Hi David,
Sorry for using incorrect nomenclature. Calling the lenses telecentric is probably a mixup. But fact is that the issues you discuss are related to the M-lenses and not to the S-lenses. Many lenses are symmetric designs, they are popular as they offer very good image quality with a relative simple construction. Symmetric wide angle lenses have a large disadventage, however, they cannot be used SLR cameras as they don't allow for the mirror box.
Normal lenses for SLR-s often symmetric, a construction called "double gauss". This doesn't work with SLRs however, so a new solution to wide angles has been invented, probably by Angénieux, who called this design retrofocus. Another name for the same construction is inverted telephoto. The design increases the distance between the exit pupil and the focal plane. The retrofocus/inverted telephoto design is often referred to as telecentric, but I feel it is a missnormer.
All DSLR wide angles are essentially retrofocus type lenses, and this design is necessitated by the mirror box. Zeiss calls this design Distagon.
Now, retrofocus, inverted telephoto or Distagon type lenses have some advantages
- Arbitrary long distance between outlet pupil and focal plane
- Less vignetting than symmetric designs
- Low beam angles, so they work well digital sensors
But, they also have disadvantages
- Larger size
- Longer as the main part of a Distagon type lens is outside the lens mount
- Difficult to make
- Distortion is much more difficult to control
Symmetric lenses are more compact but they have some issues, especially with digital sensors
- Symmetric lenss have short distance between exit pupil and focal plane
- Thus they have large beam angles
- The large beam angle in combination with the cover glass is a cause of astigmatism, the reason that Leica M cameras have thin cover glass. The M8 had extremely thin cover glass leading to inferior IR-supression.
- Natural vignetting on symmetric designs follows cosine four law, while Distagon types can reduce vignetting significantly.
- As stated in your article the large beam angles of short symmetrical designs don't work well with deep sensor wells.
Later generation of wide angles for the M are slightly retrofocus designs, as there was a need to make space for exposure meter cell of the Leica M5.
I would suggest that you check the information above with any knowledgable optical engineer.
Now, Leica has a problem with lenses that are not designed for electronic sensor. The made several design choices:
- Very thin cover glass
- Coding on lenses enabling automatic correction of vignetting and lens cast
- A sensor optimised for lenses with large beam angles
But, nothing of the above applies to the Leica-S series as it is a DSLR, and SLR cameras cannot use lenses with large beam angles because they need room for the mirror.
So, for the Leica-S all the discussion of beam angles is totally irrelevant. Talking about the Sony A7r is also totally irrelevant as it is clearly intended to be used with lenses designed for it, even if it works perfectly well lenses designed for SLRs.
Getting back to the DR question. 79 dB is stated. You can convert dB to EV by dividing by 6.02. Now 79 dB is 13.12 EV, I would'n call that 13 to 14 EV, about 13 EV would be a truthful statement. Nikon D750 has a DR of 13.73EV while the Leica M (240) has 12.34 EV, which indicative of the CMOS lagging modern Sony developed sensors a bit. It is possible that the design have been improved, but the stated 79 db is 13.12 EV. The DR is coming from the sensor and it is not related to the processing pipeline, although it seems that some manipulation may be quite frequent.
As you see, claiming advantages of the CMOSIS sensors on the Leica-S is clearly irrelevant. On the other hand the CMOSIS sensor is probably very benefitial for the classic M lenses. But classic M-lenses cannot be used on the Leica-S.
These are the reasons I strongly suggest that the article has a lot of bad information. Clearly, it is not your fault, it seems quite clear that you got that information from Leica marketing.
Some good reading here:
http://toothwalker.org/optics/vignetting.htmlhttp://en.wikipedia.org/wiki/Angénieux_retrofocus
http://www.lensrentals.com/blog/2011/03/the-development-of-wide-angle-lenseshttp://www.lensrentals.com/blog/2014/06/sensor-stack-thickness-when-does-it-matterBest regards
Erik
Erik,
I'm sorry you feel my information isn't accurate. All the facts, figures and reasoning came directly from my conversations with engineers at Leica who are directly responsible for the work with CMOSIS designing the chip. Interestingly, you are the only person to take issue with my write-up since it has been posted a few weeks ago, but I am happy to address your specific issues.
You are absolutely correct. The S lenses are telecentric designs, but the snippet of text you quoted from my article only talks about M lenses on an M240 and using those same lenses on a Sony A7r. I didn't make mention of the S lenses.
I think it would be interesting to explore in further detail just how telecentric the S lenses (or other MFDSLR lenses) actually are. Sure, I'd imagine a 180mm to be close to 100% telecentric, but perhaps an ultra-wide like the 24mm might not be totally telecentric.
In my article I said "around 79db" as the early testing samples Leica received from CMOSIS were from the first wafer. Dr. Zimmer seemed to be hinting at slightly better performance when the sensor goes into full production. Likewise, there are other factors in the imaging pipeline that contribute to overall system DR. The range of 13 to 14 stops is a reasonable estimate based on current testing and knowledge of future optimization.
We will have to see when real-life comparisons can be made. I'm as anxious as you to see the results.
Agreed. I don't like disinformation either. That's why I fact check and speak directly to the source.
David
