CL cyan drift with wide-angle lenses?

[lct]

Advertisement (gone after registration)

A fair point - but what "other Bayer-filtered cameras" are you comparing to, that also replicate all the other unique characteristics of the Leica M digitals, particularly the M9?

- short lens register/back-focus (< 40mm) - eliminates all mirrored SLRs

- no AA filter

- thin (0.8mm) cover-glass/IR filter - eliminates Sony Alphas (1.5mm)

- no "designed for digital" telecentric wide-angles

- full-frame (since IFS did not regularly occur with the cropped M8) - eliminates Fuji and Olympus bodies [...]

 

My Sony A7s modded by Kolari Vision. No Italian flag, at least none visible, even with difficult lenses like SA 21/3.4 or CV 21/4. But cyan shifts it has more or less in the corners depending on lenses used. Never felt the need to use Cornerfix though but i'm no pixel peeper.

Edited December 12, 2017 by lct

[adan]

V-e-r-r-y interesting!

 

It is of note that the A7 series (and some Fujis) use "back-illuminated" sensors, which also reduces the depth (the "wells") that light has to penetrate to get to the light-sensitive part of the silicon (as thighslapper alluded to). Something of a misnomer, because the sensor is still obviously illuminated from the front (towards the lens), but the sensor architecture is "flipped" to put the circuitry on the back, thus backwards from traditional CMOS/CCDs.

 

And this Sony diagram also suggests an additional effect - reflections off the circuitry on the front of traditional sensor structure kicking, for example, green-filtered light into a blue-tagged pixel, which would also foul up the reading of colors at low angles of incidence ("Italian Flag"):

 

https://www.sony.net/SonyInfo/News/Press/200806/08-069E/images/no_flash.gif

 

(That is of course a schematic, so one does need to be careful in interpreting it for real pixels.)

 

Of course the A7s, with only 12 Mpixels, also has larger, more widely spaced pixels - that may make some difference as well.

 

On a slightly different issue, it has also been suggested that front-lluminated sensors tend to produce warm/red casts (my bête noire with the M240 color, and my Canons), perhaps from the same phenomenon that allows the Foveon sensor to work - red/IR light can penetrate deeper into silicon - so if the light-sensitive areas are deeper in the structure....

 

http://keyassets.timeincuk.net/inspirewp/live/wp-content/uploads/sites/13/2010/08/FOVEON_X3_sensor1.jpg

 

Seems like back-illuminated CMOS sensors solve a number of problems - another one being "fill-factor," or the percentage of the sensor surface area where light is not blocked by the circuitry, which increases ISO sensitivity. I expect they will eventually become the norm, making life much easier for Leica's problem wide-angles (and everyone else, with any lens).

 

Maybe in the next generation. But possibly less relevant to the cropped CL (back on topic).

Edited December 12, 2017 by adan

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[jonoslack]

Seeing as there seems to be no consensus as to any issue here, I'm going to continue happy in the knowledge that there's no problem worth worrying about.

Exactly! That’s what I’m doing too!

[Jared]

V-e-r-r-y interesting!

 

It is of note that the A7 series (and some Fujis) use "back-illuminated" sensors, which also reduces the depth (the "wells") that light has to penetrate to get to the light-sensitive part of the silicon (as thighslapper alluded to). Something of a misnomer, because the sensor is still obviously illuminated from the front (towards the lens), but the sensor architecture is "flipped" to put the circuitry on the back, thus backwards from traditional CMOS/CCDs.

 

And this Sony diagram also suggests an additional effect - reflections off the circuitry on the front of traditional sensor structure kicking, for example, green-filtered light into a blue-tagged pixel, which would also foul up the reading of colors at low angles of incidence ("Italian Flag"):

 

https://www.sony.net/SonyInfo/News/Press/200806/08-069E/images/no_flash.gif

 

(That is of course a schematic, so one does need to be careful in interpreting it for real pixels.)

 

Of course the A7s, with only 12 Mpixels, also has larger, more widely spaced pixels - that may make some difference as well.

 

On a slightly different issue, it has also been suggested that front-lluminated sensors tend to produce warm/red casts (my bête noire with the M240 color, and my Canons), perhaps from the same phenomenon that allows the Foveon sensor to work - red/IR light can penetrate deeper into silicon - so if the light-sensitive areas are deeper in the structure....

 

http://keyassets.timeincuk.net/inspirewp/live/wp-content/uploads/sites/13/2010/08/FOVEON_X3_sensor1.jpg

 

Seems like back-illuminated CMOS sensors solve a number of problems - another one being "fill-factor," or the percentage of the sensor surface area where light is not blocked by the circuitry, which increases ISO sensitivity. I expect they will eventually become the norm, making life much easier for Leica's problem wide-angles (and everyone else, with any lens).

 

Maybe in the next generation. But possibly less relevant to the cropped CL (back on topic).

I don’t see any way that back illuminated vs front illuminat d could have any impact on colors, certainly not as long as the manufacturer is doing a decent job in determining what to write out to the DNG or raw file. They can pick whatever balance they want in the design. Obviously, there will be some variations in color filter arrays both in terms of sensitivity and band pass, so either way the manufacturer must determine how to weight the individual R, G, and B values.

 

The fill factor, though, is the big reason why BSI chips are an improvement on front-side illuminated chips. That gives you a higher quantum efficiency which is particularly critical for chips with small pixels. Keeps the noise down.

 

One interesting drawback to BSI, or at least a complication, is that most BSI chips are thinned—polished down to remove light absorbing substrate. That thinning process is often not perfectly even. It leaves polishing marks on the chip. As a result, a flat field map needs to be applied that is individual to each chip. This is much like the bad pixel map stored in most cameras to address hot and cold pixels. One more way raw isn’t as raw as we think.

 

Of course, this is all purely academic. The pictures still work fine.

[sandymc]

In that case, why don't we see this on other Bayer-filtered cameras? The Italian Flag aberration can be corrected by slightly off-centering the lens (for instance by sticking a bit of Sellotape to the flange.

 

Interesting enough, Italian flag has been seen on other cameras - in fact, the phrase was coined for the old Kodak 14n

 

Sandy

[jaapv]

That is not surprise; the Kodak sensors of M8 and M9 are closely related to the Kodak 14 sensors.