The first shot is very nice! I find the others do have too narrow depth of field, which I don't find adds to these particular shots. Still a neat effect, but somehow (for my taste), less effective as photographs.
denoir wrote:
I should add that I'm not following that advice myself as I can't resist the temptation of shooting at max aperture, but I've seen over and over again that it's problematic. My biggest problem though are geometrical distortions and projections. I don't know the exact nodal points (i.e exit pupil position) for the lenses I use which becomes problematic when you shoot close ups. Furthermore when doing close up shots (as part of wide angle panos) the fact that you have spherical geometry rather than a planar focal plane becomes obvious. I've been thinking about if one could perhaps motorize a tilt/shift lens and hook it up to to the gigapan to get correction for the non-planar focal "plane". ...Show more →
Using a motorized solution to shift the lens and the back (you couldn't use a gigapan) would preserve the flat focal plane, but your pano size would be limited by the size of your image circle. For modern 35mm TS lenses, that's ~67mm, or about five shots.
You'd only get really huge panos if you used large format lenses on a view camera, in which case it would make more sense just to shoot film instead of dragging around the home-brew gigapan, a view camera, a DSLR back adapter, a DSLR, and two tripods and fifty pounds of sandbags to dampen the vibration from your motor unit.
Possibly overkill, or splitting hairs unnecessarily, but here is a comparison of image handling that I produced today, and thought it might be of interest regarding post processing of the images that make up the panoramas.
This set features four shots taken with a humble Canon G6 as individual 4MP fine jpg files, four shots, processed with LR3 and PSE8. All three of the variations below are from the same original four shots.
All shots were vertical, at the widest 7.2mm focal length (35mm -- FF equiv.), at f/4.0.
The first version is straight OOC from the original jpg, stitched on the Auto setting in PSE8. You'll notice it is stretched, most noticably on the sides, increasing the image magnification slightly (more evident in the 100% close-up further down the page). The stitch was imported into LR3, and slight modified for exposure and not much more.
The second version was OOC as well, but stitched in the Cylindrical setting in PSE8. Imported into LR3 for similar tweaks in PP, but slightly more extensive, including +13 on Manual Distortion and -3 on Vertical.
There isn't much difference in details between 1 and 2, but I'd say the look has been enhanced in 2, but compressing the sides, removing some of the added "barrel" distortion from combining the images, and the slightly richer color in 2.
The third version was imported into LR3 where each individual file was (mildly) processed as though it were for web reproduction, including some NR, and Sharpening to 20. Distortion was increased to +17.
It is always hard to predict how images will reproduce on web forums. In this case, all the images are lighter and sharper than on my LR3 and Photobucket screens. Please don't evaluate the images on whether the exposure or color matches. The point is whether or not pre- or post- image processing will provide the results you want.
Here are the 100% details of the above variations. The first shows the increased magnification (looks more zoomed-in) and the reduced fine details (noticably in the roses and roof vents) due to no added sharpening. The second shows the benefit of PP on the stitched image. Greatest improvement (although slightly over-sharpened) comes from the third version where the images are processed before stitching.
Keep in mind that PSE8 introduces exposure and sharpening controls of its own to "fix" and blend images. I'm sure other programs do the same, so pre-stitch PP is sort of a learning experience to determine how much is enought without overdoing it.
The net result is that it is possible to generate quality images via panoramas from modest low-MP cameras like the G6 (and even P&S), as well as the earlier DSLRs like the Canon 1D and Nikon D1X.
But . . . straight OOC jpgs are somewhat lacking, and to get the most out of them, they will benefit most from processing prior to stitching, although PP on the stitched image is not without merit. I usually shoot RAW in DSLR and make my mods first in LR3 before moving to the stitching, but have been curious how jpgs would do instead. Not bad at all!
I'm prepping-up for an Alaskan cruise where I hope to get some decent panos with the 60D and G6. At least I now know I have a fighting chance, if the G6 is all I feel like carrying around. And jpgs, yet! Who woulda thunk it?
Today, I was doing some testing/comparison between the D7000 (which I keep trying [unsuccessfully] to like) and the D1X (which I like, seemingly beyond explanation). During the course of the shots and the camera adjustments, I (naturally and compulsively) made a couple quick panos at the end with the D1X.
8 vertical shots with 35/1.8 DX, jpgs at f/4, ISO 125.
PS -- I see after posting that these again come out lighter (highlights blown) and sharpened, much more than what I placed on Photobucket. I'm trying to get a handle on where this elevated exposure and sharpening are coming in -- is it the FM site? Photobucket manipulating? It's particularly noticeable with jpgs, as they can't tolerate much PP with showing visible fringing and other artifacts. Anyone have any ideas how to better prepare images to get accurate "look" of the originals?
carstenw wrote:
Obik, neat shot! Is that a swimming pool? Where is one meant to get in?
Glad you like it.
Yeah, it's a pool. There are four entrances--two service entrances off to my left, a truck entrance off to my right, and the main entrance is on the other side of the pool, out of frame on the right side. I'm probably gonna shoot the opposite of this pano (one light in the middle, dark bamboo on either side) since I'm not entirely happy with the composition.
One more technical test type of shot. The background can be found in this thread with some guy making an 8x10 digital back for product photography. I wondered why he did not just use a scanning back and it was pointed out that strobes + scanning backs were a bad combination. So I was curious if one could work with a panorama shot + strobes for product photography.
I jerry-rigged a crude soft box with two strobes (the built in on the 7D and an external slave) with a metal camera case serving as a reflector. I should point out that I know nothing about setting up lighting so it's probably all wrong:
It woked... sort of. There were some stitching issues, but the strobe thing worked as expected. I would still not really recommend such a set up for product photography. Any close ups require very exact adjustments of the pivot of the lens as to minimize parallax, and you still can't avoid getting a spherical focal 'plane' (although that's not too problematic when you shoot at small apertures).
Luka -- Great images and very informative too! Thanks, as always.
Here are a few overcast panos from my recent trip up the Alaska panhandle via cruise ship. All with Canon 60D and 15-85/3.5-5.6 IS lens, except the Ketchikan hillside that used the Zeiss Contax 50/1.4.
8-frame panos from the Samsung Galaxy Tab 10.1. This Android tablet has a 3MP rear-facing camera which can stitch in either orientation (examples below are portrait mode for a less elongated result). The pano is never flawless, but it's easy, automated and fast.
The point is whether or not pre- or post- image processing will provide the results you want.
