Quick question for those who use RF lenses on Sony BSI sensors

Discussion in 'Mirrorless Digital Cameras' started by Karim Ghantous, May 10, 2021.

  1. If you have a Sony camera with a 24x36 BSI sensor, such as an A9 or A1, do you find that corner performance with rangefinder lenses improves compared to cameras with non-BSI sensors?
  2. I haven't noticed any difference using Leica and Zeiss rangefinder lenses (for Leica). I have owned both types, starting with an A7ii. My first BSI sensor was in an A9, and all subsequent models. A BSI sensor doesn't affect the focus. The purpose of BSI is to allow larger cells by reducing the amount of space on the front used by circuitry. Rangefinder lenses shorter than 50 mm suffer corner degradation due to the cover glass, which is the same for all Sony cameras, BSI or not. Lenses designed specifically for Sony are designed with that cover glass as an additional element. When measuring MTF for these lenses, a 2 mm thick glass must be inserted near the focal plane.

    My Summicron 50/2 is a so-so performer with a Sony. The Summicron 90/2 performs superbly, hampered only by its weight, the need to enter the focal length for IBIS, and manually engage focus magnification. The Summicron 90 (v2) has less contrast than a Sony or Zeiss Batis or Loxia 90'ish lens.
  3. The technology of BSI simply improves the photon-efficiency of a sensor and enables a greater pixel density. BSI sensors still employ micro-lenses. So why should they work any better with lenses not designed for them?

    One of the great - and unfulfilled - myths about MILCs was that they would enable the use and design of 'true' wideangle lenses. That myth has been well-and-truly busted, since the necessary micro-lens array still needs lenses to have a fair degree of telecentricity.

    Sure, you can get infinity focus with an old RF lens on a MILC, but the corners will still look fringey, heavily vignetted and crap!

    The only hope for RF wideangles is with Foveon sensors, but with Sigma holding the rights, I wouldn't hold my breath for huge progress in that direction.

    Does this crude diagram explain anything?

    / / / / / / - light rays
    | | | | - photosites
    Last edited: May 13, 2021
  4. / / / / / / - light rays
    ∩∩∩∩ - microlens array
    | | | | - photosites

    The microlens array helps reduce the oblique ray problem. Microlens array present = less need for telecentricity.
    It's, as Ed said, the cover glass that introduces spherical aberration with short flange to sensor distance lenses, i.e. all rangefinder camera lenses shorther than approx. 50 mm.
  5. Thanks, Ed.

    Because of what BSI actually means. I had a hypothesis that because there was less 'stuff' in the way of the photo-diodes, the corner performance would be dramatically increased. It seems I was wrong.


    Let the ASCII wars begin!
  6. Absolutely wrong!
    The microlens array is the very reason why telecentricity is needed.
    Any lens will throw an off-centre image to the side and reduce the on-axis light intensity.

    Viz: a simple magnifying glass and its image of the sun. If the sun is oblique to the lens, then its image will also be oblique, and the on-axis light intensity is reduced compared to if the magnifying glass was simply removed.

    And that's why Leitz had to change the geometry of the microlens array and offset it inward at the edge of the sensor in their digital Leicas.
  7. Ed made no mention of spherical abberation.

    Why do you make up stuff to support your flawed arguments?
  8. You're just full of it.

    Microlens arrays are put in place to cope with the wide spread in angles of incidence - i.e. the opposite of telecentricity - and sensor pit or well depth. Anything else - i.e. what you say - is pure B.S.

    That you do not know about or understand the problem Ed mentioned says a lot about you.

    Go find a rodeo, cowboy. Shoo!
  9. I suggest you read this.
    Then go and tell that conpany they've got it completely wrong.

    Check your facts before exploding into a blind argument just for the sake of contradiction. Because it's you that's full of it!

    Why don't you shoo, and stop spreading misinformation all over these fora like slug-slime?
  10. The effective thickness of the cover glass is proportional to the sine of the angle of incidence. This has an effect similar to spherical aberration, but in the reverse direction, causing unsharp corners and edges. The angle of incidence also causes parallax between the Bayer filter and the sensor, leading to color shifts.

    Micro lenses are "blazed" to improve light collection by the sensor cells. This reduces vignetting and parallax, but has no effect on distortion of the focal plane. All told, Leica 35 mm lenses perform poorly on an MILC due to their symmetrical design and short backfocus. Zeiss ZM lenses perform considerably better, but not nearly as well as MILC lenses.

    The closer the rear node of the lens to the sensor, the greater the angle of incidence away from the center. Rangefinder lenses were designed for film, which does not have thickness and parallax issues. Angle of incidence tends to cause vignetting, which is less obtrusive than fuzzy corners and odd colors. If the physical diaphragm is near the rear node, vignetting is minimal because the projection of the opening remains circular.

    MILC lenses shorter than 50 mm tend to have a reverse-telephoto design, which moves the rear node further from the image plane. My Zeiss Loxia 21 and 25 are nearly the same size as the Loxia 85 if the hoods are removed. Loxia lenses are designed to incorporate the sensor cover glass, which eliminates color and focus issues.
    Last edited: May 18, 2021
  11. Have i mentioned before that you are full of b.s.? I have, haven't i?
  12. Q.g. Just put away that dog-eared copy of 'How to win an Internet argument for Absolute Dummies'

    Refute, rebuke, insult, confuse. That's all the endless mantra we ever get from you.

    So you're saying that SVS-Vistek are full of BS too?

    Do you even own a camera? Never seen you post a picture in support of your loony theories. Nothing in your gallery either.
  13. Errm. That didn't seem to stop NASA and Hasselblad from sticking a glass reticle/pressure plate in front of the film taken on the moon. And that's not the only mission-critical camera that's used a glass pressure plate. Nor does using a glass carrier seem to affect the corner sharpness of an enlarger. Nor do the cover glass or glass platen of a flatbed scanner cause the edges of a scan to be less sharp than the centre.
    A perfect description of increased telecentricity.
  14. 1. I had to look up that Hasselblad Reseau plate. Very interesting. I assumed it had to exist but I never thought about it.
    2. I have completely forgotten how a negative is mounted. Is it between two pieces of glass? Does the distance of the glass from the film (essentially zero) make a difference though? Same question for the Reseau plate.
    3. Yes, but the light source moves in parallel to the platen, so you would not expect any corner smearing. The film carriers that I know of have no glass of their own, although that would not make any difference to corner sharpness AFAIK.
  15. It's the other way around. "Telecentricity" is or is not (binary). "Angle of Incidence" is a continuous value. The only "telecentric" device I know is an optical collimator. That said, there's no point in trying to call out popular misuse of terms. I'm fine with "almost telecentric", etc.

    Hasselblad used Reseau plates in special backs with the focal distance set to accommodate the effect of the glass plate. Like for any SLR, Hasselblad lenses have a long relief distance to accommodate the mirror, which reduces the angle of incidence effect to a point it can usually be ignored. Glass increases the effective focal distance by about 1/3rd its thickness.

    I've tested Hasselblad lenses on a Sony A7Riii, and find their image quality is close to that of native Sony lenses, with no corner effects, very little (if any) CA, and no noticeable vignetting. That holds true for Nikon SLR lenses too, but their performance is disappointing in other ways (corner softness is built in). A notable exception is the 55/2.8 MicroNikkor.
  16. What Ed said. You have to correct for the spherical aberration caused by inserting a glass plate, or make do with the lesser image quality caused by it.
    Doesn't matter that you do not think so.
    Not that will matter to you. But hey!

    O, and yes: your evidence for your misconception of the role of microlens arrays is not bs as such, rather a complaint about microlenses causing light loss. Something else. Boohoo!
    Not that it would be relevant, but can you find many sources that agree, or even share the same sentiment?
    It still makes your statement about microlens arrays any less incorrect. It was and is completely backwards.

    Do you own a camera... Migh, what a strong retort.
    Last edited: May 19, 2021
  17. Ed said: "All told, Leica 35 mm lenses perform poorly on an MILC due to their symmetrical design and short backfocus."

    I am not knowledgeable to make any comment on the technical aspects of this thread. But it seems the Leica 35mm work wonderfully on the MILC camera I use. Am I blind?
  18. It does depend on the size of the sensor. Leica M lenses work great on Fuji X series cameras, for example. Some even perform well on medium format cameras, no joke. But, I haven't seen a comprehensive comparison yet. It would be a large project, as useful as it would be.
  19. In another thread I shot a few M compatible lenses with ultra short backfocus on a Nikon Z6ii, which is BSI and (probably) a Sony sensor. You may think differently, but the corner smearing was much less than I expected. Quite a bit less than the same wide lenses would produce on a friends Sony a7iii although the experiments were not under the exact same conditions.
    I was under the impression that the glass/filters in front of the sensor bears the fault, but that's just what I heard.

    I wouldn't hesitate to mount a VC 28mm F2 Ultron or 35mm Summicron on the Z6ii if I needed to bring a smaller package than the kit zoom I own.
    Karim Ghantous likes this.
  20. That comparison is very useful, and I thank you for taking the time to do it. That kind of thing makes buying decisions much easier.

    There are some surprises there, too. Unsurprising: old wide-angle Nikkors don't perform well on digital cameras. Surprising: the 1970s Leica 50/2 M performed slightly better than the Nikkor. I didn't see that one coming.

Share This Page