Do Tilt shift lenses correct perspective or distort

ChrisR · Post by **ChrisR** » Sat Sep 17, 2016 4:38 am

Until further notice, Distortion is where straight lines become curved. Bye bye!

ray_parkhurst · Post by **ray_parkhurst** » Sat Sep 17, 2016 9:38 am

AK, I think this misunderstanding is very simple to explain:

Viewing a coin with a normal lens with the coin at an angle, makes the coin appear oval. The coin is not oval, so this view makes the coin appear stretched out of shape.

Viewing the coin through a properly-configured shifted lens makes the coin appear round. Thus shifting has the effect of stretching the shape of the coin back to round as compared with viewing the coin with normal lens.

elf · Post by **elf** » Sat Sep 17, 2016 2:30 pm

ray_parkhurst wrote: Viewing a coin with a normal lens with the coin at an angle, makes the coin appear oval. The coin is not oval, so this view makes the coin appear stretched out of shape.

But, the coin is oval when viewed at an angle. There is no stretching involved.

ray_parkhurst · Post by **ray_parkhurst** » Sat Sep 17, 2016 3:02 pm

elf wrote:
ray_parkhurst wrote: Viewing a coin with a normal lens with the coin at an angle, makes the coin appear oval. The coin is not oval, so this view makes the coin appear stretched out of shape.

But, the coin is oval when viewed at an angle. There is no stretching involved.

This is true. But we know the coin itself is round and that's how we expect it to appear. If it appears oval, then it appears as if the coin has been stretched out of round versus what is expected. No actual stretching has occurred, it just appears that way. We can use the shift technique to "correct" this apparent stretching so that the coin will appear round, as we expect it to. The shifted (round) image has the appearance of being stretched versus the non-shifted (oval image).

ray_parkhurst · Post by **ray_parkhurst** » Sat Sep 17, 2016 4:02 pm

austrokiwi1 wrote:I actually personally prefer Rays original ( held for years) point of view that Key-stoning produces distorted images and therefore Tilt and, I believe to be consistent, shift correct the perceptual distortion.

AK...Nothing I have said is contradictory or inconsistent. And indeed, I fully agree with the statement above. Keystoning does produce distorted images and shift can be used to correct this perceptual distortion, doing so by distorting the image back to its expected shape. Your subject of this thread asks if shift corrects perspective, or distorts. The answer is "both"...it corrects perspective by distorting the image.

rjlittlefield · Post by **rjlittlefield** » Sat Sep 17, 2016 4:11 pm

ray_parkhurst wrote:it corrects perspective by distorting the image.

And yet shooting with shift lens does exactly the same thing as cropping from wideangle.

So (the prof asks...) exactly where does the distortion occur? Does shooting with wideangle distort? Does cropping distort?

--Rik

ChrisR · Post by **ChrisR** » Sat Sep 17, 2016 4:41 pm

ray_parkhurst · Post by **ray_parkhurst** » Sat Sep 17, 2016 5:44 pm

rjlittlefield wrote:
ray_parkhurst wrote:it corrects perspective by distorting the image.
And yet shooting with shift lens does exactly the same thing as cropping from wideangle.

So (the prof asks...) exactly where does the distortion occur? Does shooting with wideangle distort? Does cropping distort?

--Rik

Sorry, I left out the implied phrase "vs the uncorrected image". The uncorrected image is the natural image looking directly at the coin, ie the one where the coin appears oval.

The act of shifting the lens causes the image to distort vs the natural image. A little bit of shift causes a little bit of distortion, making the oval coin is a little less oval. Shifting such that the coin and sensor planes are parallel will distort the image sufficiently to make the coin appear round.

If I made a video that goes from natural image, to a little shift, then a little more, then full shift, then back again, the coin would change appearance from oval to round and back again.

rjlittlefield · Post by **rjlittlefield** » Sat Sep 17, 2016 6:49 pm

ray_parkhurst wrote:The act of shifting the lens causes the image to distort vs the natural image.

Hhmm, interesting...

It seems to me that shifting the lens just shifts the image on the sensor. Is that not correct? If it is correct, then how does that cause the image to distort?

--Rik

ChrisR · Post by **ChrisR** » Sat Sep 17, 2016 8:19 pm

Ray there's little point endlessly describing what happens, it's kindergarten stuff that everyone knows. What people are asking you for is precise definitions and what terminology you appropriate for exactly what, and why.
That would be Misappropriate, for some of us.

Otherwise you're just saying the image alters with camera position/angle. This we know. The coin won't look round unless the sensor's parallel to it, - we know that too.

ray_parkhurst · Post by **ray_parkhurst** » Sat Sep 17, 2016 9:10 pm

rjlittlefield wrote:
ray_parkhurst wrote:The act of shifting the lens causes the image to distort vs the natural image.
Hhmm, interesting...

It seems to me that shifting the lens just shifts the image on the sensor. Is that not correct? If it is correct, then how does that cause the image to distort?

--Rik

Yes, of course it is correct. I should have been more precise. My saying "shifting the lens" was an oversimplification of "imaging the coin using the lens shift technique". Shifting the lens will move the image toward a different vantage point, so the camera must also be pointed away from the coin in order to keep the coin in the frame. This changes the relative tilt between the coin and the sensor. Of course the technique is actually applied in opposite order. It is known that the sensor and coin should be parallel, so the camera is pointed such that the sensor and coin planes are parallel, and the lens is shifted the appropriate amount to move the coin into the frame. So what is really happening is the coin is tilted vs the sensor in the natural image, while it is flat vs the sensor in the shifted image. It is the difference in tilt that makes the coin appear different between the two images.

austrokiwi1 · Post by **austrokiwi1** » Sun Sep 18, 2016 12:14 am

This changes the relative tilt between the coin and the sensor

and.....

So what is really happening is the coin is tilted vs the sensor in the natural image, while it is flat vs the sensor in the shifted image.

RAY: As I have understood your last post; those particular quotes infer that shift changes the relationship between the subject and sensor. If that is indeed your meaning then, as far as everything I have read, viewed online and more importantly experienced, it is incorrect.

My experience has been that if I set up my camera so the sensor is tilted is respect to the subject I am going to get Key-stoning. If I then use shift I still get Key- stoning. Shift doesn't change the sensor subject relationship. Tilt on the other hand does change the relationship. As I understand it (Note: I could be wrong with regards to tilt but I don't want to get into a long discussion about it) Tilt has that apparent effect by by adjusting the field of focus with respect to the sensor! Personally I don't like using Tilt in this manner with coins)

Have a look at this video from Adorama (to save time move to the 9.15 minute mark): you can see clearly in that demo that as long as the sensor and subject are kept parallel shift has no effect on the presence-of / lack-of key-stoning. If you angle the sensor with respect to the subject you get Key-stoning whether you apply shift or not.

https://www.youtube.com/watch?v=9TTB6hKxKzE

Edit : when I set up a pseudo-Axial lighting using shift, for coins. I place the coin ensuring that it is level and centered in the cameras sensor( without shift applied). I usually check to ensure the camera is also leveled( pointed down at the coin) Once I am sure the sensor and coin are parallel I then apply shift. This way I can get the maximum shift without having the dark edge of the image circle intruding into the field of view. The coin at this point is usually well out of the center of view I then move it so, with respect to what is seen in the view finder, the coin is centered. This set up process works well for me, and it is very similar( I would say the same) to the set up shown in the Adorama video (at the point the camera is position so it is facing the center of the rectangular target) Then I place the lighting.

From experience I have found that this is the most effective process(for me). However it is dependent on long focal length lenses. I first tried this with the Mamiya 645 80mm macro. I recall when I first tried it out I thought the 80 mm and shift was wonderful It was only after time I realized the lens its self wasn't as sharp as I would like and the 80 mm soon became unused. Up till a few weeks ago I relied on the Oly 135 bellows lens but the color correction was not great and the image circle that lens projected wasn't quite as big as the Mamiya 645 lenses. I switched to the mamiya 645 120mm F4. Its a huge improvement but it still isn't that great. The lens housing is physically too thick, at 1X the lens has doubled in length and it is big and clunky. Furthermore my example shows IQ deterioration at 1-1. When used in combination with shift with coins it performs best at 0.6X magnification and less. I am now suspecting that at low magnification the DOF is sufficient to render any fault with the lens unimportant. This means on my full frame camera I am only using shift with the 120mm on coins and medals of at least 40mm in diameter

Ray, you and some other coin photographers appear to prefer lenses of around 80mm in focal length. My experience has been that shift with such lenses is not practical. in which case I believe your preferred technique of tilt is better

ray_parkhurst · Post by **ray_parkhurst** » Sun Sep 18, 2016 5:12 am

austrokiwi1 wrote:RAY: As I have understood your last post; those particular quotes infer that shift changes the relationship between the subject and sensor. If that is indeed your meaning then, as far as everything I have read, viewed online and more importantly experienced, it is incorrect.

If you keep the sensor and coin parallel, then as you shift the lens, you need to move the coin (physically) to keep it centered in the frame. So the physical relationship changes.

If you keep the coin centered in the frame, then as you shift the lens, you need to tilt (rotate?) the camera+lens to keep the coin centered in the frame. So again the physical relationship changes.

austrokiwi1 wrote:Edit : when I set up a pseudo-Axial lighting using shift, for coins. I place the coin ensuring that it is level and centered in the cameras sensor( without shift applied). I usually check to ensure the camera is also leveled( pointed down at the coin) Once I am sure the sensor and coin are parallel I then apply shift. This way I can get the maximum shift without having the dark edge of the image circle intruding into the field of view. The coin at this point is usually well out of the center of view I then move it so, with respect to what is seen in the view finder, the coin is centered.

This process seems well-reasoned.

austrokiwi1 wrote:Ray, you and some other coin photographers appear to prefer lenses of around 80mm in focal length. My experience has been that shift with such lenses is not practical. in which case I believe your preferred technique of tilt is better

Yes, I prefer 75-80mm, physically smaller lenses (enlarging, duplicating). Short lenses have shorter working distances so I can set things up using a small microscope stand. The small size can still allow shift since you don't need to move the coin very far to move away from the lens.

Note that the so-called tilt adapters, when attached to the camera, don't just tilt. Because their tilt axis is well in front of the sensor, tilting also causes shifting of the sensor. So as I tilt the camera, I must move the coin to compensate the shift. The combined effect works very well.

austrokiwi1 · Post by **austrokiwi1** » Sun Sep 18, 2016 5:46 am

So as I tilt the camera, I must move the coin to compensate the shift

Yep that is what I referred to as your tilt method

If you keep the coin centered in the frame, then as you shift the lens, you need to tilt (rotate?) the camera+lens to keep the coin centered in the frame. So again the physical relationship changes.

Edit: I actually hadn't read Rays reply properly. Ray you still have not come to terms with how shift works. Look at the Video: with shift used as I do and as demonstrated in the video there is absolutely no need to use tilt ( or rotation). Tilt is completely unnecessary and, if used, would lead to key-stoning!!!

what I do with coins is covered from 9.15 to 11.12. in the video. The rest applies to buildings and I suspect you have become confused by the instructors reference to "rotation" when He has demonstrated what happens with a building.

You have convinced me on one thing... that you have never had any experience using shift!!

Yes but even photographing building and statues with shift often requires that you move back to find the best position. The key is shift allows you to find a position in which the sensor and image remain parallel while at the same time avoiding Key-stoning.