Ad
Ad
Ad
Pages: [1] 2 »   Bottom of Page
Print
Author Topic: shift lenses for MF digital  (Read 8787 times)
BJL
Sr. Member
****
Offline Offline

Posts: 5146


« on: January 10, 2007, 03:28:57 PM »
ReplyReply

MF digital is rather lacking in shift lens options, so I have a "modest proposal": use sensors in a format like 33x44mm (as in the delayed Pentax system) with existing wide angle lenses having the larger image circle of 645 format, and shift the sensor around within the image circle.

This would seem to reproduce what happens with a shift lens: the lens has an image circle larger than the format, and shifting the lens allows one to record a rectangular region from various parts of that over-sized image circle.

Putting that somewhat smaller sensor into a 645 sized body and moving it with servo-motors seems feasible. The motors moving the sensor could perhaps also provide sensor based anti-shake, and maybe even Sony A-100 style dust removal!

Sensor based rear tilts and swings also seem possible, doable with any lens as increased image circle size is not needed.
Logged
howiesmith
Guest
« Reply #1 on: January 10, 2007, 03:58:55 PM »
ReplyReply

Quote
This would seem to reproduce what happens with a shift lens: the lens has an image circle larger than the format, and shifting the lens allows one to record a rectangular region from various parts of that over-sized image circle.

[a href=\"index.php?act=findpost&pid=94979\"][{POST_SNAPBACK}][/a]

True.  I have a question though.  While there may be enough image circle, doesn't lens performance for many MF lenses drop off significantly as you approach the unshifted image circle and dramatically as you go beyond the unshifted image circle?  Couldn't the image taken through the side if the shifted MF lens be rathr poor?
Logged
peterhorsley
Newbie
*
Offline Offline

Posts: 23


« Reply #2 on: January 10, 2007, 09:00:33 PM »
ReplyReply

Quote
Couldn't the image taken through the side if the shifted MF lens be rathr poor?
[a href=\"index.php?act=findpost&pid=94983\"][{POST_SNAPBACK}][/a]



There's a lot of extra coverage to use.

If it was a feature of a DB another use could be for automated stitching for 'panoramic' purposes or to regain the designed field of view the lens.

Even 36mm x 48mm sensors on a 6x6 body could make use of the shift movements.  The 'Blad SWC would get a whole new lease on life.

Sorting out the SLR's view finder might might be clumsy but workable for shifting.  Tilt could be tricky.  This has to be easier than designing and selling a fleet of T/S lenses.

Neat idea.

Peter
« Last Edit: January 10, 2007, 09:03:43 PM by peterhorsley » Logged
Ray
Sr. Member
****
Offline Offline

Posts: 8887


« Reply #3 on: January 10, 2007, 09:07:34 PM »
ReplyReply

Quote
MF digital is rather lacking in shift lens options, so I have a "modest proposal": use sensors in a format like 33x44mm (as in the delayed Pentax system) with existing wide angle lenses having the larger image circle of 645 format, and shift the sensor around within the image circle.
[a href=\"index.php?act=findpost&pid=94979\"][{POST_SNAPBACK}][/a]

BJL,
If the lens is really, really high quality right to the edges, this would be a good idea.

My experience with the Canon TS-E 24, is that, not only is the lens not particularly sharp in the centre, it's woefully inadequate at the edges of the image circle, when shifted.
Logged
Ray
Sr. Member
****
Offline Offline

Posts: 8887


« Reply #4 on: January 10, 2007, 09:15:42 PM »
ReplyReply

In any case, we already have a method of shifting the sensor with the use of L-brackets (from Really Right Stuff). The camera on its tripod is shifted in the opposite direction to the lens shift with each exposure.
Logged
Kirk Gittings
Sr. Member
****
Offline Offline

Posts: 1550


WWW
« Reply #5 on: January 10, 2007, 10:22:46 PM »
ReplyReply

Quote
My experience with the Canon TS-E 24, is that, not only is the lens not particularly sharp in the centre, it's woefully inadequate at the edges of the image circle, when shifted.

This is not my experience at all. My 24 TS is the second best 35mm lens that I own, my "go to" lens when I am shooting architecture in DSLR.
« Last Edit: January 10, 2007, 10:25:07 PM by Kirk Gittings » Logged

Thanks,
Kirk

Kirk Gittings
Architecture and Landscape Photography
WWW.GITTINGSPHOTO.COM

LIGHT+SPACE+STRUCTURE (blog)
John Camp
Sr. Member
****
Offline Offline

Posts: 1258


« Reply #6 on: January 10, 2007, 11:22:24 PM »
ReplyReply

Can somebody tell me why fixing it in Photoshop doesn't work better than shift lenses? I know about using tilt for DOF, which Photoshop can't do, but the shift stuff seems to work fairly well...or not?

JC
Logged
Ray
Sr. Member
****
Offline Offline

Posts: 8887


« Reply #7 on: January 10, 2007, 11:26:33 PM »
ReplyReply

Quote
This is not my experience at all. My 24 TS is the second best 35mm lens that I own, my "go to" lens when I am shooting architecture in DSLR.
[a href=\"index.php?act=findpost&pid=95027\"][{POST_SNAPBACK}][/a]

Fair enough! But is it the second best lens you own when you are not shooting architecture? The use I am talking about is stitched panoramic shots where one is effectively using the whole image circle of the lens, from one side to the other.

I've seen reports on the net of some photographers using a lens like the TS-E 24mm as a standard lens (with FF 35mm) because there's no vignetting in the corners. That may be true. There's no vignetting in the corners and no substantial image degradation in the corners or at the edges of a 43mm diagonal. It's an advantage. However, the lens is not sharp, period. At least my copy is not sharp and Photodo's copy was not sharp. There have been many reports on the net that this lens is not sharp.

Maybe you are lucky and got an aberration.  
Logged
ericstaud
Sr. Member
****
Offline Offline

Posts: 384


WWW
« Reply #8 on: January 11, 2007, 12:18:25 AM »
ReplyReply

Quote
Can somebody tell me why fixing it in Photoshop doesn't work better than shift lenses? I know about using tilt for DOF, which Photoshop can't do, but the shift stuff seems to work fairly well...or not?

JC
[a href=\"index.php?act=findpost&pid=95035\"][{POST_SNAPBACK}][/a]

There has been a recent thread discussing this.  Photoshop works, but it has drawbacks...

-You are stretching the pixels appart and loosing resolution.

-Without a reference image it can be difficult to get the proportions correct. (It is easy to take a building with square windows and accidentally make them rectangles).

-You are adding a step of retouching which costs somebody time and money.

-The image you preview in the field (and maybe show to your client) does not represent the final image and the final crop.

I had thought I could shoot some "quick and dirty" jobs using a DSLR, but they become a pain to retouch later.  The Schneider and Rodenstock lenses used on the medium format shifts systems are also very sharp and have virually no barrel or pincushion distortion.
Logged
Kirk Gittings
Sr. Member
****
Offline Offline

Posts: 1550


WWW
« Reply #9 on: January 11, 2007, 12:46:00 AM »
ReplyReply

Ray,

Actually I do those stitches all the time all the time on interiors and sometimes on exteriors too, shifting left then right or sometimes 8:00 and 4:00 for a slight lens "fall" and left right shift or 10:00 and 2:00 for a slight "rise". So I do use the lens at extreme shifts routinely. There is also little fall off on this lens at the corners so there are no stiching problems in that regard either. I do this with a 45 TS too. They both make  stitching the images in PS very very easy and I don't have to fool with the nodal point. These are good lenses.

I am willing to accept the fact that maybe I got lucky with the lenses, but understand after shooting 4x5 for 28 years (and running) for a national clientèle, my standards are pretty high.

John,

It depends on what you are shooting, how much correction you need and what kind of quality you need. For instance I do small touch ups of PC all the time in PS, but the amount of correction that would be needed on a building with a wide fixed lens would be extreme and introduce interpolation artifacts in the part of the image that gets stretched. Now there may be just sky there and that would be ok, but there may be important detail there that can get weird. The rule of thumb I tell my students is: if you have to stretch the image more than a third of the width of the frame (this is really pushing it too) you are going to have problems.

PC lenses are still important to architectural photographers for that reason just as view cameras were (and still are).

I am new to this forum and would like to share some of my overview of architectural photography where shifts are most commonly used and needed. For many reasons, all of the architectural photographers that I personally know (and I know allot all over the country from teaching and lecturing) are using full view cameras (with film or Betterlight backs) or Canon FF. This includes two friends of mine that are regulars for Architectural Digest.
« Last Edit: January 11, 2007, 12:51:06 AM by Kirk Gittings » Logged

Thanks,
Kirk

Kirk Gittings
Architecture and Landscape Photography
WWW.GITTINGSPHOTO.COM

LIGHT+SPACE+STRUCTURE (blog)
BernardLanguillier
Sr. Member
****
Offline Offline

Posts: 8089



WWW
« Reply #10 on: January 11, 2007, 02:26:49 AM »
ReplyReply

Quote
Sorting out the SLR's view finder might might be clumsy but workable for shifting.  Tilt could be tricky.  This has to be easier than designing and selling a fleet of T/S lenses.

Peter
[a href=\"index.php?act=findpost&pid=95015\"][{POST_SNAPBACK}][/a]

Peter,

Interesting idea, but:

- sensor tilt isn't the same as lens tilt since it does affect objects projected shape,
- you'd have to have some sort of live preview to be able to vizualise the effect of the sensor tilt, since it wouldn't show in the viewfinder.

Regards,
Bernard
Logged

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

Posts: 8887


« Reply #11 on: January 11, 2007, 03:17:41 AM »
ReplyReply

Quote
It depends on what you are shooting, how much correction you need and what kind of quality you need. For instance I do small touch ups of PC all the time in PS, but the amount of correction that would be needed on a building with a wide fixed lens would be extreme and introduce interpolation artifacts in the part of the image that gets stretched. Now there may be just sky there and that would be ok, but there may be important detail there that can get weird. The rule of thumb I tell my students is: if you have to stretch the image more than a third of the width of the frame (this is really pushing it too) you are going to have problems.
[a href=\"index.php?act=findpost&pid=95044\"][{POST_SNAPBACK}][/a]

Absolutely! I just tried using the PS perspective controls on the following image (free transform/distort). It's impossible without losing most of the image as well as introducing massive interpolation effects.

Canon 5D, 15mm lens at f11.

[attachment=1508:attachment]
Logged
LMO
Newbie
*
Offline Offline

Posts: 37


WWW
« Reply #12 on: January 11, 2007, 03:58:54 AM »
ReplyReply

The biggest problem with doing this in photoshop I feel is poor lenses.. If you have just a little CA or other "defect" in the image it will be severely worse when you start dragging the pixels in photoshop. And if your lens has just a little distortion you might end up with something that looks more like a baloon..

If you keep in mind the cropping when you compose this seldom is a big problem (unless you really need the resolution..) . I have shot interiors this way for several years. But lens artifacts just kills the image..

But if you want to spend your evenings doing anything else than PS perspective control I think doing it in camara is much much better..

Lasse

Quote
Can somebody tell me why fixing it in Photoshop doesn't work better than shift lenses? I know about using tilt for DOF, which Photoshop can't do, but the shift stuff seems to work fairly well...or not?

JC
[a href=\"index.php?act=findpost&pid=95035\"][{POST_SNAPBACK}][/a]
« Last Edit: January 11, 2007, 04:06:49 AM by LMO » Logged

Lasse Mørkhagen
http://www.morkhagen.net

I may not have gone where I intended to go, but I think I have ended up where I needed to be. -Douglas Adams
Ray
Sr. Member
****
Offline Offline

Posts: 8887


« Reply #13 on: January 11, 2007, 06:41:10 AM »
ReplyReply

Quote
So I do use the lens at extreme shifts routinely. There is also little fall off on this lens at the corners so there are no stiching problems in that regard either. I do this with a 45 TS too. They both make  stitching the images in PS very very easy and I don't have to fool with the nodal point. These are good lenses.
[a href=\"index.php?act=findpost&pid=95044\"][{POST_SNAPBACK}][/a]

Kirk,
My TS-E 24 suffers from significant vignetting and poor resolution at the edges and corners. This is not a problem for stitching, however, because the overlap areas are always around the central area of the image circle.

Stitching the 2 or 3 images you get using shift on these lenses is a breeze. The following image was stitched automatically with Panavue's Image Assembler. Nothing could be easier and quicker. I show it here with no modification or crop. This is how it looks, straight from the stitching program.

[attachment=1512:attachment]

But look at the strong vignetting at the edges. However, vignetting can be fixed in PS (filters/distort/lens correction). It's the resolution fall-off I don't like. Let's look at the bottom left corner, lightened so we can see more clearly.

[attachment=1513:attachment]

This is a 100% crop with jpeg compression at maximum quality. It's just not sharp and the noise is so bad in the shadows one would think I'd used ISO 800 instead of ISO 100.

Let's look at the top right corner. The fact that this is also not sharp doesn't matter in this image because I'll probably lose that corner when I apply some perspective correction in PS. Nevertheless, that corner is very unsharp.

[attachment=1514:attachment]

Just for comparison, let's have a look at the centre of the image. This is reasonably sharp. At least as good as my Sigma 15-30 at 24mm, perhaps marginally sharper. I haven't applied any fancy sharpening routines to this image. Just the default sharpening in ACR. This is another 100% crop.

[attachment=1515:attachment]

For this image I used a 5D with attached L bracket which slides on a clamp fixed to the tripod. This set-up enables one to effectively shift the sensor around the image circle, but in a rather clumsy manner. Since Image Assembler stitches such images perfectly, whether the lens is shifted in relation to the sensor, or the sensor shifted in relation to the lens, I sometimes don't bother loosening the clamp to slide the camera body along. It takes more time and in a situation like the above, one doesn't want a tourist wandering into the scene   .
Logged
peterhorsley
Newbie
*
Offline Offline

Posts: 23


« Reply #14 on: January 11, 2007, 08:38:51 AM »
ReplyReply

Hi Bernard

Quote
- sensor tilt isn't the same as lens tilt since it does affect objects projected shape,

[a href=\"index.php?act=findpost&pid=95050\"][{POST_SNAPBACK}][/a]

Solution: point camera up or down until shape is correct. Shift sensor to re-compose subject.  (The limited rise and fall of old Linhof field cameras used to require this sort of shenanaghan)

Quote
- you'd have to have some sort of live preview to be able to vizualise the effect of the sensor tilt, since it wouldn't show in the viewfinder.

[a href=\"index.php?act=findpost&pid=95050\"][{POST_SNAPBACK}][/a]
Solution: see forthcoming P+ and Leaf live view.

But I understand your point; tilt would be tricky to use.


Peter
Logged
Kirk Gittings
Sr. Member
****
Offline Offline

Posts: 1550


WWW
« Reply #15 on: January 11, 2007, 11:09:13 AM »
ReplyReply

Ray,

I appreciate the issues you are having with that lens. Is it that soft at the edges with all apertures? That softness would not be acceptable to me. If it was mine I would replace it.  They are not all that bad.
Logged

Thanks,
Kirk

Kirk Gittings
Architecture and Landscape Photography
WWW.GITTINGSPHOTO.COM

LIGHT+SPACE+STRUCTURE (blog)
Ray
Sr. Member
****
Offline Offline

Posts: 8887


« Reply #16 on: January 11, 2007, 11:33:06 AM »
ReplyReply

Quote
Ray,

I appreciate the issues you are having with that lens. Is it that soft at the edges with all apertures? That softness would not be acceptable to me. If it was mine I would replace it.  They are not all that bad.
[a href=\"index.php?act=findpost&pid=95115\"][{POST_SNAPBACK}][/a]

Kirk,
Shots were taken at f8. Perhaps I should have used f11 or even f16. I'll check some other images. I'm sorting through about 15,000 that I took on my last trip.

The problem in buying a lens such as the TS-E 24 is that one can't expect one's local store to stock it (in Australia anyway). One has to order it in. For me to reject such a lens, I'd have to have a good reason. At the time I bought the lens, I did not have the full frame 5D. With the D60 (and later 20D) there was no such problem.

I have some stitched shots with this lens, used with the D60 & 20D, which are far sharper than any single shot could be (from edge to edge), so the lens does have a use.

I'm disappointed because, when upgrading to the 5D, my main concern was how well my Sigma 15-30 would perform at the edges. It seems to perform better than the TS-E 24.

Better at the edge of the image circle, that is. Of course, the TS-E 24 has a larger image circle, so the comparison is perhaps unfair.
« Last Edit: January 11, 2007, 11:37:12 AM by Ray » Logged
Kirk Gittings
Sr. Member
****
Offline Offline

Posts: 1550


WWW
« Reply #17 on: January 11, 2007, 11:45:40 AM »
ReplyReply

Ray,

I understand completely. Though I don't live in the outback, I may as well be. I live in New Mexico. Everything I buy almost is mail order. I have a routine with lenses. I only buy from big mail order houses with good return policies. The day I get them I test them against my sharpest lens and decide whether to keep them. I also first look to see if there is any evidence of the box being opened etc. I don't waste time testing someone else's reject. To date (cross my fingers), I have never had to return a new Canon L lens. I have, though, with used and after market lenses both for the Canon and View Camera lenses.

Also, I am less concerned with the wider apertures. The only serious photography that I do is always on a tripod of non-moving objects, so f/11 or f/16 is no problem for me.

I apologize all for hijacking this thread to talk about DSLR shift lenses. I felt the need to comment on the negative statements about the TS 24.
« Last Edit: January 11, 2007, 11:47:44 AM by Kirk Gittings » Logged

Thanks,
Kirk

Kirk Gittings
Architecture and Landscape Photography
WWW.GITTINGSPHOTO.COM

LIGHT+SPACE+STRUCTURE (blog)
BJL
Sr. Member
****
Offline Offline

Posts: 5146


« Reply #18 on: January 11, 2007, 12:36:43 PM »
ReplyReply

Thanks for all the comments. One that took my fancy was about using sensor shifting for stitched panoramics: for example up to 33x88mm with two 33x44 frames or 44x99mm with three, so on.

In fact this seems an intriguing approach for photographing stationary subjects in formats that would require prohibitively expensive sensors to do in a single shot, and so to revitalize some larger format lenses (645 and up, but particularly 6x6 and up) that have been "orphaned" by the digital transition. Four stitched frames could cover up to 6x9, and so on.

My new idea is a camera that takes lenses in variety of formats larger than its native format, but maybe flexibility would be best served by this being done with a "sensor shifting back" for a modular view camera system. And once you are doing multiple shifts, maybe it would be better to use an even smaller and far less expensive sensor, joining more frames. Or to do the ultimate in stitching: joining the thousands of long thin frames from a scanned linear sensor.
Logged
gkroeger
Jr. Member
**
Offline Offline

Posts: 51


« Reply #19 on: January 11, 2007, 02:36:53 PM »
ReplyReply

Quote
Sorting out the SLR's view finder might might be clumsy but workable for shifting.  Tilt could be tricky.  This has to be easier than designing and selling a fleet of T/S lenses.

Neat idea.

Peter
[a href=\"index.php?act=findpost&pid=95015\"][{POST_SNAPBACK}][/a]


Once we get high resolution digital view finder's, tilt and shift of the sensor will be the way to go... but don't hold your breath

Glenn
Logged
Pages: [1] 2 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad