Criteria for testing 31 meg back with T/S?

Good question - much more likely to be an issue with wide lenses.

fotografz said:

Okay, despite the chance of making a Jackass of myself ... I have a question about use of the H3D-II/31 on a technical camera using T/S.

This came about because Mark K asked me to try my new 31 back on the Rollei Xact2. Normally I don't use the 31DB on the Rollei because I also have a H3D-II/39, and that's the back I always use.

However, neither Mark or I can recall actually seeing examples of the reported micro-lens color cast issue from a H3D/31 posted anywhere.

Is it an issue with all lenses? Or is it a W/A issue?

I tried it with Rodenstock 90mm APO digital lens @ f/32 and 1/125th using consistent strobes to light a color chart ... and induced severe Tilts and Shifts in a couple of different torture configurations ... and neither Mark or I could detect the color cast issue.

So, the question is ... how should the test be done to check for this? What needs to happen to see the color cast?

Any help here would be appreciated.

Click to expand...

i've only seen it happen with wide angle lenses. the 50 shift i have for the mamiya 645 doesn't produce the color cast on the Aptus 75, but the 35mm will produce some when shifted using the Horseman. An evenly lit white wall shows it on my 35 the best

JimCollum said:

i've only seen it happen with wide angle lenses. the 50 shift i have for the mamiya 645 doesn't produce the color cast on the Aptus 75, but the 35mm will produce some when shifted using the Horseman. An evenly lit white wall shows it on my 35 the best

Click to expand...

Thanks Jim. I have a Schneider 28/2.8 APO Digitar that I'll try with a flat white subject.

Marc, it has to do with the telecentricity of the lens and its interaction with the microlenses. Basically, sensors with microlenses can tolerate a small critical angle of incidence before a false color will register at that pixel location. Standard pixel sites capture photons that are relatively perpendicular to it. Microlenses are used to boost the signal sensitivity at the pixel sites, and behave like extra-large "cather's mits" that scoop up nearby photons at slightly more oblique incidence. Have you ever seen a baseball catcher try to fool the umpire that a ball just outside the strike zone was over the plate? That is similar to how microlenses pick up photons. However, if the angle of incidence is beyond a critical angle (small, typically 10 degrees or so), the photon gets called a "strike" when it is not, and a false color registers at the pixel site, creating the color cast.

The reason why some lenses yield color casts more easily than others is due to their telecentric design. Basically, the farther away the exit pupil is from the sensor, the more perpendicular the impact of the photon on the sensor site, and the less problem you will have with color casts. Typically, the non retro-focus wide angle lenses have exit pupils close to the sensor, and thereby cause the color cast issues that people see. Longer focal length lenses naturally push this exit pupil further away from the sensor. However, the newer digital lens designs, such as the Rodenstock APO 28mm HR lens, are designed with this image-telecentricity in mind (and is primarily what characterizes a "digital" lens, although there are other optimizations).

As for tilting and shifting, it is the same principle. Lenses with longer exit pupil distances from the sensor will be able to tolerate more shifting/tilting. Again, this will be more likely for the longer focal length lenses, and for the newer digital lens designs.

David

Jim... The aptus 75 does not use microlenses. I am curious which 35mm you were using on the Horseman.

David thanks for this explanation. The one question I have for you is why would the 28mm HR lens be a better choice since the HR lenses have a smaller image circle. This should be more problematic when trying to shift? Mark

mark1958 said:

Jim... The aptus 75 does not use microlenses. I am curious which 35mm you were using on the Horseman.

David thanks for this explanation. The one question I have for you is why would the 28mm HR lens be a better choice since the HR lenses have a smaller image circle. This should be more problematic when trying to shift? Mark

Click to expand...

Mark, the reference to the 28HR was a non-sequitur with regard to shifting...sorry about that. It was just the first lens that came to mind as an example of how the wide digital lenses are now being designed with longer exit pupils.

You are absolutely right with regard to tilt/shift using a MFDB. It does not have a large enough image circle in that case. Some folks use these lenses with Cambo and other setups that allow a Canon 1DsIII to capture the image, since there is much more room for movements.

So, to get my comments back on track....Marc is looking for a lens that WILL demonstrate severe color cast using T/S. To do this, he should use a lens with the shortest possible exit pupil distance from the sensor (NOT the digital/HR lenses), which would typically be a non-retrofocus ultra-wide non-digital lens.

David no problem. THe HR is one of the newest designs that I am aware of in terms of when these lenses were released, and just wanted to make sure I was following your reasoning.
Mark

David Klepacki said:

Mark, the reference to the 28HR was a non-sequitur with regard to shifting...sorry about that. It was just the first lens that came to mind as an example of how the wide digital lenses are now being designed with longer exit pupils.

You are absolutely right with regard to tilt/shift using a MFDB. It does not have a large enough image circle in that case. Some folks use these lenses with Cambo and other setups that allow a Canon 1DsIII to capture the image, since there is much more room for movements.

Click to expand...

fotografz said:

Any help here would be appreciated.

Click to expand...

Try processing in Flexcolor instead of Phocus. I noticed casts in DMR files when using the 28mm PC shifted. If I opened these files in Phocus, they would be corrected automatically.

BTW, I think the problem with casts would be worse with large shifts, rather than tilts.

Robert

Rob.. thanks ..THe samples that Marc showed me looked as if shifts did not cause as much of a problem in terms of casts.