the afternoon, the sun was already starting to get lower in the east, and consequently the left part of my panorama would be facing the sun, with a backlit foreground, while the right part of the panorama would be sunlit. Quite a wide in illumination. Solution: take EV bracketed shots (as raw).
Third, because of the orientation of the bell tower, one of the pilars would prevent me to take the wide panorama from a single point of view. Solution: take half of the pictures from the north-west facing side, and move to the north-east side for the remaining shots… hoping it would not make the stitching to problematic.
Anyways, I proceeded. Back home, I selected the shots that would make my panorama, and set everything flat in RawTherapee regarding exposure:
Okay, I didn’t optimize my exposure well enough before I took the shots. Not enough time for that, I didn’t want my significant other to get tired of waiting for me.
I exported the images and loaded them into Hugin:
- let Cpfind to find the control points automatically,
- told Hugin to clean the control points (those in the sky, and those with too low a correlation), and to fine-tune the remaining,
- optimized position and barrel distortion,
- optimized photometric information for HDR, fixed exposure,
- in the OpenGL preview, center and straighten the panorama,
- set the optimal canvas size (22131x3611 pixels)
- set the TIFF output to “Exposure fused from any arrangement” (as well as “Blended layers of similar exposure…” in case I want to try fusing the exposures with masks in Gimp)
- set the enblend options to “–primary-seam-generator=nft --blend-colorspace=IDENTITY”
- STITCH!
The enfused panorama (downsized):
The three blended exposures (downsized):
Okay, not bad at all, though the overall exposure is quite low, but I can correct that in post.
Long story short, I edited the enfused panorama in RT and Gimp (some tone mapping, some local lab improvements here and there, some noise reduction, final resizing and sharpening…) and here’s my current result (resized to 3000 pixels height):
[edit: the correct one]
I think I should try optimizing the exposures from the raw files in RawTherapee, and try again. What do you guys think?