Pages: 1 2 [3]   Go Down

Author Topic: Interesting discussion with Hasselbald's product manager Ove Bengtsson  (Read 9473 times)

vjbelle

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 636

Hi Graham..... my issues were with my 150mm Digitar at shutter speeds between 1/15 to around 1/90 (don't hold me absolutely to this).  I made direct comparisons to my attached STC vs. the FPS and it was very noticeable.  I don't have all of the image stuff in front of me right now but it was so impacting that I immediately sent the camera back.  Alpa has put out a lot of propaganda since and I don't blame them since they want to mitigate an issue that, to me, can't be completely mitigated.  I have never looked back and, again, would never own that camera. 

Victor
« Last Edit: June 30, 2016, 01:06:34 pm by vjbelle »
Logged

Graham Welland

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 722

Btw, there's a credit hold on my camera gear piggy bank for an X1D, 30, 45, 90 combo  ;)
Logged
Graham

hjulenissen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2051

I believe that global electronic shutter is hard to do without some serious compromises. I believe that Eric Fossum have said something to that effect, and the mere lack of such products despite the obvious benefits seems to indicate that it is "hard".

If Sony (or someone else) were to make the break-through that allowed global shutter with minimal compromise, would they not introduce it in high-volume 1" (or cell-phone) sensors first? I am suggesting that we would see global shutters in the RX100Mx, then the A7-series, then (possibly) the MF sensors.
The electronic issue I see is reading the data off the sensor and then writing it into flash, preferably after applying fpn removal (for commercial reasons). One needs to manage the reading, buffering writing etc at the same time as the camera stays available for a new capture.

My feeling is thqt the electronic design to achieve high throughput needs to be carefully managed; I don't know whether  Sony license a downstream reference "camera in a box" complete with Jpeg generator for this sensor although I expect they do it for low-end devices.

Edmund
If this sensor is just a "small sensor made bigger", then perhaps the same supporting IP/electronics can be (re-)used (although at greater cost and/or lower speed)?

If Sony supplies everything from sensor to flash storage, does this not mean that manufacturing digital cameras become more like designing film cameras? A small team can design a camera that enables great images, they compete on ergonomy, size/weight, price, eco-system?

Intuitively, I'd assume that a CMOS sensor (complete with integrated A/D converters) could be connected to any standard System-On-a-Chip (ARM/MIPS/...) with sufficient i/o bandwidth connected to a decent memory buffer and flash storage running Android/[realtime embedded OS]. Sonys contribution might then consist of a purely software (platform-agnostic) generic library of image processing functions and sensor driver. If the EVF is a part of the package, it might make things slightly more complex, but not too much.

Turning a "working" development board (say, 80% solution) into a highly polished product (say, 99% solution) is a lot more than "19% work" :-). You want a user interface/experience that is intuitive, feels "snappy", does not look (too much) like a rip-off of your competitors. You want to utilize hardware efficiently so as to maximize still-image throughput, buffer utilization, EVF latency and image quality. Auto-focus. You want to maximize battery life. Any piece of external/3rd party stuff that you need to interface with (lenses, memory cards, flash, Wifi, Bluetooth, NFC) has the potential for infinite amounts of testing matrixes and improved functionality.

-h
« Last Edit: June 30, 2016, 02:33:29 am by hjulenissen »
Logged

eronald

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6642
    • My gallery on Instagram

If you're in Paris, I'm ready to invest $1K in doing a Sony sensor exercise - I agree that they probably have ranges, and assume that  within a range one uses the same interface, with a possibility of having various data path widths.

I believe that global electronic shutter is hard to do without some serious compromises. I believe that Eric Fossum have said something to that effect, and the mere lack of such products despite the obvious benefits seems to indicate that it is "hard".

If Sony (or someone else) were to make the break-through that allowed global shutter with minimal compromise, would they not introduce it in high-volume 1" (or cell-phone) sensors first? I am suggesting that we would see global shutters in the RX100Mx, then the A7-series, then (possibly) the MF sensors.If this sensor is just a "small sensor made bigger", then perhaps the same supporting IP/electronics can be (re-)used (although at greater cost and/or lower speed)?

If Sony supplies everything from sensor to flash storage, does this not mean that manufacturing digital cameras become more like designing film cameras? A small team can design a camera that enables great images, they compete on ergonomy, size/weight, price, eco-system?

Intuitively, I'd assume that a CMOS sensor (complete with integrated A/D converters) could be connected to any standard System-On-a-Chip (ARM/MIPS/...) with sufficient i/o bandwidth connected to a decent memory buffer and flash storage running Android/[realtime embedded OS]. Sonys contribution might then consist of a purely software (platform-agnostic) generic library of image processing functions and sensor driver. If the EVF is a part of the package, it might make things slightly more complex, but not too much.

Turning a "working" development board (say, 80% solution) into a highly polished product (say, 99% solution) is a lot more than "19% work" :-). You want a user interface/experience that is intuitive, feels "snappy", does not look (too much) like a rip-off of your competitors. You want to utilize hardware efficiently so as to maximize still-image throughput, buffer utilization, EVF latency and image quality. Auto-focus. You want to maximize battery life. Any piece of external/3rd party stuff that you need to interface with (lenses, memory cards, flash, Wifi, Bluetooth, NFC) has the potential for infinite amounts of testing matrixes and improved functionality.

-h
Logged
If you appreciate my blog posts help me by following on https://instagram.com/edmundronald

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600

If that is the case, the next generation of 135 style mirrorless cameras with a 50 MP Sony back side illuminated CMOS sensor could compete with crop frame medium format cameras. Lenses could be smaller and this is significant, since the lens often dwarfs the mirrorless camera.
On the question of how large a format is needed for "50MP worth" of resolution, meaning about 8000x6000, so 6000 l/ph [lines per picture height] in more traditional measures: I expect lens resolution and diffraction limits on usable f-stops to be the limit.  After all, current 1 micron phone-sized pixels would fit about 50MP into a 2/3" format sensor, but the promised resolution would only be achieved at f-stops under about f/2, or f/1 by the most pessimistic analysis, and it might be very hard to control aberrations at those low aperture ratios. (The dynamic range of such small photo-sites is another issue of course, but at least I see hope for techniques like incremental read-out of photo-sites overcoming that.)

Some numbers and speculation:
In 24x16 format ("APS-C", Nikon "DX", Sony "E"), the diffraction limit on resolving that 6000 l/ph is about f/2.8-f/5.6, where maybe aberration control in good modern less designs can handle it.
In 36x24 format, the diffraction limit on resolving that 6000 l/ph is about f/4-f/8, probably within the comfort zone of aberration control in good modern lens designs, including better zoom lenses.
In 44x33 format, it is about f/5.6 - f/11, where it is relatively easy to control aberrations. In fact even 100MP only needs about f/4-f/8, where adequate aberration control is also probably quite attainable.

In the long run, I can see the case for formats bigger than about 44x33 diminishing greatly, if DR at a given pixel size improves as much as I think it can. If so, then Hasselblad is positioning itself well with this new EVF camera system.


P. S. About format names for the larger, more "serious" sensor formats: why not just use dimensions in mm, akin to the use of cm with MF film formats and inches with LF film?  It seems that there will be just a few "larger than mainstream" digital ILC formats:
36x24, 44x33 and 54x40 (and at Leica, 45x30).


P.P.S.  Do we discuss the X1D in the "Compact System Cameras" forum?!
« Last Edit: July 05, 2016, 12:02:14 pm by BJL »
Logged
Pages: 1 2 [3]   Go Up