Luminous Landscape Forum
Equipment & Techniques => Digital Cameras & Shooting Techniques => Topic started by: fike on March 26, 2010, 05:16:27 pm
-
I am considering adding a tilt shift lens to my repertoire. I have been reading about them and they sound ideal for the type of work I like to do--environmental panoramics in the forest with lots of complex foreground and soaring vertical lines with lots of crisp focus. That's what I like.
I am considering the Canon 17 mm F/4 L Tilt Shift.
The tutorials at http://www.cambridgeincolour.com/tutorials.htm (http://www.cambridgeincolour.com/tutorials.htm) have helped me understand tilt shift lenses better, but I have some questions.
Here is how I am considering using it.
1) I intend to mount my camera in the portrait position on a pano head (I think I may not really need the yaw control as much any more if I am content with the shift giving me enough additional vertical coverage).
2) I was going to make shift panos (left and right) without moving the rotating platform, basically three frames overlapping a bit in the middle frame to make smallish (roughly square) panos.
3) I figured I could use the pano head to spin the camera around the nodal point (do they work the same on tilt shift lenses?) and stitch two or more of these shifted panos together to make a larger panoramic image.
4) So far I have only used the shift mechanism to capture a wider field of view. I guess I could add-in the tilt to improve depth of field for my wilderness scene, from foreground to the horizon (speaking theoretically, here).
So presuming that everything I said above is generally practical and sensible, here are some questions:
If I am making a pano in the portrait orientation (as usual) and I use the shift mechanism vertically, to increase the pixels in the Y axis (with effectively 2.5 rows of photos), would I need to modify the tilt between the bottom row shift and the middle row and top row shift? I can imagine some pretty weird results in the foreground where the depth of field would be shallowest.
How would using a spherical panoramic head and moving the tilt shift camera in the pitch rotation change the behavior of the tilt shift lens? All the tutorials discuss generally use of the lens with the camera mounted level (for instructional purposes, I understand why they do this.) I can't even begin to imagine the possibilities of angling the focal plane while tilting the camera. The geometry is simple, but the practical application (if there is one) seems more complex.
-
I believe the physics questions implied might very well cause some heads to explode
-
I am considering adding a tilt shift lens to my repertoire. I have been reading about them and they sound ideal for the type of work I like to do--environmental panoramics in the forest with lots of complex foreground and soaring vertical lines with lots of crisp focus. That's what I like.
I am considering the Canon 17 mm F/4 L Tilt Shift.
The tutorials at http://www.cambridgeincolour.com/tutorials.htm (http://www.cambridgeincolour.com/tutorials.htm) have helped me understand tilt shift lenses better, but I have some questions.
Here is how I am considering using it.
1) I intend to mount my camera in the portrait position on a pano head (I think I may not really need the yaw control as much any more if I am content with the shift giving me enough additional vertical coverage).
2) I was going to make shift panos (left and right) without moving the rotating platform, basically three frames overlapping a bit in the middle frame to make smallish (roughly square) panos.
3) I figured I could use the pano head to spin the camera around the nodal point (do they work the same on tilt shift lenses?) and stitch two or more of these shifted panos together to make a larger panoramic image.
4) So far I have only used the shift mechanism to capture a wider field of view. I guess I could add-in the tilt to improve depth of field for my wilderness scene, from foreground to the horizon (speaking theoretically, here).
So presuming that everything I said above is generally practical and sensible, here are some questions:
If I am making a pano in the portrait orientation (as usual) and I use the shift mechanism vertically, to increase the pixels in the Y axis (with effectively 2.5 rows of photos), would I need to modify the tilt between the bottom row shift and the middle row and top row shift? I can imagine some pretty weird results in the foreground where the depth of field would be shallowest.
How would using a spherical panoramic head and moving the tilt shift camera in the pitch rotation change the behavior of the tilt shift lens? All the tutorials discuss generally use of the lens with the camera mounted level (for instructional purposes, I understand why they do this.) I can't even begin to imagine the possibilities of angling the focal plane while tilting the camera. The geometry is simple, but the practical application (if there is one) seems more complex.
Shifting using a T/S lens is rather pointless on a properly adjusted spherical pano head. That said, it can be done. You will need to shift the camera body the opposite direction of the lens shift in order to keep the entrance pupil in the same location to eliminate parallax errors.
Tilting is a lot more interesting, but will be very time consuming since you will need to reset it for each frame. With practice you may be able to shoot a pano in a reasonable amount of time.
The tilt will affect the whole image circle, so if you combine a tilt with a shift, you'll need to check the critical focus over the entire image circle. In other words you'll need to set the tilt, shift max up, adjust tilt, shift max down, adjust tilt, shift right, adjust tilt, shift left, adjust tilt and repeat until focus is correct for the entire image circle (or at least the part that you'll be using)
The tilt is the relationship between the lens and the sensor. How the camera/lens combination is oriented is not relevant. What will change dramatically is where the focus plane is located relative to the subject when the pano head is rotated.
-
Keep in mind a 17mm lens is awfully wide for stitching unless you're using a cropped-sensor camera. Like, freakishly wide, getting into the range where a rectilinear projection just doesn't look good at all.
Even on a cropped sensor camera, I think the 24mm would be a lot more useful for stitching, regardless of whether you're using shift or a rotating pano head. I used the 24 PC-E quite a bit for stitching when I was shooting cropped-format. Now on full-frame it's really wider than I would want to use for stitching in most cases.
Disregard the above if you're wanting to do those 360-VR spherical panos for the web (but you don't want a shift-lens for those, you want a fisheye).
One more thing about stitching: either use shift, or rotate the camera; but don't do both. As elf said it just doesn't really make sense to do both. In fact my experience is that you get worse results than if you hadn't shifted.
There are times when it makes sense to use tilts when doing a rotational pano, but at 17mm you should have plenty of DOF, I doubt you're going to need tilts in most situations.
-
For single row rectilinear, side-shift and stitch is better than pan-and-stitch (no distortion, no cropping off corners).
With a very short lens, you have so much depth of field, you would not need tilt to get it all in focus.
I had contemplated using tilt on the foreground of a typical "beech and cliff" (or lawn and house, church yard and church, carpet and wall...) picture, for single vertical row shift and stitch, to get the foreground in focus (the hope is that the tilt only affects the focus, so aligning the two shots requires no distortion)... I do not know if or how pano-stitching software would cope with combining these vertical rows horizontally for a multi-row shot.
For cylinder stitching, longer lenses work better.
What exactly are you contemplating doing, on what type of scene?
-
Here's a shot I made with the TS-E 17 in landscape orientation, shifting up and down to make 3 photos. There is almost no distortion with this lens, so shifting the lens instead of the sensor didn't really have any negative impact.
(http://farm3.static.flickr.com/2523/3914670157_cf29a71ed8_b.jpg) (http://www.flickr.com/photos/razortm/3914670157/)
-
Here's a shot I made with the TS-E 17 in landscape orientation, shifting up and down to make 3 photos. There is almost no distortion with this lens, so shifting the lens instead of the sensor didn't really have any negative impact.
...and you had stacks of DOF without tilting?,,, or could you, or did you, use tilt for the bottom picture to get the foreground in focus?
-
...and you had stacks of DOF without tilting?,,, or could you, or did you, use tilt for the bottom picture to get the foreground in focus?
I used tilt and infinity focus to get all of the water in focus, but I left the tilt and focus alone when I shifted for the top and bottom photos. When you shift up 12mm, the area in focus is shifting up 12mm as well, which is practically unnoticeable.
-
I used tilt and infinity focus to get all of the water in focus, but I left the tilt and focus alone when I shifted for the top and bottom photos. When you shift up 12mm, the area in focus is shifting up 12mm as well, which is practically unnoticeable.
Leaving the tilt setting saves any possible problems,,, but what about a floor/wall ceiling or three walls shot? Cylinder or sphere panos should be able to cope, but, if the original images are OOF, they the software could not fix it.
-
Here's a shot I made with the TS-E 17 in landscape orientation, shifting up and down to make 3 photos. There is almost no distortion with this lens, so shifting the lens instead of the sensor didn't really have any negative impact.
(http://farm3.static.flickr.com/2523/3914670157_cf29a71ed8_b.jpg) (http://www.flickr.com/photos/razortm/3914670157/)
This type of scene lends itself well to T/S. I don't think it will work as well with a typical forest scene where you want to get foreground objects as well the trees in focus. My first attempt using a T/S lens on a scene where trees were reflecting on a river was a complete bust. The desired plane of focus was really a reversed L so it was not possible to have it all in focus at once.
Tilt was set on the center image in this one, then both horizontal and vertical shifts were used.
(http://img.photobucket.com/albums/v649/etfrench/19504.jpg)
I think if you analyze your images carefully you can tell whether or not a T/S lens will be usefull to you. If you can get everything into a single plane of focus, the T/S lens will work very well for you. If not, I would look at focus stacking software like Zerene Stacker or TuFuse.
-
Interesting comments and responses. Thanks!
I am planning on using a cropped sensor camera for my work. So, it would behave something more like a 28mm lens. It is an interesting thought to try the 24mm version. The 24mm is a little cheaper and a little faster. One reason that I was considering the 17mm is that I typically walk-around with a 24-105 and I like to conserve resources. If I can add a new focal length to my quiver of lenses while adding tilt shift, that would be ideal. Adding a 24mm TS would be great, but slightly redundant.
I plan on using this for woodland scenes and stands of trees. I really like the effect of perfectly focused foreground along with perfectly focused trees shooting up in front of me. Someone mentioned the woodland scene having a an arrangement more like an "L-shape" than a plane. I need to think about that.
As for the suitability of panoramic rotation with a TS lens: I can see that pitch rotation (using a spherical head) would be impractical if you were planning on shifting up and down on a camera mounted in the portrait orientation. But if I do that, wouldn't I need to pivot the camera to move left and right. It seems to me that without moving the camera, you can shift your way into making a "plus-sign" shaped image where the corners that would be needed to make a square image would be left out. I am basing this on the fact that you can only shift in one direction at a time. Is that correct for the cannon lenses?
-
The 1.4x teleconverter on the 17mm also looks like a pretty decent way to increase its flexibility.
-
The 1.4x teleconverter on the 17mm also looks like a pretty decent way to increase its flexibility.
NO, dont do that if you care about image quality.
Works sort of ok with the 24TS II but not with the 17 at all.
TES
-
So my research has continued to progress. It seems that most of the responses here are discussing making shift panos while the film plane remains in a fixed position. As I understand it, this technique can generate a roughly square mosaic, and it DOES indeed appear that mixing in a tilt with this approach would be complicated and difficult (there better not be clouds in the sky because by the time I am done messing with the lens, they will move).
The approach I am considering is kind of a hybrid between panoramic stitching and shift stitching, and I am not sure if my reasoning is flawed. Let me try to explain my idea.
Let's say I am photographing a wall with framed art on it, and the floor in front of it that has a beautiful persian rug with fine detail. I want the rug and the framed art to be in perfect focus. I can make all sorts of compromises with higher apertures (don't get into focus stacking, it isn't always practical), but with a tilt shift lens, I should be able to get both in focus by making two images and merging them in post processing.
First, I would shift the camera down to take the lower half of the scene, including the floor with the carpet on it. I would tilt the lens to draw the focal plane inward towards the camera, rendering the floor in perfect focus. I would take the first image.
Second, I would reset the tilt to zero and then shift the lens upward, focus on the framed art and take the top half of the scene.
These two images would have different focus planes, but they would intersect at the base of the wall. I would then use photoshop to merge the two images at the base of the wall.
Now, what if pivoted the camera around the entrance pupil (nodal point) and completed the same procedure to the left and to the right of the original shifted panoramic. I would have 2 rows of 3 images. I could stitch the lower row and the upper row separately, and then stitch them together. I realize that this procedure couldn't cover a very wide angle because as the camera pivots, the focal plane would move too much, but otherwise, it seems like it might work (I realize that this might not work well with the 17mm, so perhaps the 24mm is preferable).
Do I understand the tilt shift lens capabilities correctly? Am I crazy?
-
First, I would shift the camera down to take the lower half of the scene, including the floor with the carpet on it. I would tilt the lens to draw the focal plane inward towards the camera, rendering the floor in perfect focus. I would take the first image.
Second, I would reset the tilt to zero and then shift the lens upward, focus on the framed art and take the top half of the scene.
These two images would have different focus planes, but they would intersect at the base of the wall. I would then use photoshop to merge the two images at the base of the wall.
I think that would work...
The effective focus distance would be different, so you would have to scale to get the two exposures lined up... but that would be child's play in this example, as you can cut off at a straight line at the bottom of the wall.
It might be more complicated if there was furniture in the shot.
-
With the panoramic capabilities offered by digital stitchers, and for the kind of scenes you are interested in (nature landscapes), I think using a regular lens with proper panorama stitching would be cheaper and could even produce better results.
Stitching images from a TS lens is very easy and will preserve the rectilinear projection of the lens. This is good, and even necessary, for arquitecture and rectilinear subjects in general.
But for nature subjects such as trees, yo could be more interested in a non-rectilinear projection that keeps the trees vertical, but compresses their rectilinear thickness in both extremes. With a rectilinear projection, if the angle of view is wide trees on the right and on the left will display much thicker than the trees in the middle, right in front of you. This distortion is easily minimised by using some other kind of projection (see a lot of them here: PTAssembler Projections (http://www.tawbaware.com/projections.htm)). And if any non-rectilinear projection is desired for your application, a TS lens doesn't make much sense because it will not save you the time and effort it would with rectilinear projections in mind.
Regards
-
When you tilt, the entrance pupil (some call it the nodal point) is going to move both horizontally and vertically, so if you want a parallax free stitch you'll need some method of moving the entrance pupil back to the horizontal and vertical pivot points of the spherical pano head. My setup allows this, but in practice I found that it was just too much of a headache to move it the precise amount required.
I would suggest that you shouldn't try to force fit a T/S lens into your current style of shooting, but should explore how to exploit the strengths of the T/S lens and see where it takes you. I took a brief tour through your site and I didn't see any images that could have been taken with a T/S lens using tilts.
-
When you tilt, the entrance pupil (some call it the nodal point) is going to move both horizontally and vertically, so if you want a parallax free stitch you'll need some method of moving the entrance pupil back to the horizontal and vertical pivot points of the spherical pano head. My setup allows this, but in practice I found that it was just too much of a headache to move it the precise amount required.
I would suggest that you shouldn't try to force fit a T/S lens into your current style of shooting, but should explore how to exploit the strengths of the T/S lens and see where it takes you. I took a brief tour through your site and I didn't see any images that could have been taken with a T/S lens using tilts.
That's some of the specific feedback I was hoping for. So you say that the entrance pupil moves with a tilt. Does it also move with a shift? Entrance pupil calibration of a T/S lens would be a PITA.
My thought on the use of shifts in my style is that I could tilt the bottom of the focal plane slightly in towards the camera to give the perception of a slightly greater depth of field from foreground to tree trunks through the woodland shots that I like so much. perhaps a shot like this could benefit from a tilt, http://marcshaffer.net/fine-art-panoramics...land-creek.html (http://marcshaffer.net/fine-art-panoramics-for-sale/highland-creek.html) or this one, http://marcshaffer.net/fine-art-panoramics...mmer-creek.html (http://marcshaffer.net/fine-art-panoramics-for-sale/dry-summer-creek.html) .
As for force fitting T/S into my panoramic style, you may be right that I am coming at T/S from a particular viewpoint. It seems to me that using a T/S with panoramic techniques would require a substantial simplification of the shooting methods to avoid going nuts with fine, detailed adjustments, as you mentioned.
-
That's some of the specific feedback I was hoping for. So you say that the entrance pupil moves with a tilt. Does it also move with a shift? Entrance pupil calibration of a T/S lens would be a PITA.
My thought on the use of shifts in my style is that I could tilt the bottom of the focal plane slightly in towards the camera to give the perception of a slightly greater depth of field from foreground to tree trunks through the woodland shots that I like so much. perhaps a shot like this could benefit from a tilt, http://marcshaffer.net/fine-art-panoramics...land-creek.html (http://marcshaffer.net/fine-art-panoramics-for-sale/highland-creek.html) or this one, http://marcshaffer.net/fine-art-panoramics...mmer-creek.html (http://marcshaffer.net/fine-art-panoramics-for-sale/dry-summer-creek.html) .
As for force fitting T/S into my panoramic style, you may be right that I am coming at T/S from a particular viewpoint. It seems to me that using a T/S with panoramic techniques would require a substantial simplification of the shooting methods to avoid going nuts with fine, detailed adjustments, as you mentioned.
Those two could easily be taken with T/S, but I'm not sure you would gain much from what you have already. I think focus stacking would probably be easier and quicker for subjects that are static.
I have a Hartblei 80mm Superrotator, and shifting and tilting moves the entrance pupil. The entrance pupil just moves the amount of the shift, but tilts are moving it at an angle. The precision that you rotate around the entrance pupil determines how much parallax you'll have. Most landscape images can tolerate quite a bit of parallax error if you want to spend the time in PP blending the seams. Most of the stitching software can handle handheld shots, so it may not be as big a deal as it sounds. My setup, http://luminous-landscape.com/forum/index....rical+pano+head (http://luminous-landscape.com/forum/index.php?showtopic=34499&hl=Spherical+pano+head) , has the camera mounted on an L bracket to allow moving it horizontally and vertically to correct for shifts and tilts. I also added stops for maximum shifts so I could do it quickly. After I did that, I realized I had half of a view camera, so I added the bellows and front movements. With Nikon lens, it will focus on infinity. In practice, the T/S lens is quite a bit slower to set up for each frame than just using the pano head to rotate. I'm sure people that have used LF cameras with movements will be able to set up and shoot quite a bit quicker and are probably quite puzzled about why I'm saying it's so tedious to do Checking the focus accuracy on a DSLR is very hard to do compared to an LF. My camera has 10X liveview, but can't scroll the view without exiting the 10X mode. Going through the process of adding a tilt, verifying the focus at the corners, and repeating until the focus is perfect is a very time consuming task with a DSLR.
I should add that there is probably nothing more fun in photography than manipulating the focus plane. It justifies the expense. I wouild recommend looking at the new Hartblei SuperRotators with the Zeiss optics. I think they more than equal the Canon and Nikon T/S lens optically and they have a lot more functionality.
-
Those two could easily be taken with T/S, but I'm not sure you would gain much from what you have already. I think focus stacking would probably be easier and quicker for subjects that are static.
I haven't had good luck with focus stacking for woodland shots. Light filtering through gently moving leaves is impossible to deal with for HDR and focus stacking because you are usually left with halos.
Most landscape images can tolerate quite a bit of parallax error if you want to spend the time in PP blending the seams. Most of the stitching software can handle handheld shots, so it may not be as big a deal as it sounds. My setup, http://luminous-landscape.com/forum/index....rical+pano+head (http://luminous-landscape.com/forum/index.php?showtopic=34499&hl=Spherical+pano+head) , has the camera mounted on an L bracket to allow moving it horizontally and vertically to correct for shifts and tilts. ...
You are confirming my suspicion that the small movement of the entrance pupil in a shift or tilt will not be likely to create large problems for a cylindrical stitch, particularly where landscape scenes are concerned. Buildings and interior architecture may be a bit different.
This statement from The Digital Picture Review of 24mm T/S (http://www.the-digital-picture.com/Reviews/Canon-TS-E-24mm-f-3.5-L-II-Tilt-Shift-Lens-Review.aspx) confused me a bit though.
[!--quoteo(post=0:date=:name=The Digital Picture)--][div class=\'quotetop\']QUOTE (The Digital Picture)[div class=\'quotemain\'][!--quotec--]Ideally, for this technique to work best, the lens position (not the camera position) should remain constant for these shots - to avoid parallax in foreground objects. There are various means of adjusting the camera the 12 degrees in both side-to-side directions (vertical is more difficult) to accomplish this, but using an Arca-Swiss compatible quick-release clamp and plate make this task relatively easy. Just measure and apply a couple of small indicators on the clamp and plate. Then move the camera the opposite direction of the shift movement - so that the lens remains in its place.[/quote]
Do people do this with an L bracket and a focusing rail turned sideways like you would for a stereo photographic setup? I have a bunch of RRS pano head stuff, but I don't immediately have a way to move the camera vertically to accommodate vertical shifts. Is this amount of precision needed when you can use a stitcher to clean up minor misalignments?
-
I went back through my years of panoramic work, and I tried to figure out what focal length I used most frequently. Generally speaking I have used focal lengths between 40mm and 70mm for most of my best work.
For this reason, I figured that the 17mm and the 24mm TS-E lenses would not fit well with my desire to meld some of my pano techniques with new tilt/shift techniques. I actually decided that the 45mm f/2.8 TS-E would be a better place to begin learning. At half the cost and 85% the quality-level, I figure it is a better fit for the experimentation I want to do.
I realize that it doesn't have the flexibility of tilting on an axis parallel to the shift without some minor modifications that aren't appropriate for the field, but that doesn't really bother me too much. I think I might want to make them parallel, but that should be something I can tackle myself.
-
that sounds about rite - mine are 50 to 100 on full frame
i think there's a thread on the 45 as a landscape lens not too long ago
why not rent a 24 and 45 from lensrentals.com for a week before making the investment?
-
I went back through my years of panoramic work, and I tried to figure out what focal length I used most frequently. Generally speaking I have used focal lengths between 40mm and 70mm for most of my best work.
The focal length only determines the number of tiles for a given field of view, and thus the resollution per tile. Perspective is determined by your shooting position.
For this reason, I figured that the 17mm and the 24mm TS-E lenses would not fit well with my desire to meld some of my pano techniques with new tilt/shift techniques. I actually decided that the 45mm f/2.8 TS-E would be a better place to begin learning. At half the cost and 85% the quality-level, I figure it is a better fit for the experimentation I want to do.
I have both (and the 90mm), and use them depending on the desired output resolution. The 45mm is fine, but does exhibit chromatic aberration (which is well adjustable in the better Raw converters). The 24mm II is optically superior, plus the added in-field rotation capability between cross- and parallel tilt/shift, but the 45mm will still out-resolve it per tile due to it's larger magnification factor. The 24mm does exhibit extremely low distortion and flare, which may be important depending on the situation.
The 17mm seems to be a good solution in cramped situations, with restricted stitching opportunities, only.
Cheers,
Bart
-
The focal length only determines the number of tiles for a given field of view, and thus the resollution per tile. Perspective is determined by your shooting position.
Cheers,
Bart
Yeah. I am generally shooting to make prints around 24"x40" at 300 dpi. That ends up being around two rows of five to seven images shot in portrait orientation on my cropped sensor camera.
Depending on the overlap, with a 17mm lens that is well over 200 degrees of rotation--with the distortion of that wide a field of view, it would not be ideal for realistic rectilinear panoramic prints. It is just too wide in the print size I am targeting.
At 24mm it would still be near 180 degrees...very wide for every day panoramic work.
45mm, on the other hand gets more around a 90 or 100 degree field of view with six or seven portrait frames.
With the 45mm lens I can always go wider by shooting more frames. I can't do the same with the wide angle versions.
The 90mm looks intriguing. I see that it has very high IQ, but all these lenses are at least quite good, so I figure for my first foray into TS, 45mm seems the best middle-ground (at $1,200 I don't want to call it a compromise).
-
Yeah. I am generally shooting to make prints around 24"x40" at 300 dpi. That ends up being around two rows of five to seven images shot in portrait orientation on my cropped sensor camera.
The amount of overlap, and crop versus full 24x36mm frame, are important parameters. I shoot with a Canon 1Ds3, but the desired output resolution should determine your choice.
Depending on the overlap, [...] 45mm, on the other hand gets more around a 90 or 100 degree field of view with six or seven portrait frames.
These are convenience versus quality factors that only you can decide on. Some jobs require the ability to shoot fast, others require to shoot with an enlargement capability/potential (when in doubt, go for resolution, i.e. longer focal length, if subject movement allows).
With the 45mm lens I can always go wider by shooting more frames. I can't do the same with the wide angle versions.
The 90mm looks intriguing. I see that it has very high IQ, but all these lenses are at least quite good, so I figure for my first foray into TS, 45mm seems the best middle-ground (at $1,200 I don't want to call it a compromise).
This (http://www.xs4all.nl/~bvdwolf/temp/OPF/615-617_S1.jpg) is a recent 244 degree horizontal FOV, cylindrical projection, TS-E 24mm II one (light changes fast around dawn time and the FOV angle and light quality gradient is extreme, hence the choice for 8 tiles with the 24mm). Here (http://www.xs4all.nl/~bvdwolf/temp/OPF/3041-3045_S.jpg) is another 5-tile one with the 45mm I shot that same morning as the fog was clearing, but the FOV was determined by my choice of magnification capability and requirements for a FOV angle.
Cheers,
Bart
-
Here's three horizontal shots on a 1dsMKIII with a 17mm T/S. Cropped a little off the bottom. I just shifted and did not make any adjustment for lens height. Align layers in Ps was all I used. This gives a very very wide image. I've done a bunch of images stitched like this and they seem to work like a charm every time.
-
NO, dont do that if you care about image quality.
Works sort of ok with the 24TS II but not with the 17 at all.
TES
Really? Most confused as I have been using the 17mm + 1.4 Kenko for quite a while and had really good results. In the real world, I can't tell the difference between that combo and my regular 24mm TS-E II. Certainly no clients have complained and the prints have been pretty large.
Mike.
-
I DO NOT shift the lens when shooting panoramas whether they are single row or multi row. Whether they are made hand held or on a panohead. As a matter of fact, Autopano will often produce unacceptably bad results when working on shifted images. Other stitchers may work ok with them. Autopano has the ability to correct for convergence from tilting up and also can make a rectilinear image from a panoramic shot (within reason.) And of course you can adjust for convergence in post if stitching with another program that doesn't do this. Shifting a lens on a pano head will throw it out of whack if you move it off of the rotation axis. Thus buying a shift lens for this application would be a waste.
Here is an example of several 17mm images stitched together in order to get this entire building in this "impossibly wide" rectilinear picture. This was shot hand held using the 17 TSE unshifted and a 5DII. The resulting file is a 195 megabyte RGB 8 bit tif.
The last image is using the 17mm TSE lens fully shifted and shows what that would take in from the same spot. So I'd guess that my stitched image is about what one would get from a 10mm or wider shift lens
-
Shifting a lens on a pano head will throw it out of whack if you move it off of the rotation axis. Thus buying a shift lens for this application would be a waste.
Just a question; in your example, would shifting UP to correct verticals then using the pano to get the width not work? No movement from nodal point and full benefit of corrected perspective distortion rather than relying on software.
I am only just starting to work with panos so am looking for all the help possible
Mike
-
Shifting a lens on a pano head will throw it out of whack if you move it off of the rotation axis. Thus buying a shift lens for this application would be a waste.
Hi Alan,
That's not necessarily the case. Sure, Autopano has problems, Photoshop probably as well. However, programs like PTAssembler (and I assume PTGUI) and Hugin can correct for a shifted lens with the 'd' and 'e' parameters. While they can be found by the optimizer, the solution becomes more difficult to find automatically. Therefore is helps to make a note of the amount of shift and apply it as calculated pixel dimensions. For example, a 5mm shift on a 6.4 micron pitch sensor is 781.25 pixels. Works like a charm.
Just a question; in your example, would shifting UP to correct verticals then using the pano to get the width not work? No movement from nodal point and full benefit of corrected perspective distortion rather than relying on software.
Mike,
With the right software, not a real problem, just some inconvenience due to a more difficult solution finding to what's called an ill posed problem (multiple solutions possible, only 1 of which is correct).
Cheers,
Bart
-
I guess it is pretty counter intuitive to me that one would even consider using a pano head and then rotate vertically or horiontally with the lens off the rotational axis. But perhaps this can be compensated for. In any case unless there is an object that is near to the camera, it won't matter as these kinds of stitched images can even be produced hand held.
-
I guess it is pretty counter intuitive to me that one would even consider using a pano head and then rotate vertically or horiontally with the lens off the rotational axis. But perhaps this can be compensated for. In any case unless there is an object that is near to the camera, it won't matter as these kinds of stitched images can even be produced hand held.
True - stitching software is getting better by the month.
I guess my feeling was, if you are doing a simple horizontal pano, getting the verticals 100% correct with a TS lens first then panning should be better than the s/w trying to fix verticals later. However, I have yet to play with this in the real world so all theory for me
-
I guess it is pretty counter intuitive to me that one would even consider using a pano head and then rotate vertically or horiontally with the lens off the rotational axis.
Hi Alan,
The lens is not off the rotational axis (the one that matters). A typical scenario is with the camera in portrait orientation, shifted up to capture height with perfect verticals. Not shifting up would put the horizon in the image center, not a very useful setting in most cases. Then a horizontal rotation through the entrance pupil will allow to get whatever width is required, while maintaining perfect verticals.
Cheers,
Bart
-
Hi Alan,
The lens is not off the rotational axis (the one that matters). A typical scenario is with the camera in portrait orientation, shifted up to capture height with perfect verticals. Not shifting up would put the horizon in the image center, not a very useful setting in most cases. Then a horizontal rotation through the entrance pupil will allow to get whatever width is required, while maintaining perfect verticals.
Cheers,
Bart
One could just as likely shifted the lens to the left or to the right. It's the same principle - rotate left or right on one axis for stitching vs. tilting up and down and stitching on the other. Yes if you really are determined to use a shift and then stitch, you can probably make it work somehow. I used to do this too, but sometimes Autopano would make a very weird horizon when I worked from lenses that were shifted upwards and sometimes it would do a good job. It is probably much harder to get it to stitch from shorter lenses that are shifted a lot.
Regarding the stitched image I posted above, I also have 5 shots that were made with the lens perpendicular to the ground and shifted all the way up. I tried stitching those images in CS4 and Autopano and couldn't get it to work. So even though I own 7 TS-E and PC lenses, I am not tempted at this point to use the shift when stitching. If I need more resolution, I can simply use a longer lens and composite more images. This is why I have a two axis pano-head that lets me shoot several rows.
-
I DO NOT shift the lens when shooting panoramas whether they are single row or multi row. Whether they are made hand held or on a panohead. As a matter of fact, Autopano will often produce unacceptably bad results when working on shifted images. Other stitchers may work ok with them. Autopano has the ability to correct for convergence from tilting up and also can make a rectilinear image from a panoramic shot (within reason.) And of course you can adjust for convergence in post if stitching with another program that doesn't do this. Shifting a lens on a pano head will throw it out of whack if you move it off of the rotation axis. Thus buying a shift lens for this application would be a waste.
Here is an example of several 17mm images stitched together in order to get this entire building in this "impossibly wide" rectilinear picture. This was shot hand held using the 17 TSE unshifted and a 5DII. The resulting file is a 195 megabyte RGB 8 bit tif.
The last image is using the 17mm TSE lens fully shifted and shows what that would take in from the same spot. So I'd guess that my stitched image is about what one would get from a 10mm or wider shift lens
This is the oddest thing I have ever seen. I have never seen "pics" so small. Makes me think there is something wrong with them. Everything looks good at that size on the web. What are you hiding?
-
I ended up getting the 24mm TS-E f/3.5 II. It is easily the sharpest lens I have ever worked with. Post-processing sharpening makes already sharp images amazing.
I have been able to do some hybridized tilt-shift-pano work.
I have used a spherical head (operating mostly in a cylindrical mode) to make two row panos where the lower row is tilted 2 degrees and shifted -5mm while the upper row is untilted and shifted +10mm. This technique is challenging to work with because the bottom row focus plane (tilted) needs to intersect the top row focal plane (untilted) in a visually logical place. This isn't always easy. I'll try to remember to post some examples.
PTGUI was able to handle these rotated and shifted TS stitches without problems.
-
Do people do this with an L bracket and a focusing rail turned sideways like you would for a stereo photographic setup? I have a bunch of RRS pano head stuff, but I don't immediately have a way to move the camera vertically to accommodate vertical shifts. Is this amount of precision needed when you can use a stitcher to clean up minor misalignments?
You simply drop your ballhead clamp in the slot to a vertical alignment. Assuming you are using an Arca Swiss camera plate, the camera can now be in portrait mode and move vertically for vertical flat stitch panos. If you have an L bracket on your camera you can keep the camera in landscape orientation while moving it vertically for flat stitched panos or vertically rotate it for traditional rotated panos.
-
I ended up getting the 24mm TS-E f/3.5 II. It is easily the sharpest lens I have ever worked with. Post-processing sharpening makes already sharp images amazing.
I have been able to do some hybridized tilt-shift-pano work.
I have used a spherical head (operating mostly in a cylindrical mode) to make two row panos where the lower row is tilted 2 degrees and shifted -5mm while the upper row is untilted and shifted +10mm. This technique is challenging to work with because the bottom row focus plane (tilted) needs to intersect the top row focal plane (untilted) in a visually logical place. This isn't always easy. I'll try to remember to post some examples.
PTGUI was able to handle these rotated and shifted TS stitches without problems.
Do you have any images to share?
-
Do you have any images to share?
As my learning curve with this lens proceeds, I should have some new work this week. I've taken hundreds of experimental shots, but nothing yet that I am particularly proud to show.
I have to say, this is not an easy lens to master. I have spent a lot of time trying to work it into my shooting style, and its lack of auto focus is a problem for me. Manually focusing well, is not as easy as it may appear. I generally use the focus, recompose, shoot method with the AF decoupled from the shutter button. I also tend to move around a lot when I shoot. for that reason, a normal zoom tends to be easier for my style. If I find that the lens stays home too often, I may sell it and get a 16-35 F/2.8 L II as a more flexible alternative.
With all that said, one thing I know about a new lens (outside of the normal range) is that you don't instantly create excellent quality (technically and artistically) with any new lens. After I got my 100-400 I was initially disappointed because I didn't know how to get the most out of it. Some times it takes time to master new equipment, particularly something as complex as a tilt shift.
-
May I be so bold as to introduce a question that I've been unable to resolve so far. I'm still waiting for NPS to answer it. I hope it's not too far off-topic or ill-informed for this august company (pardon the pun, too...)
The last Nikon Pro mag had a piece on the Agnos JMBS - "Jumbo MultiBigShoot" (sic) For anyone who hasn't seen the article it's a lens clamp for the Nikon PCE range of lenses enabling the camera body to be shifted rather than the lens barrel. This - I assume - is to enable stitching without parallax errors or the distortions seen when stitching cylindrical panos even when rotating the lens about the entrance pupil. As such it seems to me like a pretty good idea. I believe it's about Eu 300.
I've previously read a few threads here and there about stitching shots from T/S lenses. As someone who does quite a bit of modest interior shooting (usually VR panos) I'm intrigued by this idea despite having no experience with T/S lenses, much less View Cameras. I currently use the 14-24 a lot for (non-VR) interiors and I'd be interested to know the answers to a couple of questions if anyone would be kind enough to enlighten me.
1) How would the horizontal fields of view compare between the 14mm end of my existing lens and a 3-shot stitch using the Nikon 24mm T/S in landscape orientation on the Agnos clamp?
2) Is the Agnos device intended, as I imagine, primarily to eliminate parallax errors? From what I've read on LL stitching shifted shots is quite a common technique (hitherto shifting lens, and thus entrance pupil, not body I assume) Hence the Agnos clamp?
3) Anyone got any other comments on this device?
Roy
-
1) How would the horizontal fields of view compare between the 14mm end of my existing lens and a 3-shot stitch using the Nikon 24mm T/S in landscape orientation on the Agnos clamp?
2) Is the Agnos device intended, as I imagine, primarily to eliminate parallax errors? From what I've read on LL stitching shifted shots is quite a common technique (hitherto shifting lens, and thus entrance pupil, not body I assume) Hence the Agnos clamp?
3) Anyone got any other comments on this device?
Roy
That device will probably do for you in a bit easier way what I do with RRS focusing rail & slider:
Build a View Camera with your Canon 5D (http://www.ronnynilsen.com/Essays/Essay/ViewCamera/)
Ronny
-
That device will probably do for you in a bit easier way what I do with RRS focusing rail & slider:
Build a View Camera with your Canon 5D (http://www.ronnynilsen.com/Essays/Essay/ViewCamera/)
Ronny
Maybe I am missing something in your tutorial, but I do a couple of dozen or so flat stitches, without a rail on Canon T/S lenses, every week for clients and simply stitch them in PS-CS4 perfectly. I own and used to use a rail to correct the small parallax shift, but since CS3 have found the rail and parallax correction unnecessary as the stitch programs can easily handle the issue. The only problem I have seen in a couple of years is where there is not enough detail in the overlap areas (like a blank wall) to lock on to.
-
Maybe I am missing something in your tutorial, but I do a couple of dozen or so flat stitches, without a rail on Canon T/S lenses, every week for clients and simply stitch them in PS-CS4 perfectly. I own and used to use a rail to correct the small parallax shift, but since CS3 have found the rail and parallax correction unnecessary as the stitch programs can easily handle the issue. The only problem I have seen in a couple of years is where there is not enough detail in the overlap areas (like a blank wall) to lock on to.
Hi Kirk,
That's possible if you don't have overlapping close foreground and background detail. Whenever you change the position of the entrance pupul, you will get a shift in perspective between foreground and background features. The blending operation of a good stitcher can often hide the flaws, especially with a liberal amount of overlap, but not always. The only way to really avoid surprises is to keep the entrance pupil stationary, thus keeping the perspective the same.
Cheers,
Bart
-
Hi Kirk,
That's possible if you don't have overlapping close foreground and background detail. Whenever you change the position of the entrance pupul, you will get a shift in perspective between foreground and background features. The blending operation of a good stitcher can often hide the flaws, especially with a liberal amount of overlap, but not always. The only way to really avoid surprises is to keep the entrance pupil stationary, thus keeping the perspective the same.
Cheers,
Bart
Sorry WADR but that is nonsense these days with flat stitches from T/S lenses. I shoot architecture and interiors professionally and very often have "overlapping close foreground and background detail". I haven't had to use my rail in a couple of years, not with CS3 or 4..
-
Maybe I am missing something in your tutorial, but I do a couple of dozen or so flat stitches, without a rail on Canon T/S lenses, every week for clients and simply stitch them in PS-CS4 perfectly. I own and used to use a rail to correct the small parallax shift, but since CS3 have found the rail and parallax correction unnecessary as the stitch programs can easily handle the issue. The only problem I have seen in a couple of years is where there is not enough detail in the overlap areas (like a blank wall) to lock on to.
Yes, for the most part things will go well doing what you do with modern SW, I to do that from time to time. But
the advantage of doing it with "a back shift" is:
1) I'm guaranteed a perfect stitch result back home.
2) Stitching becomes very easy, took me perhaps 3 min. to do it manually in CS2, and
stitching is very fast in CS4 with "reposition only" as the SW don't need to do any correction.
Besides it takes me perhaps only 10 seconds extra time during capture compared to not using
the rails.
Ronny
-
Ronnie, you may be right somehow, but I just haven't seen it. Last week I was in Las Vegas for a 2 day shoot. I did 10 flat stitches (about 1 in 6 shots) with a 24 T/S II, some of them hand held from the top of a ladder, something I do regularly when I can't use a tripod. All of them worked perfectly as expected. I had a rail in my accessorie bag in the truck that never came out. Believe me, I have very demanding clients, on high profile out-of-town shoots, if there was any doubt whether these simple flat stitches would work, I would have used the rail, but as always they stitched perfectly. If they hadn't worked, I would have been flying back to redo them-not exactly a cost effective way of doing business.
-
I second this 100%. Since using CS4 I quit using my rail. To say the truth, I don't undertand how PS does it. I tested Photomerge with a 5D2 and TS17 in interiors and exteriors using multiple planes with intersecting beams or furniture parts where you can see through. Probably a true nightmare before CS4 and rail support.
Everytime Photomerge came with a perfect stitch. I don't flatten so I can see where the software cuts each frame. Intrinsic! A true artwork of computing science. If the software distorts anything, I can't tell @100%. Of course, as I learned to trust more CS4 Photomerge the more stitching I've been doing. It's so easy!. This is one the main reasons I decided to strenghten my EOS system and forget about digital medium format. At least for the time being. Maybe Adobe owns Canon stock
Eduardo
Ronnie, you may be right somehow, but I just haven't seen it. Last week I was in Las Vegas for a 2 day shoot. I did 10 flat stitches (about 1 in 6 shots) with a 24 T/S II, some of them hand held from the top of a ladder, something I do regularly when I can't use a tripod. All of them worked perfectly as expected. I had a rail in my accessorie bag in the truck that never came out. Believe me, I have very demanding clients, on high profile out-of-town shoots, if there was any doubt whether these simple flat stitches would work, I would have used the rail, but as always they stitched perfectly. If they hadn't worked, I would have been flying back to redo them-not exactly a cost effective way of doing business.
-
Well, I guess I shouldn't be too surprised that there are no direct answers to the questions I posed. Any discussion of stitching, whether involving TS lenses or not, usually evolves into a dispute about the necessity (or not) of using a "rail" - or a pano head. Personally I use PTGuiPro to stitch, which I think is generally accepted as being at the upper end of stitching software: I tried stitching with CS3 and it was useless - I've yet to upgrade to CS4. When stitching VR panos with a fish-eye lens it's absolutely essential to set up the panhead (mine's currently a barely serviceable NN3) so that rotation takes place around the npp point. If it's too far out - and I mean by a few mm - the pano will not stitch accurately. I made dozens before I finally got this right. But that's applicable only to my own experience of the peculiarities of VR panos.
I've shot "regular" panos off a panhead or else hand-held with a subject without much in the way of coherent foreground. So I know that you can stitch acceptably with a degree of parallax error between the donor shots. But I'm a bit baffled by the contributor above who recounts shooting hand held panos with a TS lens. I assume that it was being used as a conventional lens, unshifted?
Can I try again to get some opinions about the problem as I see it? Let's put aside for the moment the issue of parallax error. If I shoot a series of shots using a wa lens in an interior and stitch them, I get - unavoidably - the characteristic curvature of horizontals above and below the "equator". So the combination of a 24 mm TS lens and the Agnos device looked to me like a perfect solution. No curvature, no parallax errors, just an effectively bigger sensor with a wide aspect ratio (assuming landscape orientation). I'm not so concerned about resolution so my main interest is in the effective FOV I could achieve overlapping 3 shots with this lens, by comparison with my existing 14-24 @ 14mm. The absence of the volume anamorphosis effect that 14mm exhibits would also be a big plus.
I would have thought that this requirement would be pretty common amongst those of us shooting in confined spaces using 35 mm gear. I confess to being at the low end of the food-chain in this respect; even a single TS lens is a big and possibly uneconomic outlay for me. Of course if it improves my results it may prove justified.
Roy
-
I believe that this is a good example of what this thread is discussing. Attached are the three images shot with a Canon 17mmT/S, then combined in CS5's Photomerge program. The result is the fourth image, if you'd like try it yourself. This shot was taken as part of a glorified walkthru of a house I'll be shooting in the near future, the architect had originally hired another photographer who wasn't able to produce an image like this (not the right equipment, etc) so now I'm involved. I feel the process works quite well, while I'd love to be able to shift the back and never move the lens, I can find no fault with the final image. Jim
[attachment=23531:6_30_10_...2_adjust.jpg]
[attachment=23532:6_30_10_...8_adjust.jpg]
[attachment=23533:6_30_10_...4_adjust.jpg]
[attachment=23534:Chang_Pa...ma_1_dc2.jpg]
-
Can I try again to get some opinions about the problem as I see it? Let's put aside for the moment the issue of parallax error. If I shoot a series of shots using a wa lens in an interior and stitch them, I get - unavoidably - the characteristic curvature of horizontals above and below the "equator".
Hi Roy,
No you don't get a wavy horizon, provided(!) that you have your panning axis perfectly aligned. You can still get rid of the 'wave' if you point your horizon (pitch parameter) at a higher or lower position than you are doing when the 'wave' strikes. The resulting stitch will have a somewhat curved top and bottom edge, so you'll lose some FOV to cropping, unless you can fill in the missing parts with CS5s content aware fill (unlikely to be successful for most interiors). However much you try to get your rotation platform leveled, there will almost unavoidably be a small amount of error, but the panostitcher should be able to figure this out automatically if you can set some vertical line controlpoint pairs.
So the combination of a 24 mm TS lens and the Agnos device looked to me like a perfect solution. No curvature, no parallax errors, just an effectively bigger sensor with a wide aspect ratio (assuming landscape orientation). I'm not so concerned about resolution so my main interest is in the effective FOV I could achieve overlapping 3 shots with this lens, by comparison with my existing 14-24 @ 14mm. The absence of the volume anamorphosis effect that 14mm exhibits would also be a big plus.
The Argos device looks like a useful addition, I just don't know how it does in practice (e.g. can you work quickly between lens+camera shifts in opposing directions). In my experience, and within the image quality limitations of a large image circle near the edges, I use the horizontal flat-stitch most. For that it was very easy to equip my RRS setup with a few simple additions to make a very compact solution that also works fast (e.g. when trying to avoid cloud-cover to change light conditions between tiles).
(http://www.xs4all.nl/~bvdwolf/temp/OPF/RRS_OrthographicStitch.jpg)
I set my camera with the left shifted lens in the clamp and shift it to the right stop, I then shift the lens to the right and the camera to the left stop. With the camera in landscape mode that's all it takes, 2 exposures and 4 shits, quite fast. The stops are pre-set at the correct width (corresponding with 2x the amount of lens shift + clamp width). The result is very easy to composite in Photoshop with mostly a relative horizontal shift (and perhas a bit vertical to compensate for a rotated sensor). This is impossible to accomplish in Photoshop under all but the more favorable circumstances without the correction for entrance pupil shifts, especially in scenario's like interiors with furniture close to the camera. A good blender may get away with it under favorable circumstances, but I'm not going to second guess physics and hope Photoshop can manage (because when it can't, you're in big trouble). Prevention is still better than having to cure.
I would have thought that this requirement would be pretty common amongst those of us shooting in confined spaces using 35 mm gear. I confess to being at the low end of the food-chain in this respect; even a single TS lens is a big and possibly uneconomic outlay for me. Of course if it improves my results it may prove justified.
The requirement is there, and so is the solution.
Cheers,
Bart
-
I feel the process works quite well, while I'd love to be able to shift the back and never move the lens, I can find no fault with the final image.g]
Hi Jim,
this is a good example of where fully automatic blending usually works (nice image by the way, given the shooting circumstances). There is little foreground detail that overlaps with the background, so a good blend goes a long way with little difficulty. As long as you can avoid the foreground/background compositions, thing will probably work out fine. But if you want total freedom in composition and perspective/viewpoint, a small preparation can save your day (and night rest ;-)).
Cheers,
Bart
-
You guys may be right that parallax is rarely a problem with today's stitching software, but the problem is not completely eliminated so I'd rather take the safe approach and shift the camera in the QR clamp to avoid it in those rare cases where it could be a problem. It takes all of 5 seconds to guarantee I won't have to worry about parallax.
I don't carry a device to allow camera shifting for vertical's, but I don't shoot those as often and in my experience parallax is less of an issue. But for horizontals it _does_ come up on occasion, at least for me shooting landscapes. I do think a special sliding rail is overkill though. I have 10mm marked on my QR clamp and I just slide the camera over in the clamp. I may not get the shift exact down to .1mm, but I don't think that level of precision is necessary (unless maybe shooting macros).
I will say, that while I prefer PTGui over Photoshop for spherical and cylindrical panos, the "Reposition Images" option in Photoshop is preferred for shift-stitching, because you don't want to use a projection for flat stitching. You can get around this in PTGui by manually setting the focal lenth to something like 1000mm, but I just don't see the point when Photoshop has a tailor-made option.
-
........ the "Reposition Images" option in Photoshop is preferred for shift-stitching, because you don't want to use a projection for flat stitching. You can get around this in PTGui by manually setting the focal lenth to something like 1000mm, but I just don't see the point when Photoshop has a tailor-made option.
Just to further describe my very complex workflow , which always follows the KISS principle. I find the "automatic" setting in CS4 handles about 95% of flat stitches perfectly. FWIW I also own AutoPano Pro and PTGUI, but never need them for simple flat stitches.
My approach is incredibly simple. I make basic adjustments in Lightroom (images synced) to the DNG files highlight both and open from Lightroom>Photo>Edit In>Merge to Panorama in Photoshop (or just left click on one of the selected images and go to EDIT IN>. At the merge window in PS, I leave it on automatic and wallah! The whole process from conception to finished merged image hardly takes more time than a single image.
-
Disclaimer.-
Last night I was stitching an image shot of a bedroom with the 17TS and no rail. The bedroom was a very tiny space of no more of 11X11 feet. There was this double bed with its closest angle at about 3.5 feet away from my lens. As you imagine, there was extreme distortion to the shape of the bed. With the 3 stitched images I managed to cover the entire bedroom from my position.
Photomerge couldn't do it. The parallax error was too much to handle. I had to do it manually and selectevely cutting my layers. I was lucky I could do it this way. I'm sure that with the rail this wouldn't have happenned.
So, for shots with objects very close to the lens or constrained spaces, I wil better use my rail.
Regards
Eduardo
-
Thank you all for the valuable comments on my questions.
Never having used a TS lens or a view camera, this remains to some extent in the realms of the imagination. I may try hiring one to try although I doubt that anyone has the Agnos clamp available for hire. A pity there's nothing wider than 24 mm in the Nikon range too.
On the "curvature" issue, I'm a bit taken aback. Of course I know that moving the entire stitched image along the pitch axis will impact this - I have done it often enough to tweak landscape panos shot at longer focal lengths. From recollection though, to do this for even a two-shot stitch using focal lengths in the range <24mm is impossible to adjust without having to crop to an unusable extent.
It will be interesting to see if exposure in the Nikon Pro mag generates enough sales for Agnos for some informal reviews of their abominably named clamp. Any volunteers?
Roy
-
http://www.nital.it/experience/jumbo-mbs.php#05 (http://www.nital.it/experience/jumbo-mbs.php#05)
Link above (in Italian but online translation makes a reasonable job of it) has some detailed information and comments by users on the JumboMultiBigShoot. The name's even offensive to type...
Roy
-
Now we need one for Canon EOS TS's
Eduardo
http://www.nital.it/experience/jumbo-mbs.php#05 (http://www.nital.it/experience/jumbo-mbs.php#05)
Link above (in Italian but online translation makes a reasonable job of it) has some detailed information and comments by users on the JumboMultiBigShoot. The name's even offensive to type...
Roy
-
Now we need one for Canon EOS TS's
Eduardo
I don't think I would like to let the tilt/shift part of the lens carry the weight of my 5DII and battery grip and L-bracket. I'll stick to my rails. ;)
Ronny
-
Re; Ronny Nilsen's comment. There's an interesting analysis by Joseph W. on the relevant thread at the DPreview site:
http://forums.dpreview.com/forums/readflat.asp?forum=1030&message=36022465
I wonder why Nikon are heavily advertising a device that may overstress an expensive lens? Definitely a legal problem for them if these comments prove to be well-founded.
Roy
-
I've thought about that. That's the beauty of the 5 series, it's low weight. Why would you bother with a battery grip and bracket with such a setup?
If it's for wireless file transfer, I'd rather wire to save weight.
Regards
Eduardo
I don't think I would like to let the tilt/shift part of the lens carry the weight of my 5DII and battery grip and L-bracket. I'll stick to my rails. ;)
Ronny
-
I've thought about that. That's the beauty of the 5 series, it's low weight. Why would you bother with a battery grip and bracket with such a setup?
If it's for wireless file transfer, I'd rather wire to save weight.
Regards
Eduardo
I want the battery grip because I like to use use a hand strap, and for for using the camera in portrait orientation with all the buttons in the right place. And the L-bracket for use with other lenses. But even without this I would hesitate to put the weight of a 5D on a t/s lens.
Ronny
-
I use a Sony a850, RRS L bracket and nodal slide, and a Mirex T/S adapter with Mamiya M645 lenses for my landscape work. I've been very impressed with the image IQ of the Mamiya lenses. In my opinion, they're on par with Zeiss when stopped down to f8-f11.
The allure of a T/S system to me (for stiched images) over a dedicated pano head is less weight and bulk, something I appreciate more and more as the years pass. ;)
This is a link to a thread with a few images taken with my rig.http://www.dyxum.com/dforum/mirex-tilt-shift_topic64818_post727567.html?KW=#727567 (http://www.dyxum.com/dforum/mirex-tilt-shift_topic64818_post727567.html?KW=#727567)
Graham
-
So the technique works pretty well, but setup and calibration is a real pain in the neck. Here are two different images. Both are two-row panoramic images. Each row was composed of three images. One of these images had both rows taken with zero degrees of tilt. The other image has 2 degrees of tilt in the bottom row and zero degrees of tilt in the top row. The yaw rotation was done with a calibrated pano head while the second row was created with a 10mm shift upward. Sorry that these sample shots aren't anything particularly special as far as composition is concerned, but they are technically interesting. Also, sorry about the file sizes. You can't see the difference if the image is too small. The difference is particularly evident in prints. One final thing...to get the alignment between rows to optimize focus, I had to blend the two rows manually.
(http://www.trailpixie.net/temp-image/tilted.jpg)
(http://www.trailpixie.net/temp-image/not-tilted.jpg)
I think this bears-out the fact that the technique can work well. Now I just need to use it on something more interesting.
-
So the technique works pretty well, but setup and calibration is a real pain in the neck.
Hi Fike,
Thanks for taking the trouble to share your test results. I see you've opted for the 24mm TSE. I'm interested in the 17mm TSE as an alternative to the Nikkor 14-24 on the D700. If I were to buy the 17mm I would get the 5D2 at the same time. A single frame with the 17mm, using shift for perspective correction, should be wider than a 14mm shot after perspective correction in Photoshop, which inevitably results in cropping.
However, the possibility of using shift to create stitched panoramas, but without using a tripod, appeals to me greatly. Since I've had some success stitching close-up, hand-held 14-24mm shots using CS5's Photomerge (but not a 100% success), I wonder if the reduction of parallax issues using the shift, even when the camera is hand-held, might be of significant benefit when using automatic stitching programs.
I know you are mainly concerned about the effects of tilt in this thread, but since you are experimenting with your new lens, I wonder if you have tried stitching hand-held shots from your 24 TSE using programs such as Autopano Pro or CS5's Photomerge.
-
However, the possibility of using shift to create stitched panoramas, but without using a tripod, appeals to me greatly. Since I've had some success stitching close-up, hand-held 14-24mm shots using CS5's Photomerge (but not a 100% success), I wonder if the reduction of parallax issues using the shift, even when the camera is hand-held, might be of significant benefit when using automatic stitching programs.
Hi Ray,
Last time I've tried, Autopano (Giga) couldn't figure out the shift. Other pano stitchers can utilize the Panotools 'd' and 'e' parameters to complete a successful stitch of shifted images. Especially when using wide angles such as from a 24mm or 17mm the virtual horizon shifts a huge (from a mathematical point of view) distance, so the software should be able to cope with that.
Cheers,
Bart
-
Hi Bart,
I've also found that Autopano can sometimes get surprisingly confused. The reason I raise this issue is because my recent experience suggests that CS5 photomerge will respond to perspective corrected shots, taken from a close distance, which are also hand-held.
It's time-consuming to correct with 'free transform' every image before stitching. Sometimes the correction is not good enough and one has to repeat the exercise and do a better job. With perseverance one can eventually get a seamless stitch.
I'm just wondering, if the perspective correction were done through the use of a TSE lens, then turning the camera right or left for the next shot, as one would with a non-TSE lens, might make the automatic stitching process in CS5 easier. Or, as an alternative, one might use the shift mechanism to avoid turning the camera. Would this method produce a better stitch, albeit one that needed later perspective correction in PS?
I'm trying to find out what's possible without a tripod. I'd rather not use a tripod if I can get away without one.
-
...
I'm trying to find out what's possible without a tripod. I'd rather not use a tripod if I can get away without one.
I don't think that handholding a TS-E will be effective for panoramic work. I can't imagine how you would shift the lens and handhold with enough precision to get the benefits of the TS lens. Controlling the knobs and maintaining a stationary position would be next to impossible. If you are planning to handhold and pan the camera, I would recommend a normal wide angle lens instead. TS-E can be fun as a handheld lens, but only for single-frame work using the tilt for focus effects and the shift for angled shots (like mirrors and stuff). Precision work with a TS-E demands a tripod.
One problem I have noticed is that a shift of the lens seems to move the entrance pupil (sometimes called the nodal point) of the lens or to distort the image in ways that make stitching nearly impossible. So if you shoot a bottom row at -5mm and a top row at +10 mm you may have parallax problems or simple misalignment problems in one row or the other. This isn't a problem if you are only shifting to make your pano. It is only a problem if you are mixing shifting with panning to make a pano.
Let me try to explain:
If you shoot two shots, one above the other, and you use the shift to offset the frames and make a small pano, you will generally have no troubles. You will have two rows of one image (2x1). Where the problem comes is when you try to do two rows of two or more images. For reasons I haven't fully explored yet, shifted pairs of images will flat stitch perfectly, but adjacent vertical pairs may not stitch well because the upper row (+10mm shift, for example) will have too much error between frames.
I am not positive that the problem is actually parallax. My hypothesis is that the lens projection distorts the +10mm image so much that the upper row corners are warped in ways that exceed the abilities of the stitcher. The reason for this hypothesis is the observation that adjacent shifted image pairs may have a have a tree repeated in both frames but that bend in dramatically different directions (say 30 degrees different).
What does this mean in practice? I found that I could generally shift no more than 12mm from the bottom row to the top row while combining a flat stitch shift pano and a panning ptstitcher pano. Adding a tilt into this equation made the process even harder and meant that I couldn't shift more than total of 10mm.
But as for handheld panos using the TS-E shift mechanism: I wouldn't expect it to work well.
-
I don't think that handholding a TS-E will be effective for panoramic work. I can't imagine how you would shift the lens and handhold with enough precision to get the benefits of the TS lens. Controlling the knobs and maintaining a stationary position would be next to impossible.
I hand hold a 24 T/S for panoramas all the time-I prefer a tripod, but sometimes that is impractical. For example I am working on a city arts commission of people working out in our amazing public landscapes. I want to emphasize the landscape-hence the panorama-but I want to shoot the people candidly. So I have learned to shoot hand held. Suppose the person working is on the right shifted frame. I line up the shot shifting back and forth paying attention to "landmarks" in the frame. I shoot the left side shifted and then carefully monitoring my overlap and horizon line etc. I shoot a series of the right side where the people are working. I then use the left side and pick out the best "action" from the right side shifted series of exposures for a merge. It takes practice but works.
-
Hi Bart,
I've also found that Autopano can sometimes get surprisingly confused. The reason I raise this issue is because my recent experience suggests that CS5 photomerge will respond to perspective corrected shots, taken from a close distance, which are also hand-held.
Yes, I understand, that can be useful for a simple workflow. Unfortunately Photoshop doesn't offer a choice of resampling algorithms for the Photomerge function like the dedicated stitchers do. Image quality could therefore suffer needlessly, unless one has enough resolution for the intended output size. It's also a bit unclear to me which images in a stack are resized. One would expect at least one image to remain at native resolution (to avoid unnecessary resampling), but the user has no control over that.
I'm just wondering, if the perspective correction were done through the use of a TSE lens, then turning the camera right or left for the next shot, as one would with a non-TSE lens, might make the automatic stitching process in CS5 easier. Or, as an alternative, one might use the shift mechanism to avoid turning the camera. Would this method produce a better stitch, albeit one that needed later perspective correction in PS?
Up to a point it might, although the image quality suffers a bit as one gets closer to the edge of the image circle. A full left/right shift in landscape orientation will however produce such a wide angle of view that other issues become manifest (notably anamorphic distortion). Itf the stitcher can handle shifted images, then it should be beneficial to use shifted tiles to composite the panorama, because less extreme resampling is needed, and thus less cropping takes place.
I'm trying to find out what's possible without a tripod. I'd rather not use a tripod if I can get away without one.
Yes, it's always good to have the different scenarios covered beforehand rather than having to deal with them at the spur of the moment.
Cheers,
Bart
-
I hand hold a 24 T/S for panoramas all the time-I prefer a tripod, but sometimes that is impractical. For example I am working on a city arts commission of people working out in our amazing public landscapes. I want to emphasize the landscape-hence the panorama-but I want to shoot the people candidly. So I have learned to shoot hand held. Suppose the person working is on the right shifted frame. I line up the shot shifting back and forth paying attention to "landmarks" in the frame. I shoot the left side shifted and then carefully monitoring my overlap and horizon line etc. I shoot a series of the right side where the people are working. I then use the left side and pick out the best "action" from the right side shifted series of exposures for a merge. It takes practice but works.
That's cool! You also have the advantage that your subjects don't realize the camera is pointed at them. Are you able to do it for more than a couple frames, or is this something that you typically do with only two or three images?
-
I hand hold a 24 T/S for panoramas all the time-I prefer a tripod, but sometimes that is impractical. For example I am working on a city arts commission of people working out in our amazing public landscapes. I want to emphasize the landscape-hence the panorama-but I want to shoot the people candidly. So I have learned to shoot hand held. Suppose the person working is on the right shifted frame. I line up the shot shifting back and forth paying attention to "landmarks" in the frame. I shoot the left side shifted and then carefully monitoring my overlap and horizon line etc. I shoot a series of the right side where the people are working. I then use the left side and pick out the best "action" from the right side shifted series of exposures for a merge. It takes practice but works.
I can see where the T/S lens would be beneficial if you were using tilt handheld, but I don't see any advantage to using shift handheld. Have you tried just rotating around the entrance pupil? It would certainly speed up the process and allow you to grab a lot more action frames and also give you a lot more FOV to play with.
-
You actually think rotating around the entrance pupil hand held is easier than what I do? Have you tried either method? I've tried both. While looking through the viewfinder, I can actually see my reference points in the landscape for stitching, whereas I would be guessing where I am with respect to the entrance pupil.
Also, since I have already shot the side where the action isn't, I can shoot the side where the action is to my hearts content (unless the light changes and then I have to shift back to shoot the other side).
The bottom line is that this works and has become a regular tool in income stream. I did a similar trick two weeks ago while shooting the baseball fields at Freedom Park in Las Vegas for the architects, standing on top of a ladder shooting handhold panoramas of adjacent baseball fields.
-
This topic is about use of T/S lenses and adapters... but if you can use a view camera and tilt the lens and then shift the sensor, life gets so much easier... and the longer Apo-Digitars give you enough image circle for about six shots. (Yes, the kit is more expensive, but it is becoming more affordable).
-
You actually think rotating around the entrance pupil hand held is easier than what I do? Have you tried either method? I've tried both. While looking through the viewfinder, I can actually see my reference points in the landscape for stitching, whereas I would be guessing where I am with respect to the entrance pupil.
Also, since I have already shot the side where the action isn't, I can shoot the side where the action is to my hearts content (unless the light changes and then I have to shift back to shoot the other side).
The bottom line is that this works and has become a regular tool in income stream. I did a similar trick two weeks ago while shooting the baseball fields at Freedom Park in Las Vegas for the architects, standing on top of a ladder shooting handhold panoramas of adjacent baseball fields.
Well, I've shot quite a few handheld panoramas and I built a spherical pano head for my camera that has front and back movements (swing, tilt, shift, rise, and fall) so I guess I have a little experience :) The entrance pupil is actually quite easy to find, just look at the front of the lens after you've set the focus somewhat close to what you'll be using. It does take some practice to perfect, but certainly no harder than the technique you're using. Rotating around the entrance pupil doesn't limit you to a single focal length or FOV. There isn't anything wrong with your technique, it just a bit more limiting than rotating around the entrance pupil.
-
Well, I've shot quite a few handheld panoramas and I built a spherical pano head for my camera that has front and back movements (swing, tilt, shift, rise, and fall) so I guess I have a little experience :) The entrance pupil is actually quite easy to find, just look at the front of the lens after you've set the focus somewhat close to what you'll be using. It does take some practice to perfect, but certainly no harder than the technique you're using. Rotating around the entrance pupil doesn't limit you to a single focal length or FOV. There isn't anything wrong with your technique, it just a bit more limiting than rotating around the entrance pupil.
There are a couple of nice things about his approach, and the biggest is that parallax error doesn't matter. because it is a flat stitch (with very little overlap) there is no need to worry about foreground elements having parallax error. No matter how good you are, rotating around the entrance pupil is imprecise when you are handholding. The other nice thing about his approach is the ability to photograph scenes of people without them being aware that you are pointing at them.
-
The other nice thing about his approach is the ability to photograph scenes of people without them being aware that you are pointing at them.
Interesting, I never thought of that.
-
FWIW here is an example-shot over a tall fence form the top step of an 8 foot ladder leaned against a fence post. That puts me up at about 13.5 feet. A simple hand held flat shift stitch ISO 400, f/11 at 1/800 sec. Canon 24 II T/S, giving me a nice 29.5 x 11.5 300 DPI image after some slight cropping. It was commossioned by the architects. I also sometimes do a more diagonal shift when I need some rise or fall.
-
This is a shift done with a Pana G1 and 90mm Tokina macro. The lens is attached via a Fotodiox EOS shift adapter to body then a Nikon adapter to the EOS, then lens. Allows shift and the adapter rotates for verticals. No tilt.
The image size is 2955 x 8861.
The nice thing with this setup is it is small, I always have it with me, it is relatively inexpensive, very fast to operate, and shift stitches are very easy with CS5. Having no parallax problem is almost a requirement for macro panos.
Larry