tilt shift question

wfektar said:

on the distance from the tilt axis to either the rear nodal point or to the sensor plane, depending whether you're mounting to the camera body or to the lens/adapter tripod mount. The former gives you a front tilt + fall/rise, the latter a rear tilt + rise/fall. Rear swings/tilts are usually less demanding of image circle (at least on a proper view camera, where the tilt axis is the sensor plane, where it isn't demanding at all) but it will depend on your adapter and lens. Rear swings/tilts will also affect perspective. BTW, the nodal points for telephotos and reverse telephotos (retrofocus wideangles) aren't always obvious.

Click to expand...

Christof21 said:

Hi,

I would like to extend my question. Imagine that instead of buying a T&S lens, I use a tilt adaptor on my APS-C with a FF lens. How much tilt will be possible ? Does it depend on the focal lens ? I guess the tilt will be limited by the image circle ?

Click to expand...

Joseph S Wisniewski said:

Christof21 said:

Hi,

I would like to extend my question. Imagine that instead of buying a T&S lens, I use a tilt adaptor on my APS-C with a FF lens. How much tilt will be possible ? Does it depend on the focal lens ? I guess the tilt will be limited by the image circle ?

Click to expand...

You keep confusing "tilt" and "shift". What you want for panoramas is "shift". A FF lens on an APS C camera can shift about 7mm diagonally, or a bit over 8mm left-right. FF lenses don't have that much "extra" image circle when used on APS cameras. The FF image circle is 43.3mm, the APS image circle about 28mm (depending on brand). If you're going to make panoramas by shifting a FF lens in front of an APS sensor, you would be better off to just shoot a FF camera. Then you wouldn't have to stitch multiple images.

That shift lens you were looking at has a larger image circle, over 50mm, and can therefore accommodate a larger shift. It makes shifting worthwhile.

What are you shooting panoramas of? If you're going to stitch, you'll probably get better results simply by using a regular lens and rotating the camera and lens together using a panorama adapter.

--
Rahon Klavanian 1912-2008.
Armenian genocide survivor, amazing cook, scrabble master, and loving grandmother. You will be missed.
Ciao! Joseph
www.swissarmyfork.com

Click to expand...

wfektar said:

Well, I don't know how to answer that except as "it depends", but 12 deg of tilt should not run out of image circle and will be plenty for just about anything. With 15 mm of shift (is that 15 total or +/- 15?) you will likely get vignetting but you'll be able to see it when it happens in real time.

Click to expand...

artistguy said:

An excellent thread with plenty of good and bad understanding of the subject.

As a photographer that actually owns the Samyang 24ts and the manfrotto 303 pano head, and uses both with FF and m43 I might add something to the subject.

First of all, the Samyang is around half the cost of the Canon, it's not as sharp until you're at F8 on FF, from then on it's very good. It is fully manual focus and aperture. There's a slow and deliberate way of using the lens. You really want it on a tripod, camera set to manual exposure, open the aperture to allow brighter image to focus. Then stop the aperture down, then set the exposure. Take the first image. then shift upto 10 degrees (12 is maximum but will vignette!) take another shot, and then shift the other way 10 degrees. note that you can also rotate the shift axis so that you can take vertical panoramas such as waterfalls. I use the photomerge function in photoshop to make the pano from the 3 images.

One advantage of the ts over the pano head is that you can of course tilt for increased dof whilst also shifting for panorama making. The main function of shift is actually to correct perspective, such as leaning buildings effect. The tilt can be used in the opposite way to create miniature toy effects for video and stills. The pano head adds another 2 kg to the weight of the kit to be carried to the location so I tend to take the ts as it's also a very sharp 24mm prime lens.

Click to expand...

artistguy said:

The ts also enables mega pixel images to be made, I note that some posts have said it's simpler to just take an image with a wider lens (i also use the Samyang 14mm) for field of view rather than shift for 3 images. That's missing an important point, if you start with say 16mp FF image with the 14mm lens and crop to a pano, you'll have say a 10mp image to work with. If you take 3 shifted images with the 24mm ts and merge, you'll have an image of around 30+mp to work with so if you're looking to print big and really capture the detail, the ts has the advantage. You can also shift up/down with the camera in landscape format for mega pixel square images.

As said by a previous poster, simply moving the tripod left or right to mimic shift will not have the same effect at all.

Click to expand...

artistguy said:

Using m43 with the lens means more photos to create the same angle of view image but results in a mega-mega-pixel image for really large printing.

Click to expand...

Glenn NK said:

Well that was in interesting read (many of the posts). There seems to still be some confusion about the TS lens and its functions.

I've been using mine for most of three years and rather than attempt to explain the intracacies of the lens, I suggest this link:

http://www.photozone.de/canon_eos_ff/603-canon24f35tse2

On the specific topic of stitching a panorama image, may I suggest that shifting and then sliding the body in an opposite amount is much simpler than locating the nodal point of a lens. Shifting also has the advantage (over rotating) of retaining a flat field of focus on both images.

If I were to shoot a two image panorama, I would shift the lens 12 mm to the left (for example), and then slide the camera and lens 12 mm to the right (my tripod head allows this - many standard ballhead may not). Then repeat the process by shifting/sliding in the opposite directions.

On the use of Tilt to change plane of focus, I found that tables were an awkward method. An easier solution:

1) Adjust camera to acquire the desired framing of image.

2) Using LV, focus the lens at infinity or some important distant object using the focus ring.

3) Using LV, focus the lens at some important near object using the Tilt adjustment. Do not touch the focus ring.

4) Repeat steps 2) and 3) to ensure accuracy.

Once you've tried this method, I think the tilt charts will be thrown away.

Click to expand...

Member said:

Thanks for your feedback about this lens. I wish it could be available with Fuji !

Click to expand...

Glenn NK said:

Well that was in interesting read (many of the posts). There seems to still be some confusion about the TS lens and its functions.

I've been using mine for most of three years and rather than attempt to explain the intracacies of the lens, I suggest this link:

http://www.photozone.de/canon_eos_ff/603-canon24f35tse2

Click to expand...

Glenn NK said:

On the specific topic of stitching a panorama image, may I suggest that shifting and then sliding the body in an opposite amount is much simpler than locating the nodal point of a lens.

Click to expand...

Glenn NK said:

Shifting also has the advantage (over rotating) of retaining a flat field of focus on both images.

Click to expand...

Glenn NK said:

If I were to shoot a two image panorama, I would shift the lens 12 mm to the left (for example), and then slide the camera and lens 12 mm to the right (my tripod head allows this - many standard ballhead may not). Then repeat the process by shifting/sliding in the opposite directions.

On the use of Tilt to change plane of focus, I found that tables were an awkward method. An easier solution:

1) Adjust camera to acquire the desired framing of image.

2) Using LV, focus the lens at infinity or some important distant object using the focus ring.

3) Using LV, focus the lens at some important near object using the Tilt adjustment. Do not touch the focus ring.

4) Repeat steps 2) and 3) to ensure accuracy.

Once you've tried this method, I think the tilt charts will be thrown away.

Click to expand...

Glenn NK said:

--
Glenn NK

Click to expand...

Christof21 said:

artistguy said:

An excellent thread with plenty of good and bad understanding of the subject.

As a photographer that actually owns the Samyang 24ts and the manfrotto 303 pano head, and uses both with FF and m43 I might add something to the subject.

First of all, the Samyang is around half the cost of the Canon, it's not as sharp until you're at F8 on FF, from then on it's very good. It is fully manual focus and aperture. There's a slow and deliberate way of using the lens. You really want it on a tripod, camera set to manual exposure, open the aperture to allow brighter image to focus. Then stop the aperture down, then set the exposure. Take the first image. then shift upto 10 degrees (12 is maximum but will vignette!) take another shot, and then shift the other way 10 degrees. note that you can also rotate the shift axis so that you can take vertical panoramas such as waterfalls. I use the photomerge function in photoshop to make the pano from the 3 images.

One advantage of the ts over the pano head is that you can of course tilt for increased dof whilst also shifting for panorama making. The main function of shift is actually to correct perspective, such as leaning buildings effect. The tilt can be used in the opposite way to create miniature toy effects for video and stills. The pano head adds another 2 kg to the weight of the kit to be carried to the location so I tend to take the ts as it's also a very sharp 24mm prime lens.

Click to expand...

Thanks for your feedback about this lens. I wish it could be available with Fuji !

Member said:

The ts also enables mega pixel images to be made, I note that some posts have said it's simpler to just take an image with a wider lens (i also use the Samyang 14mm) for field of view rather than shift for 3 images. That's missing an important point, if you start with say 16mp FF image with the 14mm lens and crop to a pano, you'll have say a 10mp image to work with. If you take 3 shifted images with the 24mm ts and merge, you'll have an image of around 30+mp to work with so if you're looking to print big and really capture the detail, the ts has the advantage. You can also shift up/down with the camera in landscape format for mega pixel square images.

As said by a previous poster, simply moving the tripod left or right to mimic shift will not have the same effect at all.

Click to expand...

I disagree. The projection is different (rectilinear versus cylindric) but you can use the projection you want with the pano software without problem.

Click to expand...

chkproductions said:

The term Stitching and Pano are interchanged but probably shouldn't be.

The term Stitching through a lens or sensor shift and Panos through the rotation of the camera, even though they they both take multiple images and blend them into one image, work better for different subject types.

A pano is great for stitching multiple images of a distant landscape or city skyline where most subjects are far from the lens and there is greater ability to "camouflage" any discrepancies in the blend. And you can achieve a wider angle of view.

Stitching is best when the working distance to your subjects is restricted, as in an interior or with the defined angles and lines of architecture. Also stitching through shifting is more often 2 or 3 blended images that really just give a wider visual coverage retaining a normal perspective than you would get with a one shot capture with a wide angle lens that matches the same field of view.

So it seems someone would have to really decide which type of final image they want to take in order to determine the technique, shifting or rotating, is best for them.

Click to expand...

dsjtecserv said:

chkproductions said:

The term Stitching and Pano are interchanged but probably shouldn't be.

The term Stitching through a lens or sensor shift and Panos through the rotation of the camera, even though they they both take multiple images and blend them into one image, work better for different subject types.

A pano is great for stitching multiple images of a distant landscape or city skyline where most subjects are far from the lens and there is greater ability to "camouflage" any discrepancies in the blend. And you can achieve a wider angle of view.

Stitching is best when the working distance to your subjects is restricted, as in an interior or with the defined angles and lines of architecture. Also stitching through shifting is more often 2 or 3 blended images that really just give a wider visual coverage retaining a normal perspective than you would get with a one shot capture with a wide angle lens that matches the same field of view.

So it seems someone would have to really decide which type of final image they want to take in order to determine the technique, shifting or rotating, is best for them.

Click to expand...

I think for the most part you are making distinction without a difference. A panorama is any image that shows a particularly wide, unbroken view (search on the adjective "panoramic") (in either the vertical or horizontal direction, or both); normally, though not necessarily, with a large aspect ratio. There is no threshold for how wide or narrow a panorama must be, nor on how it is created. A single image can be a panorama, particularly if it is cropped to emphasize the width. Stitching of multiple images is one way (and certainly the most common) to create a panorama. And "stitching" includes the process of putting together images made either by rotation of the camera/lens or by panning the sensor across the image circle of a T/S lens. There is no inherent difference in the result: both are ways to create panoramas.

The difference in applicability to different subject types isn't due to the way the panorama are made, but rather with the projection. For moderately wide panoramas, and especially for architectural subjects, the rectilinear projection given by a T/S lens is fine and usually preferable. But the same panorama, with the same rectilinear projection, can be created by rotating the camera and lens.

Rotated panoramas are not limited to distant subjects, nor are shifted panos inherently better for close ones. Parallax is a potential issue any time there are both near and far objects, and with either technique some method will need to be used to minimize parallax error. I agree that both take about the same amount of time and hassle to set up; which is to say, there is little or no advantage to either in this respect.

With respect to the "normal perspective" given by using shifting, you need to keep in mind that perspective is a function of shooting position, and not the lens or the stitching method used. Further, keep in mind that you are simply panning the sensor across the larger image circle (and wider field of view) of the lens. The result is exactly the same as if you had used a larger sensor, with a diagonal approximating the diameter of the image circle to make a single shot, and then cropped it to match the aspect ratio of your stitched panorama. With the smaller sensor you are simply sampling the image circle (and the complete image that could have been made with a larger sensor) and putting the pieces back together. Therefore there is no inherent difference between that and a single image; it is just a matter whether is was created from one sample or more.

Click to expand...

dsjtecserv said:

The big advantage, and probably the only important one, with T/S shifting to create a panorama is the ability to use tilt for focus plane control while shifting. You can do that with a normal lens when rotating. And even though panos done by shifting the T/S have a limited field of view, it can also be rotated to create panoramas with an unlimited field. so using a T/S lens does have a leg up in that respect, nor mater which way you use it.

Click to expand...

dsjtecserv said:

Dave

--
http://www.pbase.com/dsjtecserv

Click to expand...

Christof21 said:

dsjtecserv said:

chkproductions said:

The term Stitching and Pano are interchanged but probably shouldn't be.

The term Stitching through a lens or sensor shift and Panos through the rotation of the camera, even though they they both take multiple images and blend them into one image, work better for different subject types.

A pano is great for stitching multiple images of a distant landscape or city skyline where most subjects are far from the lens and there is greater ability to "camouflage" any discrepancies in the blend. And you can achieve a wider angle of view.

Stitching is best when the working distance to your subjects is restricted, as in an interior or with the defined angles and lines of architecture. Also stitching through shifting is more often 2 or 3 blended images that really just give a wider visual coverage retaining a normal perspective than you would get with a one shot capture with a wide angle lens that matches the same field of view.

So it seems someone would have to really decide which type of final image they want to take in order to determine the technique, shifting or rotating, is best for them.

Click to expand...

I think for the most part you are making distinction without a difference. A panorama is any image that shows a particularly wide, unbroken view (search on the adjective "panoramic") (in either the vertical or horizontal direction, or both); normally, though not necessarily, with a large aspect ratio. There is no threshold for how wide or narrow a panorama must be, nor on how it is created. A single image can be a panorama, particularly if it is cropped to emphasize the width. Stitching of multiple images is one way (and certainly the most common) to create a panorama. And "stitching" includes the process of putting together images made either by rotation of the camera/lens or by panning the sensor across the image circle of a T/S lens. There is no inherent difference in the result: both are ways to create panoramas.

The difference in applicability to different subject types isn't due to the way the panorama are made, but rather with the projection. For moderately wide panoramas, and especially for architectural subjects, the rectilinear projection given by a T/S lens is fine and usually preferable. But the same panorama, with the same rectilinear projection, can be created by rotating the camera and lens.

Rotated panoramas are not limited to distant subjects, nor are shifted panos inherently better for close ones. Parallax is a potential issue any time there are both near and far objects, and with either technique some method will need to be used to minimize parallax error. I agree that both take about the same amount of time and hassle to set up; which is to say, there is little or no advantage to either in this respect.

With respect to the "normal perspective" given by using shifting, you need to keep in mind that perspective is a function of shooting position, and not the lens or the stitching method used. Further, keep in mind that you are simply panning the sensor across the larger image circle (and wider field of view) of the lens. The result is exactly the same as if you had used a larger sensor, with a diagonal approximating the diameter of the image circle to make a single shot, and then cropped it to match the aspect ratio of your stitched panorama. With the smaller sensor you are simply sampling the image circle (and the complete image that could have been made with a larger sensor) and putting the pieces back together. Therefore there is no inherent difference between that and a single image; it is just a matter whether is was created from one sample or more.

Click to expand...

I agree with everything so far. Thanks for the detailed information.

Member said:

The big advantage, and probably the only important one, with T/S shifting to create a panorama is the ability to use tilt for focus plane control while shifting. You can do that with a normal lens when rotating. And even though panos done by shifting the T/S have a limited field of view, it can also be rotated to create panoramas with an unlimited field. so using a T/S lens does have a leg up in that respect, nor mater which way you use it.

Click to expand...

Sorry, english is not my native language. You start saying that this is an advantage to be able to tilt while shifting. Then you say that normal rotating can give the same result (tilt + rotating = titt + shift ?).

So your conclusion is that the shift is useless in fact, right ? Sorry if I misunderstand.

The tilt, in the opposite, can not be replaced. Tillting has some advantages over focus stacking (and vice versa).

If you had to conclude, do you recommend a T&S (for landscape/pano photography) ? Can it make a big difference ?

Do you recommend a T&S to correct perspective (architectural shots) ?

Click to expand...

dsjtecserv said:

Christof21 said:

dsjtecserv said:

chkproductions said:

The term Stitching and Pano are interchanged but probably shouldn't be.

The term Stitching through a lens or sensor shift and Panos through the rotation of the camera, even though they they both take multiple images and blend them into one image, work better for different subject types.

A pano is great for stitching multiple images of a distant landscape or city skyline where most subjects are far from the lens and there is greater ability to "camouflage" any discrepancies in the blend. And you can achieve a wider angle of view.

Stitching is best when the working distance to your subjects is restricted, as in an interior or with the defined angles and lines of architecture. Also stitching through shifting is more often 2 or 3 blended images that really just give a wider visual coverage retaining a normal perspective than you would get with a one shot capture with a wide angle lens that matches the same field of view.

So it seems someone would have to really decide which type of final image they want to take in order to determine the technique, shifting or rotating, is best for them.

Click to expand...

I think for the most part you are making distinction without a difference. A panorama is any image that shows a particularly wide, unbroken view (search on the adjective "panoramic") (in either the vertical or horizontal direction, or both); normally, though not necessarily, with a large aspect ratio. There is no threshold for how wide or narrow a panorama must be, nor on how it is created. A single image can be a panorama, particularly if it is cropped to emphasize the width. Stitching of multiple images is one way (and certainly the most common) to create a panorama. And "stitching" includes the process of putting together images made either by rotation of the camera/lens or by panning the sensor across the image circle of a T/S lens. There is no inherent difference in the result: both are ways to create panoramas.

The difference in applicability to different subject types isn't due to the way the panorama are made, but rather with the projection. For moderately wide panoramas, and especially for architectural subjects, the rectilinear projection given by a T/S lens is fine and usually preferable. But the same panorama, with the same rectilinear projection, can be created by rotating the camera and lens.

Rotated panoramas are not limited to distant subjects, nor are shifted panos inherently better for close ones. Parallax is a potential issue any time there are both near and far objects, and with either technique some method will need to be used to minimize parallax error. I agree that both take about the same amount of time and hassle to set up; which is to say, there is little or no advantage to either in this respect.

With respect to the "normal perspective" given by using shifting, you need to keep in mind that perspective is a function of shooting position, and not the lens or the stitching method used. Further, keep in mind that you are simply panning the sensor across the larger image circle (and wider field of view) of the lens. The result is exactly the same as if you had used a larger sensor, with a diagonal approximating the diameter of the image circle to make a single shot, and then cropped it to match the aspect ratio of your stitched panorama. With the smaller sensor you are simply sampling the image circle (and the complete image that could have been made with a larger sensor) and putting the pieces back together. Therefore there is no inherent difference between that and a single image; it is just a matter whether is was created from one sample or more.

Click to expand...

I agree with everything so far. Thanks for the detailed information.

dsjtecserv said:

The big advantage, and probably the only important one, with T/S shifting to create a panorama is the ability to use tilt for focus plane control while shifting. You can do that with a normal lens when rotating. And even though panos done by shifting the T/S have a limited field of view, it can also be rotated to create panoramas with an unlimited field. so using a T/S lens does have a leg up in that respect, nor mater which way you use it.

Click to expand...

Sorry, english is not my native language. You start saying that this is an advantage to be able to tilt while shifting. Then you say that normal rotating can give the same result (tilt + rotating = titt + shift ?).

So your conclusion is that the shift is useless in fact, right ? Sorry if I misunderstand.

The tilt, in the opposite, can not be replaced. Tillting has some advantages over focus stacking (and vice versa).

If you had to conclude, do you recommend a T&S (for landscape/pano photography) ? Can it make a big difference ?

Do you recommend a T&S to correct perspective (architectural shots) ?

Click to expand...

Well, even though English is my first language, typing isn't. You didn't misunderstand, I meant to say you can't do that with a conventional lens while rotating. Sorry for the screw-up.

Click to expand...

dsjtecserv said:

I'm not attempting to recommend any lens or technique, just making some observations on their characteristicscharacteristics

Click to expand...

dsjtecserv said:

.

Dave

--
http://www.pbase.com/dsjtecserv

Click to expand...