p.1 #1 · Does stitching optically replicate a wider FOV?
So here's a question I have had for a while.
One way to get higher MP images without forking out for a super expensive high MP FF body is to stitch frames together.
I'm wondering though, if you took images at say 50mm and stitched them together, and the end result covered the FF FOV of say a 24mm lens, would the two images be the same? Assume that the MP count of the "full frame" shot and the stitched product are the same. In other words, assume that your stitching has exactly compensated for the enhanced resolution of the full frame sensor.
In particular, I know that the DOF changes with focal length. Does this mean that you would have to stop down more on the 50mm stitch shots in order to get the same FOV and DOF as a full-frame 24mm shot? How does that all work out? I know that there are two forces at play here, the DOF due to the focal length, and the DOF difference due to the full frame vs. crop body. Do the two cancel out, or is there some situations in which a stitched shot would have less DOF than a "one-shot" version?
Do any of you seasoned pano stitchers out there compensate for this in the field?
p.1 #2 · Does stitching optically replicate a wider FOV?
You will get less DOF with a stitch, given 1) equivalent f-stop and 2) to achieve same FOV. On the other hand, you will also get higher compression due to longer focal length. I view the latter as an advantage. You will also get more resolution and flexibility in composition.
p.1 #3 · Does stitching optically replicate a wider FOV?
shoenberg3 wrote:
You will get less DOF with a stitch, given 1) equivalent f-stop and 2) to achieve same FOV. On the other hand, you will also get higher compression due to longer focal length. I view the latter as an advantage. You will also get more resolution and flexibility in composition.
(1) is true, but (2) is not. The perspective of the image (what you describe as "compression") is a function of subject distance, not focal length. Thus, if one stitches together multiple images with a longer focal length to simulate a wider angle of view, the size relationship between subject and background will remain the same as if one shot a single photo using a shorter focal length from the same spot.
Two important things happen when stitching together images to form a composite with a wider angle of view, that are not mentioned above. The plane of sharpest focus is no longer planar, but curved, with the degree of curvature being proportional to the angle of view. A rectilinear lens that has been well-corrected for field curvature is actually not a constant focal length as a function of image height. Instead, the focal length of the lens is longer as the image point is more distant from the image center. This is especially important for wide-angle lenses, because one expects that when photographing a flat subject (like a wall), that the corners should be in focus at the same time the center is.
But if one stitches several images together using a longer focal length, the result will no longer match the above situation, because the method requires that you pick a spot to focus on, and lock it for all subsequent images.
The other thing that happens with this method is that one may experience difficulty matching the edges of individual images, especially if the DOF is shallow. The reason is because many lenses produce what is called "cat's-eye bokeh," a phenomenon where blur disks are football-shaped because they are produced off the optical axis. If this effect is pronounced, adjacent images will not align correctly because the blur "oval" is oriented in a sagittal direction relative to each image. Extra care with respect to lens choice, background matter, and stitching, must be taken to to ensure that this effect is avoided.
p.1 #4 · Does stitching optically replicate a wider FOV?
You are right. It is late, and I was thinking about it wrong.
Is the compression only due to the fact that longer FLs (even in a stitched situations) typcially capture less FOV?
I feel like there is more to do it than that.
p.1 #5 · Does stitching optically replicate a wider FOV?
I didn't think about the curving plane of focus. I suppose, depending on the situation, it could either work for or against your advantage. I suppose if you had a very accurate focus distance meter on your camera/lens, you could compensate for this as you shot the pano, the final result would look something kind of like a fresnel lens, where the individual elements' surfaces would sort of represent the focal plane for each shot, and each angled out a little more as you go increasingly off-center. However, I think also that as long as the effective FOV of the stitch is a reasonable amount, the amount of focal plane curvature is minimized.
I think I have experienced some degree of these effects in past panos. For example, sometimes there are areas in my pano where the image gets softer than the rest. But it's not consistent, like how one whole frame would have been blurred by missed focus or camera motion. Thinking back on this, it is possible that it was a DOF thing, and that the stitcher chose the frame with a different DOF rather than the sharpest part of each frame for the final pano.
I guess there is no perfect solution to a higher resolution/larger sensor body.
p.1 #6 · Does stitching optically replicate a wider FOV?
if you are doing a traditional panorama with sufficient DOF it won't matter.
Herb...
Ruahrc wrote:
I didn't think about the curving plane of focus. I suppose, depending on the situation, it could either work for or against your advantage.
p.1 #8 · Does stitching optically replicate a wider FOV?
but possibly insufficient DOF. i feel torn about shooting multirow panoramas with my Zeiss 50/2 Makro because of that. i'd like to go slightly longer but DOF is even worse and i don't feel like focus stacking then stitching. it's bad enough doing HDR panoramas.
Herb...
dasrocket wrote:
longer FL=less distortion=easier and more continuous stitching.
p.1 #9 · Does stitching optically replicate a wider FOV?
All really comes down to how much resolution you really need, and if your producing a crazy gigapixel file simply to say you've made a crazy gigapixel file.
I mean, yes, its pretty cool and all to say you've stacked 100+ images, and with my Gigapan head I've done that (took forever!) but realistically how useful is it ? You could never print something that large, and other than just showing a downrezed version and including the 100% crop for people to oooo and ahhh over, its all a bit silly when you consider the tradeoffs.
DOF issues, blending issues, sheer time spent etc
I'm all for pano's, but really think that given the current per frame resolution we've got today, that for "almost" any scene, a quick and simple 2 or 3 row pano is plenty. I dont really find myself doing more than 20 frames ever. Just starts to take too long and theres no additional return.
Getting 100+ mp for a smaller pano is plenty for a 40x60" print, and again, realistically how many images really are you going to do that large ? (they aren't cheap, and I'm out of wall space)
As such, I think there is a "just right" goldilocks type area for panos. Heck, if I had a PhaseOne P65+ back I'd probably just shoot a simple 3 shot pano and call it a day, having all the resolution I'd need.
HerbChong wrote:
but possibly insufficient DOF. i feel torn about shooting multirow panoramas with my Zeiss 50/2 Makro because of that. i'd like to go slightly longer but DOF is even worse and i don't feel like focus stacking then stitching. it's bad enough doing HDR panoramas.
p.1 #10 · Does stitching optically replicate a wider FOV?
HerbChong wrote:
but possibly insufficient DOF. i feel torn about shooting multirow panoramas with my Zeiss 50/2 Makro because of that. i'd like to go slightly longer but DOF is even worse and i don't feel like focus stacking then stitching. it's bad enough doing HDR panoramas.
Herb...
True enough.
I was envisioning small aps on a tripod with longer exposures rather than quicker pans.
The less distortion and overlap against the less DOF works better for images where the subject is consistently further away near infinity.
