Canon Testing a 75+ Megapixel EOS-1 Body? [CR1]

[Loswr]

Shouldn't Canon be creating something greater than 23 megapixels before jumping with the big boys?

Yeah, because they didn't already create (your word) a 120 MP APS-H sensor... :

BTW, who are the 'big boys'? There's only one for FF...and it's not Nikon.

 

[RGF]

Could be a foveon type sensor - could each photo sensor thrice, once for each color.

Just another crazy idea based upon no info but typical of how the industry plays games with marketing claims.

 

[Mt Spokane Photography]

ummm... the way i see it... i should buy some of western digital, seagate, etc stocks as if this rumor comes true LOL

You'd be better off to invest in Lexar or Sandisk. There is no profit in the old rotating disk technology, and while Samsung makes great SSD's (so does Micron), its just a miniscule part of Samsung's total business, tripling their profit on CF and SD cards would not even show on their bottom line.

 

[Lee Jay]

Shoot mRAW or sRAW, problem solved.

mraw and sraw are very poor implementations. The DNG spec with its lossy-compressed raw mode is much better. It's still demosaiced like mraw and sraw but it's in object space (like full raw and unlike mraw and sraw), it's linear (like full-raw) and it can be any resolution up to and including full resolution, yet reduced in size by a factor of 3 or more even at the same resolution. Reduce the resolution by a factor of 2 in each direction and compress this way and the file sizes are reduced by more than a factor of 10.

 

[Lee Jay]

Ugh, with a RAW file size of what? 120MB?

So? Storage is cheap, and my computers are fast.

Storage isn't cheap. It's cheaper than it's ever been, but it's not cheap.

Sure, you can get a 3TB hard drive for $120. Let's call it 4 cents a gig.

But that's not the whole story.

Many of us have recycled our last desktop computers ever. All my desktops, at work and at home, are now laptops.

To get some storage capacity, I went for larger, more expensive machines with two internal hard drive slots. Each hard drive costs about twice as much - 8 cents a gig.

For backup, I have two machines. So we're up to 16 cents a gig.

I also keep two external backups on large external drives, one at home, one at work. So we're up to 24 cents a gig.

I shot 700 shots on Sunday (yesterday). At around 7 images a gig, this would mean I'd have used 100GB yesterday, or 0.24*100GB = $24. That's more than I would typically spend on film in one day when I was shooting that way (typically two rolls a day at $10 each including processing).

So, while we may get a lot more than in the film days, I wouldn't call storage cheap in the context of digital shooting styles and when you include modern computers, storage, and backups.

Now, shooting JPEG L/M on my 20D is cheap, at about 1000 shots per gigabyte (yesterday cost me 17 cents). Raw on a 75MP camera? Nope, not cheap at all.

 

[photonius]

Don't forget the 70D. This may be a 37.5 Mp camera with dual pixels. Use different exposures on each half-site, and you get expanded DR 14bit or 16 (like the ML trick).

Canon could also bin the pixels normally, but for tele, if cropping is desired, the unbined version could be selected (sort of like the Nokia purview).

Another alternative is lens correction. Due to the high sampling, distortion, CA, etc. can be corrected first with little loss before downsampling for storage.

Lots of things that can be done.

You would never really be able to "unbin", as the pixel halves are each under a single microlens and color filter. There wouldn't really be any point, since you would have two halves of gree, two halves of red, two halves of blue. That would create a real oddity for digital interpolation, assuming you could get any benefit at all.

No, the latter two examples would refer to a full 75Mp sensor, not a 70D type sensor, of course you can't unbin that.
The idea is for normal use you "bin" your 75mp sensor to 35, or 18mp, for file storage. If you want ultimate reach for tele, you keep 75mp (and can crop), or canon even provides a crop mode (like nikon does for aps-c lenses).

By my calculations, a 75mp FF sensor would be 10600x7050 pixels in size, with 3.4 micron pixels. That is actually not all that bad. That is similar to a 24mp APS-C sensor in size (which is very interesting...would make sense if Canon has already produced a prototype 24mp 7D II sensor.)

I really don't see how Canon could keep using a 500nm FSI sensor design with 3.4 micron pixels. Given they have a patent for a BSI design for APS-C and FF, I wonder if these two sensors are using the same architecture.

yes, I agree, it seems unlikely a 500nm FSI design.

 

[c.d.embrey]

NO INTEREST!! Canon seems to be doing their very best to make me switch camera companies But that's OK, cameras are just tools, and I always use the best tool for my paying jobs. When I need 75+ megapixels for a job, I'd prefer to shoot MFD

 

[fisico]

Canon needs to improve the signal-to-noise ratio of their sensors dramatically before packing 75 MP onto one. With everything except AF and shooting speed, the 5D Mk III is merely a minimal improvement over the 5D Mk II. Across the ISO spectrum, there is still lots of shadow noise to get rid of. So my hope is that they overhaul everything else before adding pixels.

 

[LetTheRightLensIn]

Ugh, with a RAW file size of what? 120MB?

So? Storage is cheap, and my computers are fast.

Yeah, but it will take 10 hours to backup 10 photos to another HD.
And maybe 10 months to back up a a trip.

Shoot mRAW or sRAW, problem solved.

Nah, the solution is to simply suffer and do those day long back-ups of each drive.

 

[LetTheRightLensIn]

It would make sense to be a 39Mpixel body for a 4x4 pixel binning. Even a 5x5 binning corresponds to a ~61Mpixel body!

Perhaps the 5D4 can now be 39MP with 6fps and 4x4 binned 1080p and 2x2 binned 4k while this 75-80MP monster is the 1DXs specialty $$$ camera.

 

[LetTheRightLensIn]

Don't forget the 70D. This may be a 37.5 Mp camera with dual pixels. Use different exposures on each half-site, and you get expanded DR 14bit or 16 (like the ML trick).

Canon could also bin the pixels normally, but for tele, if cropping is desired, the unbined version could be selected (sort of like the Nokia purview).

Another alternative is lens correction. Due to the high sampling, distortion, CA, etc. can be corrected first with little loss before downsampling for storage.

Lots of things that can be done.

You would never really be able to "unbin", as the pixel halves are each under a single microlens and color filter. There wouldn't really be any point, since you would have two halves of gree, two halves of red, two halves of blue. That would create a real oddity for digital interpolation, assuming you could get any benefit at all.

The term MegaPixel usually refers to output image pixels, not photodiode count. Keep in mind, there are usually more real "pixels" in a sensor than can be counted from the output image anyways, and have been for some time. For example, an 18mp sensor usually has nearly 20mp actual pixels. It just doesn't seem logical for Canon to start counting their half pixels used for AF...



By my calculations, a 75mp FF sensor would be 10600x7050 pixels in size, with 3.4 micron pixels. That is actually not all that bad. That is similar to a 24mp APS-C sensor in size (which is very interesting...would make sense if Canon has already produced a prototype 24mp 7D II sensor.)

I really don't see how Canon could keep using a 500nm FSI sensor design with 3.4 micron pixels. Given they have a patent for a BSI design for APS-C and FF, I wonder if these two sensors are using the same architecture.

The sensor in Sony RX100 II is claimed to have become 40% more efficient going from FSI to BSI (although DxO seems to measure the boost a bit less so) and that is 20MP on a sensor as smaller again as APS-C vs FF which makes me think that even 24MP APS-C or FF equivalent would still only get a relatively modest boost from BSI, maybe just enough to barely notice real world at a noticeable cost increase, although a 75MP APS-C sure could get a big boost. Maybe it's just enough to start making it worthwhile. Nobody has made one bigger than the 1" sensor in the RX100 II yet though and I bet a FF BSI would be very pricey indeed (of course this entire camera sounds like a 1DXs sort of beast which tend to be insanely $$ so they might have the margin to do it).

 

[LetTheRightLensIn]

Canon needs to improve the signal-to-noise ratio of their sensors dramatically before packing 75 MP onto one. With everything except AF and shooting speed, the 5D Mk III is merely a minimal improvement over the 5D Mk II. Across the ISO spectrum, there is still lots of shadow noise to get rid of. So my hope is that they overhaul everything else before adding pixels.

Yeah at this point I really hope they conquer shadow noise at lower ISOs more than anything.

 

[poias]

Shouldn't Canon be creating something greater than 23 megapixels before jumping with the big boys?

Yeah, because they didn't already create (your word) a 120 MP APS-H sensor... :

BTW, who are the 'big boys'? There's only one for FF...and it's not Nikon.

It was great using the magical 120 MP APS-H sensor! What not to like, whether the awesomeness of 120 MP or the un-dead nature of the APS-H. Talk about relevancy!

Oh, and the big boys are all those higher mpx, high DR, superior sensors found in all modern DSLRs except for Canon.

 

[Loswr]

Oh, and the big boys are all those higher mpx, high DR, superior sensors found in all modern DSLRs except for Canon.

Meaning....Sony.

You and temple guy should get together and throw yourselves a party. But please...somewhere else, not here.

 

[bvukich]

Shouldn't Canon be creating something greater than 23 megapixels before jumping with the big boys?

Yeah, because they didn't already create (your word) a 120 MP APS-H sensor... :

BTW, who are the 'big boys'? There's only one for FF...and it's not Nikon.

It was great using the magical 120 MP APS-H sensor! What not to like, whether the awesomeness of 120 MP or the un-dead nature of the APS-H. Talk about relevancy!

Oh, and the big boys are all those higher mpx, high DR, superior sensors found in all modern DSLRs except for Canon.

Good sensors yes, and yet they consistently sell less than Canon...

If Nikon could actually build a decent body, with decent ergonomics and UI to put that sensor in, they could become a threat to Canon's market position. Although they'd need to work on their lenses too, the only shining point in that lineup is the 14-24/2.8, admittedly great, but beyond that they don't have anything notable.

 

[RLPhoto]

Shouldn't Canon be creating something greater than 23 megapixels before jumping with the big boys?

Yeah, because they didn't already create (your word) a 120 MP APS-H sensor... :

BTW, who are the 'big boys'? There's only one for FF...and it's not Nikon.

It was great using the magical 120 MP APS-H sensor! What not to like, whether the awesomeness of 120 MP or the un-dead nature of the APS-H. Talk about relevancy!

Oh, and the big boys are all those higher mpx, high DR, superior sensors found in all modern DSLRs except for Canon.

Good sensors yes, and yet they consistently sell less than Canon...

If Nikon could actually build a decent body, with decent ergonomics and UI to put that sensor in, they could become a threat to Canon's market position. Although they'd need to work on their lenses too, the only shining point in that lineup is the 14-24/2.8, admittedly great, but beyond that they don't have anything notable.

And nikon lenses mount funny. ;D

 

[Pi]

Shoot mRAW or sRAW, problem solved.

mraw and sraw are very poor implementations. The DNG spec with its lossy-compressed raw mode is much better. It's still demosaiced like mraw and sraw but it's in object space (like full raw and unlike mraw and sraw), it's linear (like full-raw) and it can be any resolution up to and including full resolution, yet reduced in size by a factor of 3 or more even at the same resolution. Reduce the resolution by a factor of 2 in each direction and compress this way and the file sizes are reduced by more than a factor of 10.

We are talking about mRAW and sRAW for sensors with very high pixel count. Whatever you do not like about the current implementation, will became a non-problem with higher pixel counts. For people, that definitely do not want to see 75mp or so anywhere in their workflow, this would be the solution.

 

[Pi]

Shoot mRAW or sRAW, problem solved.

DxO will not process mRAW & sRAW files. According to their support, it's because those formats miss some information the software requires.

Do not use DXO then.

BTW, if they want to survive, they would have to support mRAW and sRAW.

A. You're missing my point.

B. If someone gifted me a 75MP camera tomorrow morning, I would paint it blue and throw it into the sea.

And you are missing mine. If nobody told you that the actual sensor is 75mp, and the "RAW" files are 22mp, with the small "m" skipped, how would you ever know what the actual sensor resolution was?

 

[gunship01]

Would love to have a camera like this as I am sure it would recoil with every pull of the button.

That's taking pictures alright!

Just need to watch the lens (barrel) does not overheat. I'd say 4 FPS should do it.

Now make that price point around 5K and I'm in!!

 

[poias]

Good sensors yes, and yet they consistently sell less than Canon...

If sales figure is a mark of technical prowess then a Civics and Corollas are the best cars in the world.