Just Touching the Surface of Dual Pixel Technology? [CR1]

[Dylan777]

For higher iso I'm in. Otherwise just another features for video guys.

 

[Rienzphotoz]

This is fantastic ... hope to see an awesome 7D II with cool features and performance.

 

[Lawliet]

Otherwise just another features for video guys.

And for still life/high res guys. Landscape, fashion, and so on.

 

[Marsu42]

Otherwise just another features for video guys.

And for still life/high res guys. Landscape, fashion, and so on.

How so? Not to shamelessly promote Magic Lantern (again ), but focus peaking in live view is terrific for manual focus, and personally I really wouldn't know what I'd want dual pixel af in stills for as I nearly never use contrast af. Great feature if you get it for free with the latest gen cameras, but nothing to write home about unless it's in a mirrorless body.

 

[9VIII]

For higher iso I'm in. Otherwise just another features for video guys.

You wouldn't give up anything for better video performance even if it means that all youtube videos will forever suffer from constant focus hunting?

I would use the dual pixels for a more compact RGBG pixel layout with better colour accuracy, but that's just halfway to increasing resolution by 4 times and getting a perfect RGB signal per pixel (counting four photosites as one pixel).
It would be nice if camera companies would just switch to the same standards as display companies use and count groupings of three sub-pixels as one pixel.

Nice idea, but that's presuming you want to display the image on screen at 1:1 using a current generation display. The problem is people print, people display at other sizes than 1:1, and display technology changes. Compare colour CRT's with their seemingly unrelated pixel and RGB layout, LCD's with predictable pixel to RGB layout, pentile displays etc.

Take video for example. Rolling shutter is a very real problem, but roll back the clock to the very first video camera and TV - a one pixel camera with a spinning Nipkow disk. It had zero rolling shutter because the display device was a single light lit by the electrical output of the single pixel, and another Nipkow disk. Great system, but only good when matched with a specific output system.

The best is surely to get the recorded image as close to theoretically perfect as possible, then as output devices mature (by chasing that same goal), it all looks good regardless. However, with retina displays, high DPI printers and high MP cameras most of us have within reach now, the detailed arrangement of how prime colours are individually captured and reproduced has become almost meaningless.

http://forums.lenovo.com/t5/Idea-Windows-based-Tablets-and/Re-Yoga-2-Pro-13-Yellow-Color-Issues/td-p/1270427

After seeing this I'm not sure if the sub pixel layout is ever going to be less important than it is now.

 

[vscd]

A lot of these ideas sound pretty neat but at the same times some of them don't sound like dual-pixel but rather something that you can do if you just crammed in twice as many normal pixels into the camera? To put it another way, if the dual-pixel system is basically two fully functioning pixels (because that's what some of the ideas seem to be using) what's the difference?

The difference is to use different Voltages/sensitivities to each of those "half" Pixels. With just doubling the Pixelamount you can't shot HDR at one frame because all Pixels are bounded.

The idea was to use one pixel for example from -8EV to 0EV and the other from 0EV to 8EV, making a total of (theoretical) 16EV... the real range of one *single* cell couln't capture 16EV. Today you need to make 2 Shots.

 

[AvTvM]

A lot of these ideas sound pretty neat but at the same times some of them don't sound like dual-pixel but rather something that you can do if you just crammed in twice as many normal pixels into the camera? To put it another way, if the dual-pixel system is basically two fully functioning pixels (because that's what some of the ideas seem to be using) what's the difference?

I see it the same way. With enough (sub) pixels and enough computing power and some clever algorithms/ firmware it would be possible to kill all birds with one stone ... in real time.

• incredibly good hi res - unbinned
• incredibly good Hi-ISO - lower res, binned any which way
• incredibly good DR - combining low/hi ISO takes from pixel-subsets - any which way
• incredibly fast and precise AF - limited only by lens AF-drive capabilities

All of it in a "truly digital" camera body without any mechanics, noise and (internal) vibrations. End of mirrorslapping. Camera size defined by sensor-size, battery-size and ergonomic reasons (grip, balance).

Will be interesting, when Canon (as well as Nikon) finally see the light.

 

[vscd]

I see it the same way. With enough (sub) pixels and enough computing power and some clever algorithms/ firmware it would be possible to kill all birds with one stone ... in real time.

You are totally right, but you assume that every Pixel has his own ability to get controlled by the CPU. The sensor isn't working that way. You can even try to read out one single Bit of your computermemory, but don't affect the other bits by reading from the responding bytes!

 

[M.ST]

The dual pixel AF is only the beginning.

Canon is working very intensely on new functions for the dual pixel technologie.

You will see some new features in the upcoming 7D Mark II and the new EOS 1 series body.

 

[Loswr]

A lot of these ideas sound pretty neat but at the same times some of them don't sound like dual-pixel but rather something that you can do if you just crammed in twice as many normal pixels into the camera? To put it another way, if the dual-pixel system is basically two fully functioning pixels (because that's what some of the ideas seem to be using) what's the difference?

The difference is the dual-pixel method has both of the sub-pixels under one microlens. Two separate pixels would mean a loss of spatial resolution in one dimension or the other, or a 'stretched' image (3:1 or 3:4 instead of 3:2 aspect ratio) if interpolation isn't done. So, you wouldn't need twice as many, but four times as many separate pixels with individual microlenses. For a '20 MP' APS-C sensor like 70D, that drives pixel size down into the ~2um range – PowerShot territory.

 

[Lawliet]

How so? Not to shamelessly promote Magic Lantern (again ), but focus peaking in live view is terrific for manual focus, and personally I really wouldn't know what I'd want dual pixel af in stills for as I nearly never use contrast af.

Some of my favorite models are at their best in the inbetween moments - good luck keeping up with the girls

ML...well, with the 70D or 1Dx I get about 50% more flashes out of a battery, corresponding shorter flash durations and faster recycle times...then with a 5D3. The difference in rental & logistics fees outweight the costs of the camera by far. ML is nowhere to be seen. Now with a high res body - how long will that take? That kind of stuff makes "out of the box"/"officially supported" valuable; without such assertions ML remains a bonus, but can't be a factor in mid- to long term decision making.

 

[Loswr]

...well, with the 70D or 1Dx I get about 50% more flashes out of a battery, corresponding shorter flash durations and faster recycle times...then with a 5D3.

Sorry, but...huh? Of the three cameras you list, only the 70D has a popup flash. How does the 1D X provide shorter flash durations and faster recycle times than the 5DIII for an external flash? Since you ascribe the same benefit to the 70D, I assume you're not referring to something like using a higher ISO. Can you explain?

 

[Cali_PH]

Interesting rumor of a new sensor developed by Hasselblad and Sony paralleling some of the ideas discussed here.

"Every single pixel can have a different shutter time! This means the sensor allows a dramatic increase of the dynamic range. What sources didn’t tell me is how exactly this works and if the sensor is going to be first used by Hasselblads new medium format camera or by a new generation of FF sensors. Anyhow, its great news to see that Hasselblad is working on some exciting new tech with Sony!"

 

[Loswr]

Interesting rumor of a new sensor developed by Hasselblad and Sony paralleling some of the ideas discussed here.

"Every single pixel can have a different shutter time! This means the sensor allows a dramatic increase of the dynamic range. What sources didn’t tell me is how exactly this works and if the sensor is going to be first used by Hasselblads new medium format camera or by a new generation of FF sensors. Anyhow, its great news to see that Hasselblad is working on some exciting new tech with Sony!"

Sounds interesting…at least for static subjects.

 

[Cali_PH]

Sounds interesting…at least for static subjects.

Yes, I was wondering how that'd work, what settings one would use vs. what the camera actually does...at least for my main personal interest (landscape), I wouldn't have to worry much. But for something moving fast...you'd get some interesting mistakes. If that's something they (or someone else) has experimented with, I'm guessing the early test shots were very interesting. ;D

Of course, the rumor could be incorrect about different exposure times, and it's actually different ISO's as some have discussed here. Or just incorrect altogether.

 

[Don Haines]

Of course, the rumor could be incorrect about different exposure times, and it's actually different ISO's as some have discussed here. Or just incorrect altogether.

NO!!! Not an incorrect rumour! Say it isn't so....

I find it interesting that all of a sudden dual pixel technology has popped up in several sources as rumours and that Canon and Olymus (to a limited degree) have it on the market. Since this is something that has taken at least 5 years to go from the labs to the marketplace you can bet that everyone is working on it... my bet is that in a couple of years everyone will have it on all thier DSLRs and mirrorless cameras...

 

[AvTvM]

I find it interesting that all of a sudden dual pixel technology has popped up in several sources as rumours and that Canon and Olymus (to a limited degree) have it on the market.

My understanding is that currently only Canon 70D (and C100, possibly soon also C300) utilize "split/dual pixel"-on-sensor PD-AF with sensels on 80% of sensor surface useable for on-sensor PD-AF - as well as capturing light for image data just like any "regular" sensel. http://www.dpreview.com/reviews/canon-eos-70d/3

All other hybrid "on-sensor phase-detect-AF" implementations seem to be of the "older" (2010) "Fuji-type" http://www.dpreview.com/news/2010/8/5/fujifilmpd where only a small number [e.g. 99 for Sony NEX 5R] of "special-purpose" AF-pixels [partially masked off] are used for PD-AF purposes. These sensels do not capture light for image data, the blanks have to be filled by interpolation.

This latter approach seems to be used by a number of companies:

• Canon in EOS 650D, 700D, EOS-M and
• in an improved version II with wider coverage of sensor area in the SL-1/100D http://www.dpreview.com/reviews/canon-eos-100d-rebel-sl1/6
• by Oly in the OMD5 and OMD1 and
• by Sony - first in their NEX-5R http://www.dpreview.com/previews/sony-alpha-nex-5r/3, then also in NEX-6 and NEX-7 and maybe in a slightly different version in the Alpha A7 (but not in A7R which uses CD-AF only)
• and also in a number of Fuji cameras

Panasonic also filed an on-sensor PD-AF patent in 03/2012 http://www.freepatentsonline.com/8482657.html - from the looks of one of the illustration images, it seems to also use a finite number of designated PD-AF sensels on the sensor, but has a rather different layout with (separate) PD-AF line sensors behind transmissive image sensor layers and condenser lenses.

The Sony/Hassy rumour - if true at all - would be yet an all together different thing.

 

[Lawliet]

Since you ascribe the same benefit to the 70D, I assume you're not referring to something like using a higher ISO. Can you explain?

Its about the sync speed, the 5D3 is noticable behind there. The additional power required to balance with the increased influx of ambient light takes its toll on all fronts.

 

[Loswr]

Since you ascribe the same benefit to the 70D, I assume you're not referring to something like using a higher ISO. Can you explain?

Its about the sync speed, the 5D3 is noticable behind there. The additional power required to balance with the increased influx of ambient light takes its toll on all fronts.

I wouldn't have thought 1/3 of a stop would make that much of a difference...

 

[Lawliet]

I wouldn't have thought 1/3 of a stop would make that much of a difference...

With fast triggers its 2/3 of a stop, or a full one if you allow for the same amount of shading, difference(there is a reason the manuals are quite YMMV in that regard), I.E. twice the number of packs, no more lightweight heads, but bitubes that each cost not much less then a 1Dx. Or a D800&a nice set of lenses.
Enough difference to put it rather high on my priority list.