Right now, one can slide the camera an equal and opposite direction to compensate for the movement of the lens, twice to get a three shot set to stitch.

I'm curious to know how many folks would be interested in a clamp that would hold the lens body of a shift lens to allow easy, parallax-free flat stitching by allowing the camera to be moved while the lens is held stationary. The clamp could be made to attach to a rail like the RRS rail so the camera/lens combo would slide fore and aft to line up the axis of rotation with the nodal point/exit pupil as normal. The Canon shift lenses at least have locking nuts so the camera wouldn't move after locking and it would be soooo much more convenient than sliding cameras side to side, especially if you plan to use the shift in a vertical direction to get multi-rows, while rotating around the optical axis to get the horizontal shots for each row.

My email with this suggestion was sent to engineering at RRS and if there's enough interest, we might be able to convince them it's worth developing

I had to read your post a few times to "get it". This thing needs to look like a tripod collar.

It may work - you'll probably find a lot of variation in the many types of tilt shift bodies for a universal clamp. On the Canon TS-E's the amount of available mass to clamp onto is very narrow - less than 1/4"thick - right up against the camera body - with a rotation lock / release at 9 o'clock. Then there is the weight of the hanging camera body that would need to be moved by a small lens knob - tough to do with a 1D series Canon body - especially vertically. If anyone could do it - RRS would do a great job.

I do use my 24TS-E for three stitch pano's and use an RRS "L" bracket on the camera with a quick releases on the tripod and just slide the "L" bracket left and right in the quick release. I'm very happy not to need additional hardware.

The clamp has to attach to the lens well away from the camera body so the body can move freely - I'm looking at my 24 TSE and the square part between the mount and the focusing ring has large grooves on either side (the sides that don't have the tilt controls on). The newer ones look like they would need the clamp to have semicircular indents to grip the circular part of the lens body. Unless I had all the lenses in front of me, I couldn't tell whether there's a hope of designing a clamp that could be useful to them all - possibly with different grips on the basic U-shaped clamp that attaches to the rail. For simple horizontal 3 shot stitches, the sideways sliding of the L-bracket isn't too much of a hassle. Using portrait mode to get three verticals to stitch flat, then rotate around the "nodal" point then repeat to get each angle of a 3-row pano is where it would help some

I'm not sure how much this would really help, but it's a brilliant concept, worth exploring further.

It would indeed be hard to do, clamping to the lenses. Plus, that would put serious strain on the shift screw - it is designed for the weight of a small lens, not a camera. Seems like a BAD idea to me.

Maybe there's a better way:
---------------------------------------

The camera sits on a base (basically a different type of pano system) and there is a long flat rail UNDER the lens, pointing forward. It is adjustable and so can be pulled out to just past the front edge of the lens.
This rail has a line etched into its center, which marks the LENS longitudinal center.

Once you dial-in your desired shift, you then turn a dial on this rig which moves the CAMERA left-right (or up-down in portrait mode) and you do so until the center of the lens once again aligns with the line on the extender rail.

Done.

-But this will be heavy & expensive. There would have to be some SERIOUS optical benefit for this to be a viable thing.

Well, you could go cheaper, and just have the camera on a locking slider, instead of geared teeth.

One has to ask the question: How often do you need to take panos with high structures (causing keystoning considerations) that also have in-focus object in the foreground?

Landscapes? Not likely

Citiscapes? Maybe, but probably not that often.

Architecture? Bingo. There's your market. You might want to find some architecture or real estate forums, and float this idea there.

Thanks for thinking it through! Landscapes may not need to have much very close to the camera for parallax to be an issue for stitching multi-rows (art least my mainly wide-angle ones) The TSE24 isn't much lighter than the 5D so the weight issues are not too much of a concern for the mount. Tripod collars on long lenses let the body hang out on its own (think super tele with a 1D series on it) . It can be done with other devices but they're cumbersome and probably more costly than this would end up. It's mainly for folks trying to increase the MP count. A full pano rig is quite a few $$$ and that's the real alternative that is used today. Perhaps RR$ won't be interested enough because it would eat into their pano rig market
I may have to look up a couple of pals with machine shops

AJSJones wrote:
The TSE24 isn't much lighter than the 5D

It's a LOT lighter than a 5D with a battery pack & L-plate.

AJSJones wrote:
Tripod collars on long lenses let the body hang out on its own (think super tele with a 1D series on it) .

Sure, but only the lens-mount takes the weight. We're talking here a bout a tiny screw & threaded rail, meant to move the lens sideways. It won't last. I guarantee that RRS will say the same thing.

I've been shooting shift lens stitched panos for a few years. As mentioned above when doing landscape with subjects far away the parallax wasn't a problem. But, now doing architecture or close subjects parallax is off and the overlaps don't line up. So, I've been planning to build a lens clamp like you suggested to keep the lens stationary thereby shifting the camera body side/side or up/down. I think the point of using a shift lens is to keep the 2 or 3 image pieces parallel (making the subject lines parallel). So, I don't see any need to rotate the camera around the nodal. If you rotate around the nodal your lines won't be parallel and then there's no need for a shift lens anyway. But, maybe you have a different capture/stitch method in mind.

I was also imagining clamping onto the flat sides of the squarish center piece of the 24TSE although the newer TSEs don't have these big wide flat sides. Or, maybe doing a thin clamp around the round part just behind the focus ring where the aperture range is printed but, might be trickier to build.

I still think my way would be safer, and probably much easier.

You could likely just modify a macro-rail system. Find some way to mount it sideway (to the camera.) Then, mark the dead-center point, and add a matching mark on your camera's base.

Last: mark the two outer points that correspond to you maximum shift amount, and you're done.
(or add a sliding front guide-rail, as I originally outlined.)
--------

dendowling, two questions for you:


1: When you make these using SIDEWAYS shift, (just to be clear) is there a specific amount of shift that works best, or does one simply use the most that the lens will give? Is there some kind of math involved?


2: Have you ever made comparison shots, as I asked about earlier? That is, two "identical" panos, but one done with sideways shift, and the other done with nodal panning?

I am DYING to see if there's a noticeable difference, either in perspective, or in sharpness due to the software "faking" the perspective.

Honestly, I haven't experimented with nodal pan/stitch. When you use nodal pan/stitch do the two pieces just line up without digital correction or do you need software that bends image shape of the two pix into alignment? Also, when you do a nodal stitch are all the horizontal lines still parallel or are they curved?

The reason why I use the shift pan stitch is to avoid any software tweaking the image shape. I can align the two halves by simple up/down left/right movement in photoshop. The reason for this is because on each half I shoot two exposures (under/over) for a light and dark version that can be used to dodge/burn the image. But, the auto-stitch software won't do the light version exactly the same as the dark version so, the auto-stitched light pic never aligns with the auto-stitched dark pic.

I like your idea of just sliding the camera sideways to counteract the lens shift but, I wonder how accurate is the placement of the lens even with the guide you mentioned. The concept with the lens mount is to keep the lens perfectly stationary so you don't have to worry about the position of the lens .

Here are examples of the side/side shift stitch and up/down shift stitch. Is this possible with a nodal pan/stitch? I'm not making a challenge, just wondering what nodal can do.

Den,

Your first pic has a LOT of keystone error. What stitching software you are using? According to some posts, above, modern stitching software automatically compensates for this (though I don't see how it could be automatic, and it may well cause a noticeable loss of detail)

Did you perhaps turn off this function, on purpose, to keep maximum sharpness? -I recently did a test of basic keystone adjustment, using Lensfix, and was shocked at how much detail was lost. Never thought about it before.

I think the ultimate rig for that type of shot (exactly what I was talking about in my OP) is to use a shift lens n portrait orientation, shifted UP, plus a nodal-panning base. That combination should yield minimum keystoning, no parallax error, and maximum sharpness.

But I'm still interested in seeing results & hearing more opinions.

you haven't tried any modern software. 5 shot HDR panos where there are 5 shots in each position at 1 stop apart are no problem with anything modern. you stitch all images at once and then go away until it is done. as for simple shift alignment, there is a lot to be said for it but that's not why i am going to get a shift lens. it's to get better control of the focus plane for DOF control.

Herb...

dendowling wrote:
The reason why I use the shift pan stitch is to avoid any software tweaking the image shape. I can align the two halves by simple up/down left/right movement in photoshop. The reason for this is because on each half I shoot two exposures (under/over) for a light and dark version that can be used to dodge/burn the image. But, the auto-stitch software won't do the light version exactly the same as the dark version so, the auto-stitched light pic never aligns with the auto-stitched dark pic.

HerbChong wrote:
you haven't tried any modern software. 5 shot HDR panos where there are 5 shots in each position at 1 stop apart are no problem with anything modern. you stitch all images at once and then go away until it is done. as for simple shift alignment, there is a lot to be said for it but that's not why i am going to get a shift lens. it's to get better control of the focus plane for DOF control.

Herb...





Well you could also do DOF focus bracketing in addition to the exposure bracketing. Of course then you'd be taking up to 5 frames for HDR at each focus point in your image, and you'd want several frames at least for the focus bracketing to get the DOF your after, so we are talking up to 25 frames per stitched part of the scene, and with just a simple 5 frame pano you'd then but up to 150 frames. Yikes!

Not only would that take forever to stitch but I'm sure the light would of changed from the time the first frame is taken til the last so then that poses an issue.

It could be argued that you could selectively focus bracket just parts of the image, as with exposing less frames for certain parts, but then that gets really confusing to keep track of in the field I think.

"Old school" tilt method to increase DOF is the winner and makes a pc-e/ts lens well worth the money. Just one shot can give you the front to back sharpness you need at an average apeture you want.

AJSJones wrote:
Right now, one can slide the camera an equal and opposite direction to compensate for the movement of the lens, twice to get a three shot set to stitch.

I'm curious to know how many folks would be interested in a clamp that would hold the lens body of a shift lens to allow easy, parallax-free flat stitching by allowing the camera to be moved while the lens is held stationary. The clamp could be made to attach to a rail like the RRS rail so the camera/lens combo would slide fore and aft to line up the axis of rotation with the nodal point/exit pupil as normal. The Canon shift lenses at least have locking nuts so the camera wouldn't move after locking and it would be soooo much more convenient than sliding cameras side to side, especially if you plan to use the shift in a vertical direction to get multi-rows, while rotating around the optical axis to get the horizontal shots for each row.

My email with this suggestion was sent to engineering at RRS and if there's enough interest, we might be able to convince them it's worth developing



I'm curious what a clamp would provide holding the lens over simply doing it the "old" way of shifting the lens one direction and then shifting the camera body the other direction ?

I have marks on my clamp that shows how far to shift the dovetail plate to match what I shift the lens and it works great.

I guess if only the camera was moving it would be slightly faster but it would seem like sort of another pricey clamp to get, probably $300 knowing RRS.

I think everyone already has an L plate and dovetail clamp on their ballheads (or at least those who are serious enough to own PC lenses) so I would think it would be a very limited market for such a product.

It woudl work of course but I think to be sturdy you'd have to attach it to the lens body, and while there are some nice flat panels to do that with, you'd have to drill out the lens to attach it and how many want to drill into a $2k lens ?

millsart wrote:
I think everyone already has an L plate and dovetail clamp on their ballheads (or at least those who are serious enough to own PC lenses) so I would think it would be a very limited market for such a product.


Well, I don't, but then I have yet to buy a fancy pano rig. I am very new to panos and so far have just done hand-helds.

I am looking to improve my quality, and recently obtained a shift-lens, hence perusing threads such as this one.

Can you describe (with pics if possible) your rig? I realize that dovetails are used to mount a pano rig onto your regular ballhead. OK, that works for nodal-panned panos, but how are you using this for shift-lens panos? Something like what I outlined, above? (A sideways-mounted "macro" rail?)

Use an RRS "L" bracket on your 5d - there are center line marks on the bracket for the landscape and portrait positions. Use an RRS ballhead clamp - there are center line & 10mm graduation marks to the left & right of center. Set up on a tripod - use a bubble level. If you have those items you don't need a pano rig or marco rail for three frame shift pano's. I keep my "L" bracket on my camera - full time.

Set (slide) your camera to the left 10mm graduation - shift TS-E right 10mm - SHOOT - Set (slide) camera to center - TS-E center - SHOOT - Set (slide) camera right 10mm - shift left 10mm - ETC. The lens stays on a center line - the body moves. I don't even lock the ballhead clamp or TS-E locks.

Sorry no pics - gotta go to work to pay for camera gear! Have fun!

OK, I get it. You are just sliding the plate inside the clamp. I guess that's workable. I was thinking more along the lines of a threaded-screw micro adjuster. (like a macro rig, but sideways)

-But I don't understand how you are using the marks on the L-plate. It's the LENS that must stay centered. It seems to me that you are just guess-timating, and not keep the lens ABSOLUTELY in the same position. No?

I would definitely be interested. My current solution is the same as above with the L bracket marked (or maybe a macro rail ?) but it gets expensive fast.

This does keep the lens centered on the same axis and effectively moves the body 10mm left or right for a given shot. There was an article on this a few years back a google search might turn up.

Cableaddict wrote:
OK, I get it. You are just sliding the plate inside the clamp. I guess that's workable. I was thinking more along the lines of a threaded-screw micro adjuster. (like a macro rig, but sideways)

-But I don't understand how you are using the marks on the L-plate. It's the LENS that must stay centered. It seems to me that you are just guess-timating, and not keep the lens ABSOLUTELY in the same position. No?


You shift the lens 11mm to the left, so then you shift the camera body 11mm to the right. The net result is the lens stays in exactly the same postion.

The lens has mm markings so you can tell how far you shifted it, and the L plate has markings to tell how far you moved that.

If your L plate doesn't have markings on, its pretty easy to just measure 11mm and put a little paint mark so you know how far to slide it each direction.

It is exact right down to the micron ? No, but its accurate to at least half a millimeter and thats more than accurate enough