Henning Offline Upload & Sell: Off
|
rico wrote:
@Henning@, "optical shift" is already a reality with the Scheimpflug Principle. After tilting, the optical axis can be restored to the sensor center with a front- or rear-standard shift. It's easier to imagine if you have a LF camera handy, and see the image on the ground glass. The 35mm bellows with T/S (Contax RTS, Nikon, Novoflex) have full Scheimpflug capabilities, but the limited movements make iterative adjustments (like focus) a necessity. I'm jealous of our FM Alties who have gone whole hog with technical camera setups: expensive and bulky, but much easier to use.
I think you misunderstand me. I fully understand all the Scheimpflug concepts; I've been an architectural photographer for a very long time and still have enough Sinar components to build about 3 Sinars, some other 4x5's and lenses from 35 to 480mm. This is not about restoring an optical axis through tilting. I was talking about shifting; i.e., the subject focal plane stays parallel to the sensor. That's really the basic requirement of a shifted lens, and when using a TS-E lens for architectural work on 24x36 tilting is really only used when shooting architectural models. When you use lens shift, almost all optical systems that we use for photography shifts the optical axis relative to the sensors centre, thus making use of the steeper incident rays that the lens is capable of to illuminate the corners opposite the direction of shift. This causes the incidence angle to be steeper than the sensor is designed for, and causes mushy corners and asymmetric colour shifts, as when you use most Leica wide angles on the A7 series or other bodies not designed for them, and particularly the A7RII. Higher pixel densities bring out these faults to a greater degree. On the Canon 5DmkII and mkIII, which I used for the shift lenses before, I was never bothered by the mushiness in the corners very much, but it was there. Things are magnified with the A7RII and its 42mp sensor, and the thicker Sony coverglass. With the Sony 42mp sensor I have higher standards, and therefore it bothers me more.
By 'optical shift' I mean using the rear group as a transfer group to keep the angles of incidence to the far corners moderate, while allowing the front groups to provide the majority of the shift. Probably the best solution is to have two differential shift motions; a smaller one for the rear transfer group and a greater one for the front groups. It would make the lens very large, complex and expensive.
It might be easier (but it would result in a larger lens again) to make the total design more retro focus, so that the rear exit pupil is further from the sensor.
As I mentioned before, when I use the 90 TS-E, I really don't run into problems, because I use that lens a lot with tilt, and also, it's exit pupil is much further away from the sensor.
Btw, on the technical cameras you can't use the older, more nearly symmetrical wide-angle lenses anymore either when large shift movements are contemplated; newer lenses that are more retro focus are required and also, the pixel density is much lens on the MF backs, so mushiness is less noticeable. Overall, the same constraints apply just that on the Sony, with high pixel density and thick cover glass we run into noticeable problems sooner. The system is just not designed with shift lenses in mind.
|