Patent: Sensor Technology

[Canon Rumors]

<p>A new sensor patent from Canon has appeared and continues to add to what are sure to be new full frame sensors next year.</p>
<p><strong>NL breaks down the patent…</strong>

“Patent from Canon that varies the sensitivity of pixels by positioning the photon detection zone deeper in the design.”</p>
<p>Source: [<a href="http://www.northlight-images.co.uk/cameras/Canon_rumours.html" target="_blank">EG</a>] via [<a href="http://www.northlight-images.co.uk/cameras/Canon_rumours.html" target="_blank">NL</a>]</p>
<p><strong><span style="color: #ff0000;">c</span>r</strong></p>

 

[CaptureWhatYouSee]

Does anyone see hope in this as a realistic approach?
Realistic : cost effective + an actual improvement + ..

 

[Marsu42]

<p>A new sensor patent from Canon has appeared and continues to add to what are sure to be new full frame sensors next year.

CR quotes EG quotes NL quotes a google translation of the patent - now that's a [CR-1] for you :->

But next year? How so? 2015 isn't that far away, and depending on their sensor research and time it takes to develop a 5d3/6d2/... it might take longer than that.

Last not least, with Canon being Canon, they'll only implement costly sensor design changes if they feel they have an absolutely pressing need for it, like essential changes for a high res camera "3d".

 

[Coldhands]

I get a sinking feeling that this is more geared toward video than still.

Optical attenuation = less light. So they are sacrificing sensitivity for the sake of dynamic range. I can only see this as a worth-while tradeoff in the world of video where noise is not as noticeable, and DR boosting techniques (ND grads, exposure fusion, etc) are not necessarily viable.

Your interpretation may will almost certainly differ

 

[Marsu42]

Your interpretation may will almost certainly differ

I don't know about this specific patent, but Canon's low-light security camera sensors show the way. Magic Lantern's raw video has also shown a large demand for higher dynamic range with video.

As Canon seems to be unwilling or unable to compete with Sony's high-res sensors, the way out might be further integration of video and stills. If Canon "mirrorless" settles for a 4k resolution sensor and enables high-quality frame grabbing @120fps, that would be a product meeting a high demand for a lot of purposes.

 

[Orangutan]

Optical attenuation = less light. So they are sacrificing sensitivity for the sake of dynamic range.

Not just DR, but highlight control. It would allow videographers or landscape photographers to expose for shadows. On the other hand, combine this with BSI, and they could extend exposure latitude in both directions.

I believe things like this have been done before, but in the past it's been a mini-photosite between the regular photosites. This process might be cheaper/easier to produce since the photosites are all the same size, though some are "submerged" a bit.

 

[Woody]

As Canon seems to be unwilling or unable to compete with Sony's high-res sensors, the way out might be further integration of video and stills. If Canon "mirrorless" settles for a 4k resolution sensor and enables high-quality frame grabbing @120fps, that would be a product meeting a high demand for a lot of purposes.

I also think the future lies in the ability to combine high quality video with stills. Let's see...

 

[Woody]

CR quotes EG quotes NL quotes a google translation of the patent - now that's a [CR-1] for you :->

Well, here's the original Egami article on Canon's patent
http://egami.blog.so-net.ne.jp/2014-11-29

 

[Orangutan]

As Canon seems to be unwilling or unable to compete with Sony's high-res sensors, the way out might be further integration of video and stills.

Viewed from another angle, Sony, Nikon, et al, have been unwilling or unable to compete with Canon's DPAF sensors. (Maybe Samsung now has) For most camera owners, good-focus vs. bad-focus is a larger determining factor in deciding whether a certain photo is worth keeping than is the superior sensor. It seems more like conservative Canon is putting its R&D investment into profitable advances, rather than niche advances. Of course I'd like dramatic sensor improvements from Canon; however, as I and many others have said before, it won't happen until the market demands it. When that eventually occurs, I have no doubts that Canon will be able to deliver.

 

[IndustrialAndrew]

Next year does seem like a very short turn around time from chip manufacture, to camera design, manufacture, testing and then supplying to market. Two years seems like a short time too.

 

[crazyklaus]

Next year does seem like a very short turn around time from chip manufacture, to camera design, manufacture, testing and then supplying to market. Two years seems like a short time too.

That we first see that particular patent right now does not mean they only started working on this technology yesterday, does it?

Having some less sensitive pixels on the sensor would benefit landscape or other single-exposure HDR shooting, especially if combined with an on-sensor ADC technology.
On the other hand it would hurt low-light-performance a bit.

 

[pedro]

what I could get out of it from the link provided filtering the japanese signs through babylon translater:
A patent to extend dynamic range in the pixel that Canon transmissivity is different[a semiconductor / electronics technology]
Canon HDR Explanation / the self-interpretation of the patent document Patent public number 2014-175,553 A Pub. Date2014.9.22 An application date2013.3.11 A patent of Canon A light decrement film by the poly-silicon arranges it on photoelectric conversion department The light decrement film is different in transmissivity of the light depending on the thickness, and the thickness is different by a pixel The small pixel of the transmissivity has a small opening area A color filter of RGB either The transmissivity is R<G<B A patent of Canon That sensitivity is different by a pixel isn't the thing that various makers are rare of patent application. About a method to let sensitivity be different, various patent application will be considered to be it to evade a patent. For example, a certain maker is applying for the patent of the different ND filter of the transmissivity. Then I seem to have established the light decrement film by the poly-silicon between a filter and photodiode when it was said what the patent application of Canon was different in. Because many makers are interested in technology to let the sensitivity of the pixel be different, manufacture is expected.

 

[emag]

what I could get out of it from the link provided filtering the japanese signs through babylon translater:

I think I sprained my brain reading that............l

 

[pedro]

oh yes I can imagine ;-) maybe there is anybody out there who dominates japanese, or who knows how to transform these phrases into something like english...;-) cheers, Pedro

 

[Loswr]

Next year does seem like a very short turn around time from chip manufacture, to camera design, manufacture, testing and then supplying to market. Two years seems like a short time too.

That we first see that particular patent right now does not mean they only started working on this technology yesterday, does it?

Of course not...we see a patent when it publishes – and that's generally 18 months after it was filed (and obviously the research must have matured to a certain point to support the patent application.

 

[Mt Spokane Photography]

Next year does seem like a very short turn around time from chip manufacture, to camera design, manufacture, testing and then supplying to market. Two years seems like a short time too.

That we first see that particular patent right now does not mean they only started working on this technology yesterday, does it?

Of course not...we see a patent when it publishes – and that's generally 18 months after it was filed (and obviously the research must have matured to a certain point to support the patent application.

Patents are usually the result of years of research which then results in a patent application, and 18 months later, a patent.

This patent references a 2004 patent from Fujii (2004-32059) which also used two pixels to increase DR. This one is different in the implementation, but is the same basic idea. One pixel is a high sensitivity pixel and the other is a low sensitivity pixel, its just a matter of how the sensitivity is controlled. The net result of combining the outputs of the two pixels is to increase the dynamic range.

The technology to do this has been around for many years, its just a matter of it producing enough benefit to put into production.

I'm reading it as another one of those protective patents intended to protect the results of a lot of research.

However, I think it is quite possible if Canon chooses to go that route.

I'm thinking that their next camera will use a DPAF or a layered sensor and that this technology does not provide enough benefit compared to other things in the pipeline.

 

[dgatwood]

It seems to me that it would be easier to just crank up the full well capacity, or better yet, add a full buffer-based electronic shutter and sum multiple short exposures.

 

[hpjfromdk]

For those interested you may find the US version of this patent: http://www.freepatentsonline.com/20140253767.pdf a bit easier to read

 

[Mt Spokane Photography]

oh yes I can imagine ;-) maybe there is anybody out there who dominates japanese, or who knows how to transform these phrases into something like english...;-) cheers, Pedro

The Japanese Patent site does a better job of translating than Google, but in this case, the patent was also filed in the USA, so its there in plain technical jargon English. You still have to wade thru a ton of obscure language.

 

[mb66energy]

Does anyone see hope in this as a realistic approach?
Realistic : cost effective + an actual improvement + ..

Seems to me like a very good idea: Light sensitive pixels in the focus of the micro lenses and less sensitive pixels with SAME FULL WELL CAPACITY between the microlens focus regions. So it is a modified dual pixel design ...: If the more sensitive pixel is saturated the less sensitive pixel is still measuring valuable data. More than nice to have and something which can give us 15-20 stops of dynamic range. But perhaps not at 10 fps with 50 MPix ...

Thanks Woody for the english/US patent - just scanned it and was much better than the japanes->german translation by go og le tra ns lations.