Ad
Ad
Ad
Pages: [1] 2 3 ... 6 »   Bottom of Page
Print
Author Topic: DoF and Perspective Revisited  (Read 15459 times)
Nick Rains
Sr. Member
****
Offline Offline

Posts: 704



WWW
« on: September 04, 2009, 06:28:54 PM »
ReplyReply

A recently closed thread - which degenerated into acrimony - was 'discussing' whether a wide lens has the same perspective as a series of stitched, longer focal length images covering the same FOV from the same postion.

Now, I know the theory from books, and I've been shooting for 26 years so I know the real world too. However, I'm now puzzled and could do with enlightenment without being yelled at.

DoF - this is the same for any lens at a given f stop and depends entirely on the subject distance from the camera. That much I know and I think we all agree on both the theory and the practice.

But the perspective, if that's the right word, is not the same with a single wide lens as it is with a series of stitched narrower FOV shots. I was fiddling around with a zoom lens - 17-40 - and the relationship of objects in corners of the 17mm setting looks totally different compared to the same items from the same position at the 40mm setting. I can only conclude that a single shot with the 17mm will look completely different than that from a stitched series using the 40mm setting taken from the same position.

This is what Guy Mancuso was alluding to, before he was shouted down.

I read the thread and was going with ThierryH's theory at first until I read Guy's post with the 28mm interior shot divided into 6 panels. It's quite clear that the look of each of those panels is not the same as from a lens of longer FL that covers the same region of the subject.

I'm not totally sure why, but it's clearly the case - maybe it's to do with the fact that a wide lens has 'bend' the light a lot to project such a wide FOV onto the sensor as a single image, whereas the longer lens has to bend the light less - less distortion.

In fact the more I think about it, the more sure I am that it's to do with having to move the camera to take the stitched images. Therefore the relationship between the camera and the subject is not the same as with a single capture. The only stitched images that have the same persepective as a single capture with the same FOV are those captured with a shift lens where the camera is not moved, only the lens. I suggest that a (hypothetical) 90mm shift lens which could be shifted to cover the same total FOV as a 28mm lens would look exactly the same as single capture from the 28mm lens. OTOH, a series from a fixed 90mm lens where the camera had to be rotated would not look the same.

I'm no optical expert but I know what I see. The perspective similarity between stitched images and equivalent wide lenses is mostly a myth. It's only true if the camera does not  move at all - not just its position but it's orientation to the subject.
Logged

Nick Rains
Australian Photographer
Leica Akademie Instructor
www.nickrains.com
Graham Mitchell
Sr. Member
****
Offline Offline

Posts: 2282



WWW
« Reply #1 on: September 04, 2009, 06:35:57 PM »
ReplyReply

Are you referring to geometric distortion? This distortion may not appear in the individual shots with the longer focal length lens, but when the images are stitched together and the rectilinear geometric transformation applied, the end result will be the same. (This ignores other issues such as the polyhedral focus of stitched images versus the normal focal 'plane').
« Last Edit: September 04, 2009, 06:36:46 PM by foto-z » Logged

Graham Mitchell - www.graham-mitchell.com
Panopeeper
Sr. Member
****
Offline Offline

Posts: 1805


« Reply #2 on: September 04, 2009, 06:38:15 PM »
ReplyReply

Quote from: Nick Rains
I was fiddling around with a zoom lens - 17-40 - and the relationship of objects in corners of the 17mm setting looks totally different compared to the same items from the same position at the 40mm setting. I can only conclude that a single shot with the 17mm will look completely different than that from a stitched series using the 40mm setting taken from the same position.

Nick, what about posting those images for others to chew on, in good resolution and quality if possible, together with the stitched result?
Logged

Gabor
Nick Rains
Sr. Member
****
Offline Offline

Posts: 704



WWW
« Reply #3 on: September 04, 2009, 06:40:26 PM »
ReplyReply

Quote from: foto-z
Are you referring to geometric distortion? This distortion may not appear in the individual shots with the longer focal length lens, but when the images are stitched together and the rectilinear geometric transformation applied, the end result will be the same. (This ignores other issues such as the polyhedral focus of stitched images versus the normal focal 'plane').
Maybe that's the term I was fishing around for. I believe you have it right here.
Logged

Nick Rains
Australian Photographer
Leica Akademie Instructor
www.nickrains.com
Nick Rains
Sr. Member
****
Offline Offline

Posts: 704



WWW
« Reply #4 on: September 04, 2009, 06:53:24 PM »
ReplyReply

Quote from: Panopeeper
Nick, what about posting those images for others to chew on, in good resolution and quality if possible, together with the stitched result?

This was the thread:

http://luminous-landscape.com/forum/index....40&start=40

The image I'm referring to is towards the end of P3.

I didn't shoot any images with the zoom, just looked through the viewfinder; grab a camera and you'll see what I mean. If I have time I'll shoot the frames but I'm ducking back to the computer between Saturday morning chores around the house!
« Last Edit: September 04, 2009, 06:57:32 PM by Nick Rains » Logged

Nick Rains
Australian Photographer
Leica Akademie Instructor
www.nickrains.com
JeffKohn
Sr. Member
****
Offline Offline

Posts: 1671



WWW
« Reply #5 on: September 04, 2009, 07:12:16 PM »
ReplyReply

First it's probably a good idea to agree on some terminology. What I would call geometric distortion has nothing to do with perspective or even focal length, except to the extent that certain focal lengths tend to have certain types of geometric distortion due to lens design contraints. Namely, telephotos often have pincushion, while wide-angle lenses are more likely to have barrel distortion. These distortions are optical flaws in lenses though, and have nothing to do with perspective. In fact two 20mm lenses can have varying amounts of barrel distortion.

Geometric distortion is not what causes elements at the edge of a wide-angle shot to be stretched out. That's rectilinear distortion, or perspective distortion (I've seen it called both). Rectilinear distortion has everything to do with perspective and field of view. Note I said field of view, not focal length. Whether you get that perspective and field of view from a single wide-angle shot or by shooting several shots with a longer lens and then stitching them using a rectilinear projection, the resulting perspective will be the same.

It's true that if you just take several shots with a longer lens and align them you won't get the same look as a wide-angle shot. But that's not what pano software does; it projects the images, and as part of that process the outer portions of the resulting image will get stretched just like they do in a wide-angle shot. The wider the field of view, the more the edges are stretched; which is why even though you can shoot a much wider FOV with stitching than with the widest of lenses, you probably won't want to use a rectilinear projection for extreme FOV's.  Note that other projections such as cylindrical or spherical do not produce the same results as a rectilinear projection.

You don't need panos/stitching to prove that perspective is determined solely by position and not focal length. Take a zoom lens such as the 17-40 and mount it on a tripod. Take a shot at 17mm, and another at 40mm. On the computer, crop the 17mm image so that it covers the same part of the scene as the 40mm image, and you'll find that the perspective of the two resulting images (eg the spatial relationships and relative sizes of objects in the scene) will be exactly the same. I think DPReview even has an article covering this with sample shots if you don't feel like doing the experiment yourself. It's incontrovertible fact, and it really does amaze me how many photographers refuse to grasp this concept. Usually it's 35mm photographers who had never shot any other formats before digital came along, and you'll hear them saying something like  "I don't like DX format because I can't use a 60mm lens to get the same pleasing portrait perspective that my 85mm lens gave me on 35mm". Try shooting that 85mm portrait on 4x5" film and then tell me that it's the focal length that determines the perspective....
Logged

Panopeeper
Sr. Member
****
Offline Offline

Posts: 1805


« Reply #6 on: September 04, 2009, 07:14:01 PM »
ReplyReply

Quote from: Nick Rains
I didn't shoot any images with the zoom, just looked through the viewfinder; grab a camera and you'll see what I mean

I thought you actually made the shots. As I understand it now, you found the corner of the 17mm shot visually different from the 40mm shot of that corner. You were right.

1. When you make a shot at 17mm and then at 40mm from the same position, in the same direction, then the respective crop of the 17mm image will be identical (regarding the perspective) to the 40mm shot. Geometric distortion may play some role, but that's not "perspective".

2. When you make a 17mm shot and then you shoot the corner of that frame with 40mm, you have to adjust the camera's direction. The lens is rectilinear, it captures a rectangle perpendicular to the lens' axis. This will look very different from the corner of the 17mm shot.

It is the job of the stitcher to warp the source frames, so that they look like they had been part of a larger frame, presumed you choose rectilinear projection (as its name suggests, the rectilinear projection creates a single rectilinear mosaic/pano). Thus the result of stitching won't look like what you saw in the viewfinder.


Added: there is lots of overlap with Jeff's post above; we must have been typing at the same time.
« Last Edit: September 04, 2009, 07:16:13 PM by Panopeeper » Logged

Gabor
bill t.
Sr. Member
****
Offline Offline

Posts: 2693


WWW
« Reply #7 on: September 04, 2009, 07:39:42 PM »
ReplyReply

Quote from: Panopeeper
the result of stitching won't look like what you saw in the viewfinder.
Thanks for that, great way to putting it!  More than once have had this discussion with other photographers.  My response is always "well, just try it, you'll be convinced!"  Apparently nobody wants to do anything more than just look through the viewfinder.

Should also be noted that most stitchers now offer extremely useful projections and hybrid projections that are simply unavailable with any lens.  Some of these easily solve thorny problems with super wide shots.
Logged
JeffKohn
Sr. Member
****
Offline Offline

Posts: 1671



WWW
« Reply #8 on: September 04, 2009, 07:56:59 PM »
ReplyReply

Quote from: bill t.
Should also be noted that most stitchers now offer extremely useful projections and hybrid projections that are simply unavailable with any lens.  Some of these easily solve thorny problems with super wide shots.
Yes, I really like the Vedutismo projection in PTGui for landscapes. I tend to use that for anything too wide to look natural with a rectilinear projection.
Logged

Murray Fredericks
Sr. Member
****
Offline Offline

Posts: 290



WWW
« Reply #9 on: September 04, 2009, 08:00:25 PM »
ReplyReply

Just to confuse things further, the geometric 'look' of the final stitched image will depend on the kind of projection chosen to render the stitch into a single image.

I have assumed (never tested) that a planar projection of a stitched set of images from a longer lens that covers the same FOV as a wide lens will have the same geometric properties as the single frame from the same wide lens.

The Geometric 'look' (maybe sometimes confused with perspective?) will vary quite a lot if a spherical or cylindrical projection is chosen.

Cheers

Murray
Logged

Wayne Fox
Sr. Member
****
Offline Offline

Posts: 2883



WWW
« Reply #10 on: September 04, 2009, 08:04:27 PM »
ReplyReply

Nick,

I was sorry to see that thread degenerate to that point as well.  I believe the problem some were having with it is confusing distortion and perspective.

I also believe the problem some are having with this concept is related to how we are so entrenched in using focal length to control perspective, but we forget to do so we also have to change positions.  If we do not change positions, we actually don't change perspective, we change the field of view. But because we are so entrenched in the idea of using longer lenses to change perspective, we  believe using stitching to increase the field of view to match, we will get the benefit if the perspective change from using the longer lens ... it sort of sounds logical.  It doesn't work that way ...  perspective only changes when the camera position changes, not when the focal length changes.

In effect when using stitching we are increasing the size of the the sensor  ... we are creating a "larger format" camera and thus require a longer lens and several stitched images to match the FoV. but  to change perspective you still have to change the position of the camera.  You cannot use stitching with a longer lens as a method to change perspective.  It really is just the basic physics.

The picture you mentioned is a good example.  Were that picture taken with a longer lens capturing only 1/6th of the field of view, and then stitched, the perspective would remain the same.  What in effect you are doing is using a larger format camera (stitched = greater sensor size), and thus needing a longer lens to narrow the FoV to match.

Distortion can certainly be a problem, but distortion and perspective are not the same thing . This is especially the case with close foreground objects and/or wide angle lenses. But if the lens distorted the perspective, it isn't actually "valid" (if I dare use that term).  IF you correct out the distortion, then the images match.

I never thought about this much, but accidently learned this by first hand experience several years ago.  I was taking a picture using of a golf green in the foreground and the clubhouse in the background.  I wanted to change the perspective to increase the size of the clubhouse in relation to the green  but couldn't get further away (without a helicopter).  So I decided to use a 200mm lens to stitch an image together, believing it would accomplish the same thing.  To my surprise, the final stitched shot was virtually identical to the one I took with my 80mm lens.  Because neither focal length was a wide angle  there wasn't even any evidence of problems with distortion.  So the only way I could change the perspective of the golf hole to the clubhouse would be to use the 200mm lens, and move farther away until the golf hole was the same size.  Only then would my clubhouse become larger in perspective.  Since I didn't have a helicopter at my disposal I didn't get the shot.  

I hope I'm not being over simplistic.  It's sure easy to go out and try it for yourself.
Logged

Nick Rains
Sr. Member
****
Offline Offline

Posts: 704



WWW
« Reply #11 on: September 04, 2009, 08:06:08 PM »
ReplyReply

Quote from: Panopeeper
It is the job of the stitcher to warp the source frames, so that they look like they had been part of a larger frame, presumed you choose rectilinear projection (as its name suggests, the rectilinear projection creates a single rectilinear mosaic/pano). Thus the result of stitching won't look like what you saw in the viewfinder.

Thanks, I guess that's the missing point - stitching software will warp images to match the chosen projection and if you want a flat projection taken with multiple shots it will warp the corners to look like that of an equivalent wide angle lens. The w/a is doing the same job, 'warping' the image into a flat projection onto the sensor.

If a cylindrical projection works for you, then you'll get a different look, more like the old Noblex cameras used to get.

This is where the "textbook" theories of perspective, distortion etc get confusing - they have to be tempered with what you are actually doing with the shots in post i.e. how they will be projected onto the flat surface of a screen or print.
« Last Edit: September 04, 2009, 08:06:56 PM by Nick Rains » Logged

Nick Rains
Australian Photographer
Leica Akademie Instructor
www.nickrains.com
BernardLanguillier
Sr. Member
****
Offline Offline

Posts: 8188



WWW
« Reply #12 on: September 04, 2009, 08:28:17 PM »
ReplyReply

Quick test in a corner of my living room (full size samples).

- Comparison between single frame with 24 mm lens and 3 rows pano with 60 mm lens,
- The camera was in the exact same position, I kept it on the pano head in both samples, just switched the lens and set the pano head to central position for the 24 mm lens,
- distance to the subject is less than 1 meter (about 70 cm),
- Focus was set with live view on the book next to the fan,
- Both images were shot in M mode at f8,
- The pano was computed with PTgui in flat projection mode,
- It took about 40 seconds to shoot the pano without trying to be utterly fast, this is the result of my first and only trial,
- The computation of the pano took less than 10 minutes for this 120 megapixel image.

- single image from a 24 mm lens, slightly cropped to match the stitch result (22 megapixel):



(link to jpg quality 8, 2.5 MB full size image) http://www.flickr.com/photos/bernardlangui...923847/sizes/o/

- pano made with a 60mm lens, 3 rows of 4 images (120 megapixel).



(link to jpg quality 7, 7.5 MB full size image) http://www.flickr.com/photos/bernardlangui...724060/sizes/o/

What I see:
- the sensor of my d3x needs cleaning,
- the images are very similar,
- the pano image has less DoF since it was shot with a longer lens.
- the 24 mm lens has some geometrical distorsion that is not visible in the pano image. this is not a simple spherical one and results in subtle differences in rendition of some of the objects. Both scenes would probably be assessed by 99% of people to be totally faithful to the original.
- the 24 mm lens has some light fall off event at f8 (see upper left corner),
- the color rendition of the 24mm appears to be a bit more neutral than that of the 60mm, although this difference could also have been introduced by the pano software,
- the sharpness per pixel of the pano is on par with that of the single image and can be considered to be excellent in absolute terms.

My personal conclusion are:
- real world usage matches the theory very well, flat projection panos are difficult to distinguish from single frames shot with a wide lens,
- when high resolutions are needed for an application pano are a handy solution, the impact in shooting time is small (think of all the other things you would have had to do before taking this image) but processing time can be a problem.

Cheers,
Bernard
« Last Edit: September 05, 2009, 06:19:38 AM by BernardLanguillier » Logged

A few images online here!
Panopeeper
Sr. Member
****
Offline Offline

Posts: 1805


« Reply #13 on: September 04, 2009, 08:59:06 PM »
ReplyReply

Attached the top left and the top right corner shots of a 2x7 shot pano, in rectilinear projection. This is the work of the warping. These show, that the shot of the corners captured a larger area than the wider lens would have captured. This means, that only a part of this frame becomes part of the pano, and that part needs to be stretched in order to achieve the same proportions as a single rectilinear shot would have created.
Logged

Gabor
Dick Roadnight
Sr. Member
****
Offline Offline

Posts: 1730


« Reply #14 on: September 05, 2009, 03:29:25 AM »
ReplyReply

Quote from: Nick Rains
The only stitched images that have the same persepective as a single capture with the same FOV are those captured with a shift lens where the camera is not moved, only the lens.
This also applies to images stitched using a sliding back (or rear standard movements) on a view camera, where not the lens, but the sensor is moved. This produces much better results if any part of the subject is near the lens.

...but the theory is that the stitching software e-projects the images to try to correct the perspective.
Logged

Hasselblad H4, Sinar P3 monorail view camera, Schneider Apo-digitar lenses
BernardLanguillier
Sr. Member
****
Offline Offline

Posts: 8188



WWW
« Reply #15 on: September 05, 2009, 06:22:08 AM »
ReplyReply

Quote from: Dick Roadnight
This also applies to images stitched using a sliding back (or rear standard movements) on a view camera, where not the lens, but the sensor is moved. This produces much better results if any part of the subject is near the lens.

...but the theory is that the stitching software e-projects the images to try to correct the perspective.

Do you have any concrete comparison showing that the leading softwares are unable to do it? I believe that my post above in this thread clearly shows that there is at least one example where they do an excellent job.

My question is geniune, I have never seen such a case, but it could very well exist.

Cheers,
Bernard
Logged

A few images online here!
ErikKaffehr
Sr. Member
****
Offline Offline

Posts: 7640


WWW
« Reply #16 on: September 05, 2009, 06:50:07 AM »
ReplyReply

Hi,

The way I see it if we discuss really wide images rectilinearity is an illusion. It comes from the way the brain interprets images, it knows that structures are rectilinear. Add to this that the non peripheral field of seeing is rather small, what we see sharp is always within a small angle.

To keep parallel lines straight and parallel we need to to distort (stretch out) images which are of axis (that is seen with a large angle of view). So for wide angle panoramas we either have to stretch out the peripheral part, thus distorting surfaces, or accept curved and converging parallels by using cyclindrical or spherical projection.

A rectilinear wide angle does this stretching internally, a panorama camera using a swinging lens gives a cylindrical panorama and a fish eye a spherical one.

When stitching we can choose either mapping, except for very wide angles where rectilinear projection is not possible because it would need infinite stretching.

Best regards
Erik


Quote from: Nick Rains
A recently closed thread - which degenerated into acrimony - was 'discussing' whether a wide lens has the same perspective as a series of stitched, longer focal length images covering the same FOV from the same postion.

Now, I know the theory from books, and I've been shooting for 26 years so I know the real world too. However, I'm now puzzled and could do with enlightenment without being yelled at.

DoF - this is the same for any lens at a given f stop and depends entirely on the subject distance from the camera. That much I know and I think we all agree on both the theory and the practice.

But the perspective, if that's the right word, is not the same with a single wide lens as it is with a series of stitched narrower FOV shots. I was fiddling around with a zoom lens - 17-40 - and the relationship of objects in corners of the 17mm setting looks totally different compared to the same items from the same position at the 40mm setting. I can only conclude that a single shot with the 17mm will look completely different than that from a stitched series using the 40mm setting taken from the same position.

This is what Guy Mancuso was alluding to, before he was shouted down.

I read the thread and was going with ThierryH's theory at first until I read Guy's post with the 28mm interior shot divided into 6 panels. It's quite clear that the look of each of those panels is not the same as from a lens of longer FL that covers the same region of the subject.

I'm not totally sure why, but it's clearly the case - maybe it's to do with the fact that a wide lens has 'bend' the light a lot to project such a wide FOV onto the sensor as a single image, whereas the longer lens has to bend the light less - less distortion.

In fact the more I think about it, the more sure I am that it's to do with having to move the camera to take the stitched images. Therefore the relationship between the camera and the subject is not the same as with a single capture. The only stitched images that have the same persepective as a single capture with the same FOV are those captured with a shift lens where the camera is not moved, only the lens. I suggest that a (hypothetical) 90mm shift lens which could be shifted to cover the same total FOV as a 28mm lens would look exactly the same as single capture from the 28mm lens. OTOH, a series from a fixed 90mm lens where the camera had to be rotated would not look the same.

I'm no optical expert but I know what I see. The perspective similarity between stitched images and equivalent wide lenses is mostly a myth. It's only true if the camera does not  move at all - not just its position but it's orientation to the subject.
Logged

Jack Flesher
Sr. Member
****
Offline Offline

Posts: 2595



WWW
« Reply #17 on: September 05, 2009, 08:42:32 AM »
ReplyReply

It's why there was the argument yesterday -- I think folks are talking apples to oranges to bananas.  You have perspective, lens distortions and 3D>2D image projections.  They all affect how the image will appear.  

Regardless, the fact remains that for visual PERSPECTIVE, the only thing that affects it are the relative positions of the shooter to subject.  Lens distortions and/or projection conversion will alter how the image appears but will not alter that relative perspective...
Logged

Craig Lamson
Sr. Member
****
Offline Offline

Posts: 773



WWW
« Reply #18 on: September 05, 2009, 09:00:30 AM »
ReplyReply

Quote from: Jack Flesher
Regardless, the fact remains that for visual PERSPECTIVE, the only thing that affects it are the relative positions of the shooter to subject.  Lens distortions and/or projection conversion will alter how the image appears but will not alter that relative perspective...

(Bolding mine)


Ding Ding Ding...we have a winner....
« Last Edit: September 05, 2009, 09:01:05 AM by infocusinc » Logged

Craig Lamson Photo
www.craiglamson.com
gwhitf
Sr. Member
****
Offline Offline

Posts: 820


« Reply #19 on: September 05, 2009, 09:23:09 AM »
ReplyReply

To Nick Rains,

I stitch ALL the time. I went through a period where I stitched every frame I shot, even people shots, even handheld people shots. I did it with the Canon 45TS. That's the only lens I have any experience with, stitching. But I love the results. I don't care about any of these fancy terms about distortion -- call it what you want. I could care less. I only care about the photograph and the resulting image. I just know that the 45 stitched produces a much more pleasing image than moving in closer with a wide angle, with its "womp factor". I like to stay back a bit, and leave things flatter optically, and then stitch by shifting the 45. Ideally, you'd Liquid Nails a tripod mount onto the 45 lens, and mount the lens onto the tripod, rather than the body. That way, if on a tripod, it would line up to the pixel, in post. But even with the body mounted, (wrong technique), it's pretty easy to stitch in post if you have large overlap. I simply prefer the flatter look, even when you add in the additional post work.

I stitch with the 80 too, on the H2, but since it's not a shift lens, it's a much bigger task, even with that stupid RRS attachment. Simply put, if you pan the body, you're asking for trouble, in post. No matter what gadget or nodal point you're using. Just one opinion.
Logged
Pages: [1] 2 3 ... 6 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad