lenses for architecture in 35mm

[Rainer Viertlböck]

i just posted this little article in another forum,- might be from interest here for someone, who knows.

about lenses and architecture i tried a lot of stuff in 35mm. and ended up with some lenses which do a fairly good job, although except the sigma 12-24 not any wa lense is really good corrected,- and the bad thing here is that all this wides show more or less significant moustache distortions, which is very difficult to correct in a way, that distortion will not show up if you put a straight line in parallel to the image border. and this often is a must in architecture motifs.
with shift lenses you furthermore have the problem that the distortion becomes unsymmetric.

for 35mm i use:
sigma 12-24 / this lense is by far the best corrected wide angle lenses which ever was madefor 35mm systems. its really funny.... problem with it is that the samples are very different, most of them suffer from bad centration, which will result in unsharp edges, often just one or two edges for some unsymmetric mounted lenses inside the optic. try 5 and you will get a real good one out of them. its worth the effort.

shift lenses are good from olympus in terms of sharpness and distortion ,- but this is relative. i wouldnt say they are really good- compared with 4x5" lenses or mf digital lenses ( schneider + rodenstock ), but better than all other existing 35mm shift lenses they are.
all other wide shift lenses ,equal if nikon, canon or schneider are not sharp if you shift more then 4-5 mm and all show significant distortion. here the nikon is best,- but the lense really isnt that sharp. same about 35mm focal length. olympus is best, together wit hthe contax,, but this lense shows - as alll zeiss wides - significant distortion again. the olympus not.
worthest thing is that the oly lenses show very much CA. really very much.

i use a lot the zoerk adapter together with pentax 645 af lenses. here the 35 mm AF lense is very good. sharp and not much distortion. very good lense to stitch two images to get impressive resolutions, which equal my 33mp sensor or even can exceed it for panoramas.
also the longer lenses from pentax 645 are very good. i have the 45/55/75 and all are great. the longer ones are also corrected nearly perfect.

about the sigma 12-24 again one word:
it shows at 12mm around 2% distortion, which is NOT moustache and therefor perfectly correctable.
this is the same distortion % than shows my 5000 $ rodenstock 28HR,- which equals 21mm in 35system.
its really impressive.

[Bart_van_der_Wolf]

Originally Posted by Rainer Viertlböck

About lenses and architecture i tried a lot of stuff in 35mm. and ended up with some lenses which do a fairly good job, although except the sigma 12-24 not any wa lense is really good corrected,- and the bad thing here is that all this wides show more or less significant moustache distortions, which is very difficult to correct in a way, that distortion will not show up if you put a straight line in parallel to the image border. and this often is a must in architecture motifs.
with shift lenses you furthermore have the problem that the distortion becomes unsymmetric.

Hallo Rainer,

Althought I'd agree that it is best to start with as little distortion as possible, I'm quite happy with my stitched images. Stitching with panorama software allows to correct for complex (barrel+pincushion in the same image) distortions at the same time as the stitching of multiple images. It also allows to enhance resolution, correct for keystoning, and adjust apparent/projection perspective.

shift lenses are good from olympus in terms of sharpness and distortion ,- but this is relative. i wouldnt say they are really good- compared with 4x5" lenses or mf digital lenses ( schneider + rodenstock ), but better than all other existing 35mm shift lenses they are.
all other wide shift lenses ,equal if nikon, canon or schneider are not sharp if you shift more then 4-5 mm and all show significant distortion. here the nikon is best,- but the lense really isnt that sharp. same about 35mm focal length. olympus is best, together wit hthe contax,, but this lense shows - as alll zeiss wides - significant distortion again. the olympus not.
worthest thing is that the oly lenses show very much CA. really very much.

I haven't compared the Canon and others side-by-side, but IMO the Canon T/S-E 45mm f/2.8 has moderate CA even when fully shifted, especially when a capable Raw converter can adjust for most of it.

Bart

[Asher Kelman]

Bart,

The point you have made about stitching is so valuable for people to follow. Most of the aberrations can be corrected. "Unless one has a money-spending itch, stich!"

The question I have is should corrections be made before stitching, as in DXO, or just rely on the power of the stitching software?

BTW, you might reassure most earthlings about lesser lenses, but Rainer is IMHO, one of the most fastidious guys I have ever met as fat as choice of lenses. He tests everything in the conditions to be used and has his gear corrected to his specs.

I'm really impressed, therefore, that he totally verifies the opinion of Nicolas Claris' on the Sigma 12-24, albeit one has to be selective. I have seen Nicolas' work from the 1DsII enlarged to wall size! It was amazing. This lens from sigma, is what Nicolas, one of the most sort after Marine Photographer's in Europe, uses.

These lessons are so important for the rest of us in selecting gear.

I can attest the value of the Zoerk. I use it with my 1DII and a 45mm Pentax 6x7 lens. The series II 45mm lens is said to be sharper. However Sean Reid likes the Zoerk for the smaler MF Pentax series.


Asher

[Dave McCaughan]

At PhotoExpo yesterday Calumet was showing the new Cambo X2-Pro.

http://www.cambo.com/Html/products_p...t/Item752.html

This seems to be a very elegant solution, and it is very well made. I'm sure that if it does well more adapters will be made available. We are switching from RZ to digital and I was at the show trying to decide if we really need a MFDB or if a 1DSII will be enough. We try not to use wide lenses but every once in a while a 28mm is a necessary evil. Otherwise our 37, 50, 65, 75, 90, and 110 RZ lenses should work well. I'd be interested to hear if anyone has experience with a 1DSII for magazine work especially interior editorial.

Dave

[Craig_Lamson]

Dave, I've been using the 1DsII since its introduction for marine and rv interiors along with the Sigma 12-24. I'm very happy with the results as are my clients. I recently did an entire series of brochures that had interiors as 22"x8.5" spreads and the files, uprezzed with GF were wonderful!

Examples are here:

www.infocusinc.net

Craig

[Bart_van_der_Wolf]

Originally Posted by Asher Kelman

"Unless one has a money-spending itch, stich!"

That pretty much captures my feelings on that subject.

The question I have is should corrections be made before stitching, as in DXO, or just rely on the power of the stitching software?

In the case of DxO, I'd say let DxO do most of the dirty work, and let the stitching software handle the projection distortions which allow a seamless transition (assisted by some blending application if you don't want to spend too much time on postprocessing). Same goes for CA correction without DxO, because most Pano software doesn't specifically separate the color channels, and the CA (and several other) correction(s) can only be done on the individual shots.

CA correction on shifted images, if not already adjusted by the Raw processor*, requires the introduction of an image centre offset, or the temporary addition of some canvas space to center the symmetrical lateral CA before correction.

*) Here is an example of the apparent CA difference from Adobe Raw Converter versus RawShooter Premium on a fully (11mm) shifted TS/E 45mm f/2.8 in Portrait orientation for a 3 image stitch. It demonstrates that the Raw converter alone will already make quite a difference, even before dedicated postprocessing.

BTW, you might reassure most earthlings about lesser lenses, but Rainer is IMHO, one of the most fastidious guys I have ever met as far as choice of lenses. He tests everything in the conditions to be used and has his gear corrected to his specs.

That's why I started my reaction with agreeing that good image quality (as little distortion as possible) from the start is preferable. Having said that though, there is still hope for lesser mortals (lesser in the sense of spending power, or fewer paying customers ;-) ).

Bart



spelling "fat" to "far"

[leonardobarreto.com]

Bart_van_der_Wolf

Now I know why you are "one of the most fastidious guys I have ever met", since I don't understand a couple of things, can I aks?


1) What is projection distortion?

I think that there are two different stitch situations, one is a panorama of two or more images and the other is one made from two captures of the same image circle and then put together in to one.

I don't see why there should be any thing to adjust in the second scenario. Of course there are differences in the center and periphery of the image circle, but this is a different topic all together, no?

Any way, I will try to find this 50mm shift Mamiya lens and do some testings of my own.

[Asher Kelman]

Originally Posted by Bart_van_der_Wolf

That pretty much captures my feelings on that subject.

*) Here is an example of the apparent CA difference from Adobe Raw Converter versus RawShooter Premium on a fully (11mm) shifted TS/E 45mm f/2.8 in Portrait orientation for a 3 image stitch. It demonstrates that the Raw converter alone will already make quite a difference, even before dedicated postprocessing.

There's some residual green on the top of the ball in the RawShooter processed image.

Can you get rid of that too?

Asher

[Bart_van_der_Wolf]

Originally Posted by Asher Kelman

There's some residual green on the top of the ball in the RawShooter processed image.

Yes, I just wanted to show the difference a Raw converter can make, even before postprocessing. I forgot to mention that I magnified that part to 200% zoom, so it looks worse that it actually is at the pixel level.

Can you get rid of that too?

Yes. Depending on how obvious it is in the image, one can either desaturate that color locally (or globally if the scene allows), or postprocess the individual images with a radial size reduction of the green layer, after adding empty centering space. The lighting and background/sky in the example I showed was so problematic that I prefer to re-shoot the images anyway, so I never got around to correcting it.

Bart