Stitching with a tilt-shift lens

[pjtn]

I haven't seen a test comparing the D800 sensors tonal separation with MFD, it's generally always about resolution. I'd be interested to see how they compare.

[John Rodriguez]

Jim, I would like to keep the sensor 24x36mm for price reasons. So you're saying maybe 3 half stop increments in exposure, then stacked might create a smoother tonality? I have seen the same sort of thing done to reduce noise in an image.

Petrus, personally I find the DSLR images to be crunchy and lack the three dimensionality of film images. Have a look at some of the photos in this collection by Hin Chau http://www.hinius.net/atf.php. I believe he was using a Mamiya 7 II. Or am I seeing something else?

Here's a link to one in particular http://blinding.slideshowpro.com/albums/001/390/album-122821/cache/Shanghai-18-4445-03.sjpg_800_640_0_90_0_50_50.sjpg?1360205836

I think you're seeing something else, unless you're actually looking at the prints themselves. Fine tonal gradation isn't something that really shows up on a 600 pixel sRGB image.

[pjtn]

That is a good point. If the 'look' I'm seeing is not to do with the tonality, I'd be curious to find out what it is.

This series of photos also has the 'look' that I'm talking about:

http://www.koxvold.com/projects/los-angeles-river/

There is a smoothness and three dimensionally to the images.

[pjtn]

This series has the 'look' very strongly:

http://www.koxvold.com/projects/uae/

Or am I just looking at a certain type of film?

[Bart_van_der_Wolf]

Shifting a lens vertically, laterally or to some intermediate angle  changes the angle of view framed by the recording medium. Lenses with shift function project a relatively large diameter image circle - shifting the position of the lens or the recording medium changes the portion  of the image circle framed by the edges of the media.  Pointing the camera in a different direction has the same effect. This is simple geometry and has nothing all to do with stitching multiple frames together. Tilting the lens or the film/sensor plane can change the framing of the subject as well.

Hi Ellis,

That's correct, unless one offsets the displacement of the entrance pupil by a counter movement of the whole camera+lens. That's what the setup that I posted a picture of earlier in this thread achieves. It compensates for the horizontal lens shift, a common setup e.g. for landscape photography with some downward tilt included.

Indeed, tilting the lens will also introduce an offset, but it's only relevant if the amount of tilt is changed between shots, and the amount of off-axis displacement can be relatively limited, and can sometimes be hidden in the overlap blend. On my 1Ds3, the TS/E 45mm e.g. shifts the image center projection roughly 37.5 pixels (0.24 mm) per degree, the TS/E 90mm some 13.8 pixels (0.09 mm) per degree, and the TS/E 24mm II some 36.1 pixels (0.23 mm) per degree of tilt.

A good panorama stitcher can include these offsets (if not already compensated for when shooting) for an improved registration of the image tiles in a rotational stitch, but if only pre-compensated lens shifts are involved then stitching becomes basically simple overlaying of shifted images, even Photoshop can manage that.

Cheers,
Bart

[Bart_van_der_Wolf]

Jim, I would like to keep the sensor 24x36mm for price reasons. So you're saying maybe 3 half stop increments in exposure, then stacked might create a smoother tonality? I have seen the same sort of thing done to reduce noise in an image.

Hi,

All it takes is an average of multiple identical exposures (which obviously works best with stationary subjects). The photon shot noise will be reduced by a factor of 1/sqrt(N), where N is the number of exposures. So 4 exposures will already reduce the noise to half of that in a single exposure, and the signal to noise ratio (SNR) will double.

Of course the shadows could benefit from more photons, to improve tonality, and there exposure blending or HDR compositing may help as well. Using 3 bracketed exposures 0, +1, +2 EV will improve the shadows at least as much as 4 averaged exposures do, but the highlights will not improve if the +0 EV exposure was already 'exposed to the right'. Averaging is used a lot in astrophotography, for a reason ..., the SNR of all levels of brightness improves (with good registration).

Cheers,
Bart

[pjtn]

That's very interesting Bart. How would the 3 exposures be combined? Is it a simple matter of layering them in Photoshop and setting the opacity to 33%, 66% and 100% respectively?

[Ellis Vener]

This series has the 'look' very strongly:

http://www.koxvold.com/projects/uae/

Or am I just looking at a certain type of film?

It is scanned color negative film or possibly scanned prints made mostly from color negative film. The areas where I'd expect to see black or virtual black look more like they have higher values than that and the contrast throughout is fairly low, even on images that were clearly shot in midday light on sunny days.

I like his compositions and his ability to convey a certain emotional power (aloneness, without joy or irony) he has created in his landscapes and cityscapes. That is hard to sustain.   His portraits? While they are also well crafted, the emotional and aesthetic flatness is not my cup of tea or coffee.

[Ellis Vener]

That's very interesting Bart. How would the 3 exposures be combined? Is it a simple matter of layering them in Photoshop and setting the opacity to 33%, 66% and 100% respectively?

If you use Lightroom try the Enfuse plug in from Timothy Armes: http://www.photographers-toolbox.com/products/lrenfuse.php

[John Rodriguez]

As Ellis pointed out, he's shooting negative film.  You can try and replicate the response curves in processing. Try separate RGB curves in color mode to get the balance and one RGB combined curve in normal mode to emulate the luminosity/saturation.  There are also plugins you can try.  To replicate some of his images you might need to do some exposure blending where the usable dynamic range of C41 is larger then your sensor.  You likely wouldn't be able to get a color 100% match, as in some cases a film may separate colors that a sensor sees as the same, but you could definitely get it in the ballpark. Tim Parkin has some interesting articles about trying to replicate film that are worth a read.  He's mostly trying to replicate Velvia 50, but overall the problems are the same.  However, they're mostly color related, the contrast curves should be easier to emulate.

http://www.timparkin.co.uk/2012/08/why-you-cant-make-digital-look-like-velvia-50/

http://www.onlandscape.co.uk/2012/08/film-emulation-software/

[Ellis Vener]

Lee Varis has some interesting if very work intensive approaches to controlling color in his "10 channel color" (really it is  6 channels: RGB+ A & B from LAB + blacK from CMYK) workflow . Look up Lee Varis on Vimeo.com and then search inside them for "10 Channel Color Workflow." There are 4 videos.

Varis is using  color to create a heightened sense of depth  and then intensifying color  but the skeleton of his technique can be used to control color anyway you want to. Did I mention that it is work intensive?

[Bart_van_der_Wolf]

That's very interesting Bart. How would the 3 exposures be combined? Is it a simple matter of layering them in Photoshop and setting the opacity to 33%, 66% and 100% respectively?

Hi,

Either that, or you could try something like 40%, 60% and 100% because we're probably using gamma adjusted images at this stage. Alternatively you can use the Layer's Blend If functionality, targeting the shadows in the +2 EV layer, the Mid-tones in the +1 EV layer, but the individual layer exposures must also be adjusted if they are different.

But the correct way of combining multiple different exposure times would be to use a dedicated application that does exposure blending, or one that assembles an HDR image, e.g. Photoshop's  File|Automate|HDR merge Pro, for further tonemapping.

For the combination of multiple equal exposures one can use Photoshop Extended's Stack functionality: Layer|Smart Objects|Stack mode|Average.

Cheers,
Bart

[leuallen]

Ellis, I have used Lee 10 channel method and have one of his books. I found the technique very interesting. The work intensive part can be reduced by creating some PS actions to create the LAB and CYMK  files. I've found his technique of using the black CYMK channel to reduce the saturation of the shadows to be useful.

Larry

[John Rodriguez]

Dan Margulis is a good read as well.  He goes in depth into each color space.  Learning how to recognize each color channel just by looking at it was a real eye opener. 

[pjtn]

Wow, lots to take in here. I tried Googling to find answers but certainly found nothing like what I've been given here.