Hello, im the creator of the image, ive just registered to give you a little more info since youve shown an intrest.
Equipment used was a canon 500D + 55-250IS + 18-55IS(kit), plus a cheap tilt/pan tripod.
Rendering and post processing was done with Autopano pro 2, intel E7200 (
[email protected]), 8GB ram + a lot of free space.
Taking the pictures was done manually on the tripod, with a total of about 7-800 pictures taken at various focal lengths. The bulk is done at 250mm, or 400mm@35mm equivilant. Lower focal lengths were taken over the whole scene incase i missed any bits(fortunately not this time, but has been needed other times), and to square off the edges. Aswell as 250mm, the scene was taken at 18, 55 and ~150mm. So many different lengths are used because while the software can join images at different focal lengths, it fails if the difference is too great, so intermediate levels are used to make sure it can join all the dots.
Stitching was originally going to be done with the smartblend rendering option, but that was going to take weeks to render if it did it successfully atall. Instead i used multiband which got the rendering done in a few days (requiring ~800GB free space for temporary and destination files), but results in a little more ghosting (see the cars on the roads for example).
Each focal length was rendered seperately then combined and merged manually. The software is not yet able to select the longest focal length to use to render in any given area, so the different focal lengths are put into layers, then lower res layers are erased where they overlap with lower res layers manually. The layers were then merged, and simple post processing done on it (levels, sharpening etc).
Editing the image quickly swamped the RAM, leaving photoshop grinding on the hard disk, which meant just opening the file took half an hour or more, and any actual editing was a few mouse strokes followed by lots of waiting.
Start to finish was about 2 weeks during free time (around FT job), plus 2 days to upload. Final image file is about 9GB after compression.
Strictly, panoramics should be done with a pano head which rotates the camera about its nodal point to avoid paralax errors, fortunately i managed to avoid too many of them in this picture since the majority of the scene is far away.
It can be done with a motorised rig, gigapan is an example of one, but others are also used based on telescope mounts and a little scripting. This saves personal effort, can make sure there are no holes missed, and facilitates stitching of the image, however such things are outside of my budget for now.
Regarding the suggestion a 5D was used, i think you may be refering to the 26GP dresden image:
http://www.dresden-26-gigapixels.com/dresden26GP . But as it has been pointed out, simple point and shoot images can be used to stitch together high res images. For example before i got the DSLR i used a bridge camera worth ~100GBP which was able to do things such as
http://lifeinmegapixels.com/location.php?location=stpeters . Its biggest limitation was shot to shot speed, but that is because it was quite low end, and half decent bridge camera is going to get you as far as you want realisticly.
Hope ive not completely bored you, ill try to keep an eye out here incase anyone has any further questions.
