[Emacs]

Why you people think software method must be CDAF only? There are other approaches. In order to achieve PDAF speed these should implement two functions: defocus sign and defocus value estimation on the image.
Now take a look at the these two shots:

First is taken with Canon 135/2, the second with CZ APO Sonnar 135/2.
You can easily see different color fringes around contrast edges: ones that are behind DOF zone have green tint, these that are in front of it has purple or at least not green in case of APO. This is the direct consequence of different angle of refraction for different wavelengths. In corresponding mathematical model this is described by different PSFs (PSF=point spread function) for different wavelengths. This feature can be used (and it has been already in several papers) to estimate defocus sign.
The defocus value can be evaluated via blur estimation (not a trivial task though, but surely can be tuned good enough to be useful in practice).

[falconeye]

Therefore even if CDAF sensor is exposed to higher light intensity, the exposure time is shorter, so the light flux seen by CDAF might not be greater.
...
However in reality the resultant S/N ratio for CDAF is not proportionally 4 times larger than for PDAF as the PDAF sensor has the advantages of 1) being monochromatic 2) having larger pixels.

Again, I already covered this, summarized by C>P (you may want to go back and read my post again).

The S/N advantage of CDAF doesn't come from the light losses in AF mirrors etc. It comes from the fact that PDAF uses only two very small patches (long but narrow) "cut out" from the exit pupil (I already said so). The loss due to the AF mirror comes extra. And larger sensels do not help.

I don't know but do you the favour to now estimate that the patches are approx. F/5.6 apart and have maybe 1/5 width compared to their distance (probably less for a better phase separation). Which makes them F/28 wide (the real reason why they can be refocussed in an out of focus plane). Assuming the same height than a CDAF region of interest (which can be made larger), then PDAF uses roughly 1/10 the light of CDAF at F/2.8. Give the AF mirror, take the monochromatic sensels, and we roughly stay at that.

I have no way to predict the near future. But even w/o resorting to making use of lens aberrations, the potential of CDAF exceeds that of PDAF. PDAF throws away too much light to get into a simpler algorithmic domain.

[Ayoh]

Thanks for the reply. I know you already covered the C>P issue, understood.

And I agree there is more potential for CDAF due to the smart image processing. Also higher camera resolution is demanding higher optical effort in PDAF relay optics which may approach economic limits in the future.

The additional effective aperture limit of the AF system is a good point which I had forgotten about.
However perhaps 1/5 sensor patch width is a bit pessimistic looking at PDAF sensors with high low light sensitivity like the 1DX or 6D:

http://cdn1.mos.techradar.futurecdn.net//art/cameras/Canon/1DX/Canon_EOS_1DX...or-580-100.jpg
http://www.phreekz.de/wordpress/wp-content/uploads/2012/09/5DII_6D_AF_Sensor.jpg

Also the QE of monochromatic pixels can be in the order of 90% where an averaged bayer unit of 4 pixels is probably around ~30% in practice. This can provide an additional benefit (if not a drastic one as you noted) if reading the same area of the image plane.

So being more optimistic we can maybe reduce your 1/10 PDAF light intensity disadvantage.

Additionally there are other possible specific challenges associated with CDAF:

- Potential delays in AF acquisition introduced by a incorrect initial guess of focusing direction which leads to greater defocus.
- Large defocus which requires significant lens travel before practically measurable difference is registered (such as in telephoto or large aperture lenses). This can conflate with the previous issue.
- Limits on main sensor readout speed at high resolution. Line skipping may be needed reducing effective CDAF exposure per unit area and resolution. Although the PDAF sensor resolution is at a disadvantage here anyway due to the large pixels.

Anyway, on the whole I agree with you that CDAF will probably be the superior approach in the future when the image processing algorithms becomes good enough to minimize the effect of the above disadvantages. Perhaps there is potential for the rapidly developing depth map imaging field to help in tracking too.

BTW why use the term sensel? it does not seem to be a common term. Pixel seems perfectly good.

Again, I already covered this, summarized by C>P (you may want to go back and read my post again).

The S/N advantage of CDAF doesn't come from the light losses in AF mirrors etc. It comes from the fact that PDAF uses only two very small patches (long but narrow) "cut out" from the exit pupil (I already said so). The loss due to the AF mirror comes extra. And larger sensels do not help.

I don't know but do you the favour to now estimate that the patches are approx. F/5.6 apart and have maybe 1/5 width compared to their distance (probably less for a better phase separation). Which makes them F/28 wide (the real reason why they can be refocussed in an out of focus plane). Assuming the same height than a CDAF region of interest (which can be made larger), then PDAF uses roughly 1/10 the light of CDAF at F/2.8. Give the AF mirror, take the monochromatic sensels, and we roughly stay at that.

I have no way to predict the near future. But even w/o resorting to making use of lens aberrations, the potential of CDAF exceeds that of PDAF. PDAF throws away too much light to get into a simpler algorithmic domain.

[philbaum]

With all the information about CDAF possibilities in prior posts, it looks like we're in for interesting times ahead. Sony may be smelling an opportunity to get ahead of the 2 elephants in the field. Noone but Sony has built mirrorless FF cameras before. As quickly as the PS market is currently crumbling, a camera mfr probably doesn't want to be the last company to be making DSLRs because they may be tomorrow's dinosaurs.

[falconeye]

BTW why use the term sensel? it does not seem to be a common term. Pixel seems perfectly good.

Pixel = Picture element
Sensel = Sensor element

As pixel and sensels in a typical digital camera are different from each other, I try to not confuse one term with the other.


BTW, in your image of the AF sensor, a single linear detector seems to have roughly 1/10 the width of its length (assuming the entire bright regions are sensitive). That's capturing 1/10 the light of a square region of the size of the AF sensor length. The neighboring patches receive light from other regions. So, I feel little incentive to update my guestimate

[Ayoh]

Sensel is term coined on photo gear forums by DIY enthusiast. It is not a term used by image sensor technologists - both a 2D digital picture elements and the hardware element are referred to as pixels by convention. If you can find a scientific paper which uses the term sensel in an imaging context (or in any other scientific field) I would be happy to be proven wrong. Eric Fossum himself (inventor of the active pixel cmos architecture) also dismissed the term.

Also, its seems you were previously referring not to the aspect ratio of the PDAF sensor element but to the ratio of width to separation distance, which is implied by "distance" in this statement "maybe 1/5 width compared to their distance (probably less for a better phase separation)".
Regardless, it is the AF sensing area which matters for light collection, which was the focus of my discussion. If we consider the areas of the image plane seen by the PDAF sensors and a CDAF spot focusing readout window, I don't think there will be a 10 fold advantage to CDAF, particularly taking into account the practical reality of line skipping in current CDAF.

From PentaxForums.com: https://www.pentaxforums.com/forums/non-pentax-cameras-canon-nikon-etc/225103...#ixzz2U7ghIjxN

Pixel = Picture element
Sensel = Sensor element

As pixel and sensels in a typical digital camera are different from each other, I try to not confuse one term with the other.


BTW, in your image of the AF sensor, a single linear detector seems to have roughly 1/10 the width of its length (assuming the entire bright regions are sensitive). That's capturing 1/10 the light of a square region of the size of the AF sensor length. The neighboring patches receive light from other regions. So, I feel little incentive to update my guestimate

[falconeye]

Sensel is term coined on photo gear forums by DIY enthusiast.

You quote Fossum from the Wiki discussion at sensel - Wiktionary . And yet, it is a term recognized by wikipedia.

Unlike scientists, forum users are jumping onto any possible misunderstanding.

Therefore, I continue to use pixel for finished images (RGB), sensel for sensors (Bayer or Foveon or 3-CCD or monochrome) and dot for LCD/OLED displays and printers. All three are to be distinguished. Good luck in using one word for all three in forum discussions ... Me, I don't have that time to waste.

You asked, I answered. Why argue with me?

If we consider the areas of the image plane seen by the PDAF sensors and a CDAF spot focusing readout window, I don't think there will be a 10 fold advantage to CDAF, particularly taking into account the practical reality of line skipping in current CDAF.

Until you come up with a different number to quote, I'll stick with my factor 10 (which means more than 5, less than 20).

And again, line skipping should not come up in a discussion about the potential of both technologies. Moreover, only a dumb engineer will omit a fast readout mode w/o line skipping for small enough regions of interest (called "windowing").

Enough off-topic in this thread for my taste. It is about the end of Sony SLT. I think I leave our little discussion alone now. Thanks for pushing me into thinking things over again

[Ayoh]

Thank you for the discussion, you are most informative as always, sorry if I seem argumentative.

As for the "dumb engineer" thugh he is also constrained by not being able to have a concurrent fast windowed sensor read out and a non-windowed read out (as necessary for live view). Maybe some mode switching and buffering skillfulness could be employed but you can't do both at the same time, hence raison d'etre for a windowed readout.

More on topic though I wonder why Sony decided against going with a standard movable mirror in their SLT design. The PDAF sensor could remain in the prism housing as with the current design, but raising the mirror out of the light path during still shooting would avoid any pellicle mirror image compromises. Furthermore, they could pass a larger ratio of light to the PDAF (say 50% rather than the current 20-30%) and with current sensors the live view image and video quality would still be of sufficient quality even in low light. The PDAF during video would still work too (and in lower light).

The disadvantages of the approach seem to be:

- the need for mirror actuation (which surely can't add much to cost - if that is a concern - as such an actuator is a basic DSLR commodity)
- thicker mirror required to withstand shock loading. The extra thickness would shift the image and may introduce some ghosting with a larger offset than in a thin pellicle design. But for live view and video resolution would this really be important?
- not enough advantage over a OVF DSLR to bother. But this didn't stop making SLT which seems ever worse.

Any thoughts?

You quote Fossum from the Wiki discussion at sensel - Wiktionary . And yet, it is a term recognized by wikipedia.

Unlike scientists, forum users are jumping onto any possible misunderstanding.

Therefore, I continue to use pixel for finished images (RGB), sensel for sensors (Bayer or Foveon or 3-CCD or monochrome) and dot for LCD/OLED displays and printers. All three are to be distinguished. Good luck in using one word for all three in forum discussions ... Me, I don't have that time to waste.

You asked, I answered. Why argue with me?

Until you come up with a different number to quote, I'll stick with my factor 10 (which means more than 5, less than 20).

And again, line skipping should not come up in a discussion about the potential of both technologies. Moreover, only a dumb engineer will omit a fast readout mode w/o line skipping for small enough regions of interest (called "windowing").


Enough off-topic in this thread for my taste. It is about the end of Sony SLT. I think I leave our little discussion alone now. Thanks for pushing me into thinking things over again

[falconeye]

Any thoughts?

I like my own conspiration theory most
I.e., Sony made SLT for the sole purpose to eventually migrate the Alpha user base to the E mount.

And then of course the mirror MUST be fixed, to leave an optical advantage to the E mount once no more SLRs are made to compare with. Eventually, Sony wants to sell NEX, not Alpha.

And let's be fair. That's still 10 times smarter than Olympus who simply essentially dropped the FT mount.

[cali92rs]

Eventually, Sony wants to sell NEX, not Alpha.

Seems smart to me...let the elephants (Canon and nikon) fight it out in the FF DSLR arena.
NEX and Fuji are the only real players in the APS-C MILC arena.

[philbaum]

Seems smart to me...let the elephants (Canon and nikon) fight it out in the FF DSLR arena.
NEX and Fuji are the only real players in the APS-C MILC arena.

Noone seems to know where Pentax will position themselves in this future landscape. My fear is they will continue to make APS Dslrs till only the museums want them anymore. I think they are smarter than that, however.

[Clavius]

Noone seems to know where Pentax will position themselves in this future landscape. My fear is they will continue to make APS Dslrs till only the museums want them anymore. I think they are smarter than that, however.

The fact that nobody knows where Pentax is headed, seems like a big advantage. That's why I'm hoping they'll use that advantage and surprise everybody with something really big. Not with a me-too product with which they try to catch up with the big two eight, but with something unique that turns the table around for a while, making the other catch up to them.

[top-quark]

More on topic though I wonder why Sony decided against going with a standard movable mirror in their SLT design.

For full-time phase detection autofocus (see the demo video that Anvh posted the link to). This has considerable benefits for video and continuous AF tracking. The phase detection sensor is at the top of the mirror box, perpendicular to the sensor. If the mirror moved out of the way, phase detection AF would stop. The 0.5 EV light lost to the pellicle doesn't matter all that much - it's the difference between ISO 80 and ISO 120.

Interesting speculation that the A-mount is on its way out. I saw a recent Sony video where the Sony engineering guy said that they bought out Minolta not for the name, but for the mount. That would be a big chunk of change down the drain if they drop it. However, big businesses are concerned not just with profits but earnings per share and it may well be that Sony see no reason to throw good money after bad by continuing to invest in an area where they're not competing all that well rather than redirecting resources to the much more successful E-mount. They can continue legacy A-mount support with their rather good A-mount adapter which effectively turns a NEX camera into an Alpha one complete with pellicle mirror. Then again, the recent release of the top-specced A99 would suggest that Sony are very much committed to the Alpha platform.

[jogiba]

Since the E mount works with full frame sensor like on the NEX-VG900 and there are many full frame E mount cine lenses available now I think the future for Sony is with the E mount. I only have one A mount (35mm F1.4) lens that I use on my FF VG900 with LA-EA3 adapter and all the others are Pentax mount with PK-NEX adapter or E mount lenses. The E mount also works great with the new speed booster adapters for Pentax , Canon, Nikon and other mounts that makes fast F1.4 lenses perform like F1.0 lenses. There are also fast F0.95 E mount lenses available.
Pro Lenses| B&H Photo Video
SLR Magic Noktor 50mm f/0.95 HyperPrime Lens for Sony E-mount NEX Series Cameras SLR-5095E

Imaging Businesses


Placing image sensors, a particularly strong category for Sony, at its core, Sony is concentrating the focus of its imaging businesses on creating value-added products, while aggressively exploring new applications for its imaging technologies in both the consumer and professional markets. In terms of image sensors, the Company will continue to commercialize new sensor technologies capable of differentiating finished products, for use in a range of consumer and professional applications. The Company also plans to engage in aggressive capital investment in order to meet the robust demand for these components. At the same time, Sony is also developing technologies that further expand the range of sensor applications, including sensors capable of sensing beyond the visible light spectrum, and sensors capable of detecting and categorizing different types of information. For the professional market, Sony will continue to reinforce its professional camera lineup centering on 4K-compatible cameras, as well as cameras for cinematography. The Company will also target further business growth by extending the scope of its digital imaging technologies to new business areas such as security, sports and medical, and will reallocate resources accordingly. In the consumer market, where business conditions continue to shift rapidly, Sony aims to expand sales of value-added compact digital still cameras by introducing models that leverage Sony’s image sensor technologies to further enhance image quality, and also incorporate feature enhancements such as reduced size and weight, and higher-powered zoom. Sony will also seek to firmly maintain its number one global market share in the growing mirrorless lens camera category. Through these measures, Sony will target sales of 1.3 trillion yen and an operating profit margin of more than 10% across the image sensor, professional, and consumer categories by FY14.

http://www.sony.net/SonyInfo/News/Press/201305/13-065E/index.html

[Ayoh]

You obviously did not read my post thoroughly. All this was covered.

For full-time phase detection autofocus (see the demo video that Anvh posted the link to). This has considerable benefits for video and continuous AF tracking. The phase detection sensor is at the top of the mirror box, perpendicular to the sensor. If the mirror moved out of the way, phase detection AF would stop. The 0.5 EV light lost to the pellicle doesn't matter all that much - it's the difference between ISO 80 and ISO 120.

Interesting speculation that the A-mount is on its way out. I saw a recent Sony video where the Sony engineering guy said that they bought out Minolta not for the name, but for the mount. That would be a big chunk of change down the drain if they drop it. However, big businesses are concerned not just with profits but earnings per share and it may well be that Sony see no reason to throw good money after bad by continuing to invest in an area where they're not competing all that well rather than redirecting resources to the much more successful E-mount. They can continue legacy A-mount support with their rather good A-mount adapter which effectively turns a NEX camera into an Alpha one complete with pellicle mirror. Then again, the recent release of the top-specced A99 would suggest that Sony are very much committed to the Alpha platform.