Canon EIS 60 revealed? High end mirrorless camera with pixel fusion technology

57


*Diagram and logos for illustration purposes only*

Detailed information has just been leaked onto the internet regarding Canon’s mirrorless camera, which right now is posted all over internet forums in Asia. No idea how yet – may be a marketing leak, trying to find more.

Due for launch in the second quarter of 2011, Canon’s mirrorless is branded EIS. The first in the series – EIS 60 – is a high end mirrorless camera, not an entry level GF / NEX competitor. It will feature a 22 megapixel sensor. This is a brand new back illuminated design, 18mm x 12mm in size (approximately Micro 4/3rds sized*).

EIS stands for Electronic Image System, as opposed to the famous Canon EOS moniker which is Electronic Optical System.

Canon are developing new EIS lenses, at least two of which will be available at launch – these are the 12-75mm F2.8-4 IS Macro and 70-300mm F3.5-5.6 IS, which retract into a much smaller barrel than EOS lenses.

The camera also has a feature that windows the centre part of the sensor. This will be great for c-mounts. In this sensor windowing mode, the camera’s continuous shooting rate increases to a whopping 20fps and resolution drops to 5.5 megapixel whilst offering a tighter crop of the image – great for sports shooters.

In addition to sensor windowing, the EIS 60 has a new Canon technology we’ve not seen before. I am dubbing it Pixel Fusion for now since the leaked information doesn’t assign a name to it.

Video is sampled using ‘9 pixel fusion’ according to our source. Instead of scaling the image by binning pixels, 9 pixels are merged together to form just 1.

Visualise this as multiple square matrixes of 3×3 pixels on the full 22MP sensor, merge each block together to form 1 and you have a 1920×1080 frame but using all the data on the sensor. This is an absolutely mega idea, because it means that each pixel in video mode is effectively produced from 9. That will increase dynamic range and reduce noise. What a great way to scale the 22MP sensor.

The pixel merging technology is hardware supported on the sensor and new DIGIC according to the source, and is also used to produce higher ISOs in stills mode. The EIS 60 has ISO 100-6400 with an extended range of to 12800, 25600 in 5.5 megapixel pixel fusion mode.

Canon have further lenses in development, not all will be available at launch. However it pleases me immensely to see all those fast primes – something Panasonic just did not deliver on with the Micro 4/3rds line-up. To quote our source via Google Translate, starting from wide-angle to telephoto, here is the full list from R&D:

• 5mm F4 Fisheye
• 8-25mm F4 wide-angle zoom
• 14mm F2 pancake – which our translation of the source’s dialect hilariously termed ‘biscuit head!!’
• 25mm F1.2 pancake
• 45mm F1.5 standard biscuit head(!)
• 65mm F2.0 Macro (1:1, 2:1 is equivalent to full-size)

Canon will of course launch an EF – EIS adapter. EF lenses will auto-focus on the mirrorless body and aperture can be selected (which is electronic on EF and EFS lenses). This is not something you can do on other systems right now – of course Canon have built the electronic contacts into the ring, as only they can. The lens communication protocols are their IP and the only other company to license them so far have been RED. The camera will be compatible with most EF lenses out of the box. The source does not mention whether OIS will function on EF lenses – I expect it will be powered and operational just fine.

The EIS 60 has dual SDXC memory card slots and can capture stills whilst recording video via the main shutter button. It has a dedicated video record button alongside it.

If this leak is genuine, it sounds like the be all and end all of mirrorless cameras in 2011.

It is great to see this ‘future’ form of DSLR breaking out of the rut it has found itself in during 2010 with the glut of lower end step ups from compacts, like the GF and NEX series all pandering to the lower end of the market.

Mirrorless cameras have to continue to push the concept of what an interchangeable lens camera is forward into the all-digital future, unlike the lower end models we’ve seen. The GF1 and NEX 5 only place an emphasis on size and ease of use relative to DSLRS.

It is clear that Canon are not following suit and will make a highly advanced mirrorless system with a fantastic video mode.

* Standard Micro 4/3rds is 17.3 mm × 13.0 mm. Canon’s is 18mm x 12mm because it is in 3:2 aspect ratio not 4:3. The GH1’s non-standard M4/3rds sensor is multi-aspect, supporting 16:9 and is slightly larger than both.

Update: Here is a roughly translated version of the link given to me by my forum member. It is not the original source, that is here

Share.

About Author

British filmmaker and editor of EOSHD, Andrew works in Berlin on his own self funded filmmaking and video projects.

57 Comments

  1. Andrew, the 9 pixel fusion idea interests me too, it’s no coincidence I’m sure that the sensor width is exactly 5760 (3×1920). What I’m trying to figure out is how it would work, I’m hoping it works by binning every common color in a 3×3 cell so you would get a “superpixel” that has a complete set of R, G and B data. A traditional Bayer mask would work but it may be better to use a nonstandard one such as this: http://i73.photobucket.com/albums/i202/richtrav/alternatebayer.jpgwhere all 3 colors are represented 3 times in a 3×3 grid (this also easily bins into a 2×2 pattern that forms a traditional Bayer mask). At any rate it’s exciting news and it’d be nice if Canon had some info at Photokina

  2. Let’s hear it for RichST guys – he deserves the credit for discovering this info.The idea is intriguing and everything that Asian source stated basically makes 100% sense. If he is making this up then he damn well has the maths right! At the very least he must have thought *incredibly* hard about the specs. Another thing that stood out for me was the 18 x 12mm sensor – it makes perfect sense since this is basically a 3:2 aspect ratio Micro 4/3rds sized sensor. We all know that Canon’s sensors are 3:2 and not 4:3 like a Micro 4/3rds Live MOS.So… the way I’d summarise a camera like this is – well, hot. If this is basically a very compact 7D with EVF, some pixel fusion tricks and a better video mode, as well as being mirrorless with fast primes and adaptable to just about every lens on the market, well… who the hell would NOT buy one?

  3. Hehe, thanks but I have to confess I saw it on tucked away in another forum (why it wasn’t leading news I have no idea). The camera just makes SENSE, it’s using much more sophisticated binning and possibly cropping than what we’re seeing today and it’s using it for a variety of uses: high quality video, high ISO, ultrafast burst rates, etc. It sounds like they’ve started from scratch in designing this thing, having rethought their decade-old ideology of what a camera sensor should be. Am crossing my fingers this thing is real

  4. It is very feasible and well within current technological limits I believe. If it isn’t real then whoever thought of it in Asia should go straight to the Japanese HQ of Canon and get a bloody job!! :)

  5. great and fabulous news. except for one thing only – why, oh why, oh why oh why a m4/3 sensor size?!? Why not make it APS-C ? it could still have crop mode, but i absolutely hate the m4/3 sesor size.
    you cant use cine16mm lens with it, c-mount lens “barely” covers it and regular slr lens are never wide enough ’cause of the crop factor.
    I will never like that sensor size and canon is only making it worse for me if they put out this camera :)

  6. This all sounds feasible, except for some of the lenses. I’m not sure even Canon could create a f/1.2 or f/1.5 “biscuit head” lens. Usually you need some more depth to focus & resolve the increased light capture of a fast lens.

  7. Fascinating rumor – only one thing on this bugs me though.
    You said “I have had to temporarily remove it to prevent people from linking to that page rather than EOSHD as the story breaks.”
    So let me get this straight – you won’t post the link to the ACTUAL article because you want people to link to your page instead of the actual source of the information? Makes me wonder if you’re more worried about your blog’s popularity than you are about the information.
    Bad move, EOSHD, bad move.

  8. This is a sketchy rumour. A lot of things don’t add up, especially the far-too-good-to-be-true lens lineup of 8 awesome lenses already planned for production:1. Backlit design does not yield much improvement at the 4/3 sensor size… why would Canon waste time and money pursuing it?2. It would be impossible to design “pancake” lenses with specs of 14/2, 25/1.2, and 45/1.5. Not that those lenses aren’t doable, but they would never be small enough to be considered pancakes — at least not by what the typical defintion of a pancake lens would be. They are spec’d WAY too fast to be pancake-sized. 3. 8-25/4 WA zoom and 5/4 fisheye? Are you kidding? Nobody has the technology to produce lenses that wide or with such a large range. If Olympus and Panasonic couldn’t do it with m4/3, what makes one think Canon can do it? Canon can’t even get the corner performance right with their 16-35/2.8L MkII and this was their second shot at it.4. 12-75/2.8-4 IS? A 6x variable aperture zoom starting that wide and that fast with IS? Again, if Olympus couldn’t do it, how could Canon? 5. 65/2 1:1 macro? I mean, seriously. Who’s going to come out with a 130 mm equivalent f/2 macro as their first macro offering? What company has ever produced a 130/2 1:1 macro before? And how can Canon make such a fast macro 1:1? Olympus could only produce a 1:2 macro for their 50/2.6. It looks like someone conveniently took the 4/3 lens catalog and just “expanded” the focal length ranges or made the max aperture faster: – 7-14/4 becomes 8-25/4- 12-60/2.8-4 becomes 12-75/2.8-4- 14/2.5 becomes 14/2- 25/1.4 becomes 25/1.2- 50/2 macro becomes 65/2 macro7. There is no mention of the auto-focus technology (CDAF/PDAF, etc). This is arguably THE most critical aspect of such a camera, but not much if any details regarding it. Only the silly 22 MP super-binning sensor and fantasy-land lens lineup are specified in great deal.8. No mention of the video encoding specs nor the audio in/mic configuration, which is very important for videography.9. No mention of the lens mount specs – flange diameter, back-focus distance, etc., which is important if you’re going to announce a new lens mount and system. 10. Where are the details on the EVF specs? This would be very important for a mirrorless camera, no?11. SDXC cards wouldn’t have the speed to handle the sheer data rate generated by 22 MP for any length of burst, nor the 5.5 MP x 20 fps that this camera is claimed to achieve unless you’re expecting 45+ second buffer to card write times.The typical emphasis on unusually attractive sensor and lens specs and disregard for the other important aspects of the camera leads me to believe that it’s just some camera geek’s dream that’s not well-grounded in reality.

  9. Those lenses sound really fake to me. It’ll be amazing if true- but this doesn’t feel legit to me.

  10. As Mr Yang pointed out above, the original source is http://forum.xitek.com/sorthread.php?threadid=782342

    I did not make up the rumour, I am giving you my interpretation of it. Nothing to do with traffic my friend. The link I myself was emailed by RichST is now edited into the article and I am posting it here for you http://translate.googleusercontent.com/translate_c?hl=en&ie=UTF-8&sl=auto&tl=en&u=http://pp.520dc.com/space.php%3Fuid%3D56%26do%3Dthread%26id%3D3021&prev=_t&rurl=translate.google.com&twu=1&usg=ALkJrhirPwWwYI2VLD2l-PW69AvxaE5iAA

  11. Panasonic in one of their compacts uses this same method by combinig pixels, (lowering the megapixel output of the image) to offer higher dynamic range and better noise suppression for high iso images, whilst also giving them higher s/n.

    Brilliant idea to use it in video. . .

  12. 22 mpx on a sensor smaller than APS-C?
    Kinda too much.
    Pixel Fusion [insert trumpets here] is nothing but a variation of pixel binning, which is being used in imaging sensors (mainly CCD’s) for over a decade. Maybe even over two decades already.
    It’s pro’lly being done off-sensor.
    But instead of 3×3 binning (GRG/BGB/GRG ???) why didn’t Canon go for 2×2 (GR/BG)?
    It would require only a 10mpx sensor, with larger photosites, possible better dynamic range and SNR. It would allow for smaller image files, faster FPS, both in video and photo.
    Marketing over engineering. Bleah.
    Otherwise, a more than welcomed change in the photo-tech world.

  13. It’s a 20/1.7 projecting an image circle onto a primarily 4:3 aspect ratio sensor – not as difficult. Panasonic’s been planning for this and doing it for longer than Canon has.

    For the 14/2, we have a good benchmark to compare against with Panasonic announcing a 14/2.5 soon. Originally slated to be a 14/2.8, they went back to the drawing board to make it faster due to the market success of the 20/1.7. They essentially “maxed out” the lens design at f/2.5 designed to work primarily with a 4:3 aspect ratio. How can Canon hit f/2 on a shortened development cycle for use with a 3:2 aspect ratio and make it pancake-sized? They would have to bust out some technology that i haven’t seen Canon use before, like elements with aspheric surfaces on both sides.

  14. The amount of correction required for 14/2 and 25/1.2 lenses is very difficult to achieve when you specify extreme size constraints such as a pancake lens design. Think of how coma and astigmatism increase exponentially with aperture size. How many elements do you realistically need to have good enough corner performance for a 22 MP sensor?

  15. Regarding following comments:

    3. I have no doubt the 5/4 fisheye is doable. It might be a circular fisheye, not a fullframe one. Look at Sigma, they have a 4.5mm f2.8 DC circular fisheye lens. The 8-25/4 would be like a cropped frame 16-35/4 VR Nikkor, only that a little longer at telephoto end…

    4. This would be like the 12-60mm 4/3rds lens, with IS and little longer and telephoto end. It does not say it will be smaller in size or lighter. Why not possible?

    5. Tamron has currently a 60mm f/2 Di II 1:1 macro lens (no VC). So again, why not a 65mm f/2 EIS 1:1 macro? Just because you do not like the focal length?

    Even as unlikely this lens lineup is (like missing a 14-45mm f/3.5-5.6 standard kit lens), is not that lenses are not doable. Current examples of similar technology exist. Heck, maybe the pancakes are doable if they sacrifice distortion and 2/3’s of frame sharpness…

  16. For the record, some of the rumors he has also spread lately were a Canon 24-85 2.8L IS silent stepless aperture lens and an NEX 16-80/2.4-4 and 8-400/3.5-5.6 OS lenses. The Sony lenses are supposed to be announced in September, so we’ll see if he gets either right.

  17. 2×2 binning makes little sense with a traditional Bayer mask, it works better with alternative arrays (which could well be the case here if this is really a product under development).

  18. Yes, those lenses are technically doable, but what of the familiarity of traditional focal lengths and a “regular” fisheye. For the circular fisheye and 130 mm-eq f/2 1:1 macro, why would a company like Canon set out to design oddball lenses for their new system when they would presumably try to gain as much market acceptance as possible from the start to maximize their chances of success? The answer is that they’re unlikely to do so. Follow-up lenses can be oddball focal lengths and circular fisheyes, but not the foundation lenses of a system.

    And yes, while the lenses are technically doable, but at what cost? Like you said, they could sacrifice considerable distortion control and edge sharpness, but why would they if they want to position this as a premium mirrorless system with a 22 MP sensor?

    If they wanted to pull off good IQ across the frame and have small zooms / pancake primes, it would be optically very challenging, and therefore expensive to design and produce. Does Canon have the guts to have this lineup essentially be an “all L-glass” lineup starting in the $1000 range? Even if they did, how many mirrorless buyers are going to blow thousands on lenses? Are they going after 5DMkII customers? The Rebel T2i/550D buyers aren’t going to be able to afford this stuff.

    Let’s face it. The more one has to justify how various lenses are, in fact, technically doable, the more one should realize that the specs provided in the rumour are not realistic, and the production of such a system would be impractical.

  19. Just to be clearer, those lenses are apparently on an R&D dev programme and will not see the launch, they probably won’t be due until well into 2011. No idea if they are doable or not to be honest! Biscuit or no biscuit.

  20. I wouldn’t read too much into the responses on the forum, they give no reason for not trusting him. Seems like a healthy dose of cynicism to me! I am not ruling anything out. It all makes sense to me.

  21. 8-400mm – wow, that’s a little crazy. Well, let’s give him the benefit of the doubt, but I guess this is one reason not to trust him, isn’t it?

  22. Why would I want to go and buy that aliasing and rolling shutter nex camera ??? Dont try to turn this into an canon vs. sony thing, or whatever vs. whatever. I was just saying saying that i dont like m4/3 sensor size because it is dificult for me to fit appropriate slr lenses on it.
    and it is my own personal preference and opinion, not a fanboy attitude (likesome) for any of the camera manufacturers.
    The canon pixel fusion however is great (if its true)
    It would be wonderful if finally after two years after 5dmk2 we see the next step in video processing.. Who knows, maybe till the year 2015 , we have an APS-C sensor with no rolling shutter, no aliasing, no moire, and have it cost less that 4000 $. I bet that the technology is there, they just have to pay off their expensive BS cameras and camcorderes that they made till this day.

    cheers

  23. Why not? Canon has most definitely expressed an interest in the concept, yes the “biscuit” lenses (maybe they’re fatter pancake lenses) are a little iffy, but still a lot of the ideas behind the proposal make lots of sense. The pixel density is high enough that backwiring might improve IQ. If extensive use of “fusion” (sounds like on-sensor pixel binning to me) is going to be incorporated into the camera it may take more wiring than normal for a typical sensor, another possible reason to use backwiring.

    I’m still thinking about what the sensor mask would look like, and keep coming back to the one I posted previously (http://i73.photobucket.com/albums/i202/richtrav/alternatebayer.jpg). This pattern would be easy to bin in 2×2 blocks to form a 5.5 megapixel bayer pattern by binning just one color out of the 4 cell block, while a 3×3 array could bin 3 pixels of the same color diagonally to produce a superpixel, it just wouldn’t be a square group of 9. Or the sensor could have multiple wiring for different binning patterns, one for a 2×2 mode and a completely different one for a square 3×3 mode, I’m not a sensor designer so I’d like to hear the insight from anyone who might know how feasible that is.

    His lens proposals may seem a little off but the idea but I think the idea is to keep it small, and the theory behind the sensor is good and the details in it make some sense. Fuji has already been doing it for years. Dropping the resolution to 5.5 would double the sensitivity of the camera and cutting it by a third would produce a perfect video mode if all 3 colors were sampled for each output pixel. In both cases sensitivity would go up. Canon already uses 3 pixel binning on their VDSLRs, they just do it in one direction unfortunately

    And for those who blow this off as ridiculous (which I admit it may very well be), how many of you would have thought a month ago Canon would announce a 120mp APS-H sensor or a monster sensor the size of a sheet of paper? They’re clearly not sitting on their arses, the 60d notwithstanding

  24. I am not quite as impressed with the concept or the ideas from that poster. It’s overly complicated and incurs many other “costs” to pull it off. I cannot see how the benefits outweigh the drawbacks, and must conclude that this is too inelegant and impractical a design to either be real or be given a green light by the powers that be at Canon. And it’s not just the lenses that are a little off. I just thought about the impracticalities of the sensor. To wit:1. 22 MP on 18×12 (probably more like 17.5×11.6 mm effective) results in extremely high pixel density. 18 MP on APS-C is one thing. But 22 MP on the smaller 18×12 sensor? We’re talking a 3.0-3.1 micron pixel pitch here. That’s equivalent to the pixel density of a 40 MP APS-C sensor or a 96 MP FF sensor!2. Such extreme density results in small pixels that are not able to collect enough photons to generate a clean signal in low light as well as being limited in full well capacity, greatly reducing dynamic range. So the noise performance and dynamic range will be roughly half-way between a Canon S90 (2 micron pixels) and a Canon 7D (4.3 micron pixels).3. Aforementioned tight pixel spacing also results in diffraction starting as early as f/5.6! (Airy disk diameter is ~5.4 microns at f/4 in green light). Diffraction might not be very noticeable until f/8, but consider that you’re only getting maybe ~15-17 MP of resolution at f/5.6 and ~6-8 MP at f/8. 4. What kind of lenses would be required to resolve the fine detail required for a 3 micron pixel sensor? Again, that’s equivalent to the pixel density of a 40 MP APS-C sensor or a 96 MP FF sensor. How many lenses do you know of that can do a 40 MP APS-C sensor or 96 MP FF sensor justice? And to connect this with my previous post wherein i questioned just how feasible it was to design and produce lenses with such range, speed, and compactness, how could they make those lenses and have them perform at a level befitting of such a demanding sensor?So i still have to wonder: why is using a 22 MP sensor a good idea? I submit that it is not. … which only serves to reinforce my original assertion that this rumour is sketchy and not grounded in reality.

  25. Hmm I guess Canon’s 120mp APS-H sensor must have you in a tailspin (just kidding!)

    Seriously though, the concept of pixel binning has been around and is hardly what I’d call “overly complicated”; many video modes in cameras today utilize it and Fuji at least uses it to expand dynamic range and reduce noise. It does so at the expense of resolution (I believe the article stated that Canon’s camera would do high ISOs in 5.5mp mode). And the idea of altering the Bayer mask is not some complex fantasy I just dreamt up, Fuji has already done it specifically to better optimize its binning: http://www.fujifilm.com/products/digital_cameras/exr/features/page_03.html

    I do admit the pixel count is awfully high, and this sensor does sound like it’s something of a hybrid between an APS-C and 2/3 system in size (not that surprising considering Nikon’s EVIL system under development, or at least one of them, is rumored to be based on a 1″ sensor design). But the better small sensor cameras do pretty well for themselves when it comes to resolving quite high detail with much tinier pixel pitches

    But again you’re ignoring the advantages to binning, it lets you speed up readout if you want good video, it lets you decrease noise if that’s your objective and it might even be possible to increase dynamic range if it the adjacent pixels could be read at separate ISO levels. Yes this would all come at the expense of resolution but that’s the point, create a high resolution sensor and you can afford to cut down the mp count to suit your needs

  26. BSI (back illuminated) circuitry on a CMOS really benefits sensors with dense pixel pitches, where the circuitry starts to outweigh the photosites for physical space on the sensor. So I do not believe that 22MP is impossible with a BSI 18x12mm CMOS. And if the whole thing is made up, kudos at least to the maker-upper in recognising that ;)

  27. I’ve never been impressed with the IQ from Fuji’s sensors, my S5 Pro notwithstanding, nor with pixel binning in general, but admittedly, there is room and potential for improvement. I don’t think it will ever be good enough for top notch still image quality, and feel that lower density sensors with bigger pixels is “the better way”.

    Ultimately, it is reflected in the fact that my priorities are for maximum still image quality over video. There are likely philosophical differences at play, being that i am a stills photographer posting on a website dedicated to video… it could very well be that i am the one who is out of touch with the changing realities of convergence! ;)

  28. Hi You blog technically interestingly well. I suppose you can influence Canon to include a vari-angle lcd in their mirrorless cam so it will be an all around shooter. I mean after working/shooting all day long about anything I can include a self shot self portrait of myself as a user. All work and no play makes Jack a dull boy they say.

  29. “Pixel Fusion” is something already well known in astrophotography community as a pixel binning, and is in use almost as long as CCD astrocameras. So, nothing revolutionary, just for the first time applied in regular cameras. Astro-CCD cameras are monochrome, so there you can use also 2×2 pixel binning as well as 3×3 pixel binning, whilst with Bayer filters you have to skip 2×2 and go to 3×3 pixels binning (fusion ;)).
    And yes adessio, pentaprism bothers me, too.

  30. “Pixel Fusion” is something already well known in astrophotography community as a pixel binning, and is in use almost as long as CCD astrocameras. So, nothing revolutionary, just for the first time applied in regular cameras. Astro-CCD cameras are monochrome, so there you can use also 2×2 pixel binning as well as 3×3 pixel binning, whilst with Bayer filters you have to skip 2×2 and go to 3×3 pixels binning (fusion ;)).
    And yes adessio, pentaprism bothers me, too.

  31. Pingback: Canon mirrorless camera in 2012?

  32. Pingback: Canon EOS 1100D

  33. Pingback: Canon EIS ミラーレスシステム!?は半信半疑以下? | YOUのデジタルマニアックス dmaniax.com

  34. Pingback: broken link

  35. Pingback: Canon EIS 60 revealed? « DSLR Photography and Videography

  36. > The first in the series – EIS 60 – is a high end mirrorless camera, not an entry level GF / NEX competitor.
    > It will feature a 22 megapixel sensor. This is a brand new back illuminated design, 18mm x 12mm

    Is this a joke? 18×12 sensor to compete with NEX? LOL :D:D:D

  37. Pingback: Lens Circle

  38. Pingback: Learn how to jailbreak your iPhone 5

Leave A Reply