Canon is gearing up to finally release a high megapixel camera with 100+ megapixels [CR3]

[usern4cr]

It appears I'm waiting for the next R3 with more MP but not 100MP. The R3 was introduced November 27, 2021. Does anyone guess there will be an R3-like introduction (I do much more pictures than videos) with, say, 60 - 80 MP before end of this year? That would be one year from R3. Or does anyone guess that, say, R3II comes every two years (or more?)

I'd be very surprised to see a R3-II in 2 years (R3 is barely out now in quantity). I would expect the R1 to hit within that time and it better be a big hullabalu for Canon and most of those here!

 

[unfocused]

I would expect the R1 to hit within that time and it better be a big hullabalu for Canon and most of those here!

As usual there will be a of whining when the R1 finally hits. People have poured so many unrealistic conflicting expectations into that body that it is bound to disappoint. If folks can’t find what they want in the existing lineup they aren’t trying.

 

[usern4cr]

As usual there will be a of whining when the R1 finally hits. People have poured so many unrealistic conflicting expectations into that body that it is bound disappoint. If folks can’t find what they want in the existing lineup they aren’t trying.

I expect the R1 to make a lot of people (that can afford it) very happy. My personal guess is that it will have a fast QP BSI sensor with a reasonable MP count (~45?). It will have other bells and whistles, but I won't guess any further there. But I'm in the "smaller body" camp so I'd still prefer a R5 II to come out.

 

[JohnC]

I'd be very surprised to see a R3-II in 2 years (R3 is barely out now in quantity). I would expect the R1 to hit within that time and it better be a big hullabalu for Canon and most of those here!

Speaking of the R3, it only took me 2 weeks to get one from Canon direct (delivered 2 days ago)

 

[usern4cr]

Speaking of the R3, it only took me 2 weeks to get one from Canon direct (delivered 2 days ago)

That's good to hear that Canon has (hopefully) got enough availability now. Good for them, and for us.

 

[Michael Clark]

In the past, with FSI sensors, as your #pixels get too high you end up with all your sensor being non-pixel sensing support circuitry.
But new BSI has the potential that the full sensor well depth is always there, no matter how tiny the pixels being sensed. So I could see BSI technology allowing much higher resolutions while maintaining a similar IQ, which is a win in potential resolution at the cost of only much higher data storage requirements.

This almost begs the question of: If they introduce the same R5 successor in a 50-60MP version and a 100+MP version at the same price, which would you want? Most would opt for 100+ just assuming it's the best choice. I might opt for the 50-60MP sensor at the moment taking storage costs into account, but would probably wait to see some test results before knowing which was better for my use cases.

But either way, I'm eagerly waiting for the R5 style successor!

Canon has been using gapless micro-lenses above the circuitry for at least a decade or so. Almost all of that light previously "lost" to circuitry is now funneled into the well. BSI has advantages, as do stacked layer sensors, but the loss of area to circuitry is nowhere near the issue it was 15-20 years ago.

As to resolution: I'd get an R5 tomorrow if I could justify the cost/benefit in terms of revenue stream. But I wouldn't necessarily be thrilled with even the 45MP it offers. I do want more than the 20 MP offered by the R6, though, and there isn't much choice in between at the moment in the RF line. So I'm still shooting with a 30MP 5D Mark IV for my FF body.

I'd love a 32MP APS-C body because I would use it for a different use case than the FF bodies and would rather have a speedy 32MP APS-C sensor than an 82 MP FF camera with the much larger files sizes, not to mention cost. Unfortunately, with no vertical grip available for it, I doubt I'll be getting an R7. My old shoulder makes it difficult for me to shoot in portrait orientation without vertical controls.

I have absolutely no need for a 100 MP+ camera. Most of my work these days is displayed digitally, either web based or on relatively low resolution monitors (2K is only 3.7 MP, 4K is only 8.8 MP). I can't remember the last time I've had anything printed larger than 24x16. 30MP is fine for that.

 

[Michael Clark]

Most who are active on this site might, but taking the wider and majority usage I’m not so sure. I found it interesting that each year in Lens Rentals “Top Twenty most rented gear of the year” list the “high mp” version of a manufacturer’s model line (ie Sony A7R vs A7 for instance) never makes the list. Not once.

Rental patterns don't necessarily reflect overall sales patterns, though. For some things, different types of users rent vs. buy. For others, people buy what they plan to use frequently and only rent what they will occasionally use.

 

[Michael Clark]

I agree and it does not even need that many resources to keep it up to date. For example just get the R10 insides and put it in a slightly smaller M mount body and you have an up to date M camera. They did the same with the 90D/M6II. The insides are identical 90%.

For shooting video, maybe. But for shooting stills with an eye level viewfinder they are significantly different. Optical VF vs. detachable electronic VF that ties up the hot show when no PC terminal is provided. Dedicated PDAF array vs. on sensor hybrid DPAF.

 

[Michael Clark]

The processing time is not as relevant as it is designed for full sensor reading. Rolling shutter is the readout speed and the faster the portion of the sensor read then - in theory - the reduction in rolling shutter.
The ultimate goal/extension to this theory is a global shutter. The sensor information can be stored temporarily waiting for the processor to process it. The development of stacked sensors is to reduce the readout speed and reduce noise by having the amplifiers as close as possible to the sensor

We're talking about AF speed, not output image quality.

In terms of how many scan cycles per second the camera can do AF, it's all about processing time which is directly impacted by data volume. Even if one had a global shutter, which no camera like the R3, R5, α1, α9, or z1 yet has, it would still be about how long it takes the camera to process that information and move the AF system in response to that data.

The camera can't store that data to be processed later and use it to effectively control AF until it has processed that data in the way it needs to in order to use that data for AF purposes.

No one cares if rolling shutter influences AF speed, because the algorithms can compensate for the distorted shapes. People only care if rolling shutter affects the captured image.

No matter how fast a sensor can be read out, it will always take less time to process the data from only 2/3 of the lines that it will to process all of the lines when talking about using that data to control AF response.

 

[Michael Clark]

I've gotten to the point where I really prefer purpose-built bodies. I have the R3 and the R5 and choose which to use based on what I need/want at the time.

^^^^^^T.H.I.S^^^^^^

 

[Michael Clark]

Learn some basic maths and physics before making such comments. Read this for a start: http://www.normankoren.com/Tutorials/MTF.html Or if you find that heavy going try Roger Ciccala https://www.lensrentals.com/blog/2017/07/experiments-for-ultra-high-resolution-camera-sensors/

"So What Did We Learn Today?​

Well, mostly nothing, but there are a couple of things.

First, let me emphasize again that if we had a 150-megapixel camera and shot today’s lenses on it, the images would have more detail than that same lens on your current 36-megapixel camera."

But the $10,000 question is will you get 5X, or even 2-3X as many lp/mm by spending 2-3X as much?

As one approaches the absolute resolution limits of the lens (the so-called diffraction cutoff frequency), the gain from significantly increased sensor resolution becomes more and more miniscule.

 

[AlanF]

But the $10,000 question is will you get 5X, or even 2-3X as many lp/mm by spending 2-3X as much?

As one approaches the absolute resolution limits of the lens (the so-called diffraction cutoff frequency), the gain from significantly increased sensor resolution becomes more and more miniscule.

Few would disagree that money spent on photographic equipment follows the law of diminishing returns. But, sometimes the miniscule increase is just enough to make it worthwhile (like that extra degree of waterproofing, for example).

 

[Rocky]

There are real world limitations.

A silicon wafer and the masks used to make the chip have a high price, which has to be divided over units sold. The larger the sensor, the fewer sensors can be made out of it. Then, the larger the sensor, the fewer the cameras sold. This means higher price per unit.

Wafers aren't clean. There might be dozens of defects (in the silicon itself, dust particles, etc) on it. The larger the chip, the higher the chances there would be a defect in its area, so the number of chips that can be made out of a wafer drops exponentially with its size.

Which is why the Hasselblad H6D-400C's sensor costs $26,000.

450mm wafers are on the horizon, which might lower sensor prices. Then again, the market is shrinking. I would be surprised if digital 645 would become viable option in the next 25 years.

The manufacturing cost alone will not be $26,000. However if you add the mask cost ( at least 20 for the wafer and a few more for the micro lenses) for a limited amount of wafer(due to very low volume production) and the testing cost of individual sensor. Then the $26,000 may be reasonable.

 

[angelisland]

People salivating for more REZ here in the forums.

I am in the commercial (advertising) photography world, shooting and producing for companies like L'Oreal, LVMH, J Walter Thompson, Apple, Google, Amazon, Williams-Sonoma, Sephora, Martha Stewart…etc etc.
No one, and I mean NO ONE is asking for more pixels.

Just my two cents, from commercial experience.

Cheers.

 

[usern4cr]

Canon has been using gapless micro-lenses above the circuitry for at least a decade or so. Almost all of that light previously "lost" to circuitry is now funneled into the well. BSI has advantages, as do stacked layer sensors, but the loss of area to circuitry is nowhere near the issue it was 15-20 years ago.

As to resolution: I'd get an R5 tomorrow if I could justify the cost/benefit in terms of revenue stream. But I wouldn't necessarily be thrilled with even the 45MP it offers. I do want more than the 20 MP offered by the R6, though, and there isn't much choice in between at the moment in the RF line. So I'm still shooting with a 30MP 5D Mark IV for my FF body.

I'd love a 32MP APS-C body because I would use it for a different use case than the FF bodies and would rather have a speedy 32MP APS-C sensor than an 82 MP FF camera with the much larger files sizes, not to mention cost. Unfortunately, with no vertical grip available for it, I doubt I'll be getting an R7. My old shoulder makes it difficult for me to shoot in portrait orientation without vertical controls.

I have absolutely no need for a 100 MP+ camera. Most of my work these days is displayed digitally, either web based or on relatively low resolution monitors (2K is only 3.7 MP, 4K is only 8.8 MP). I can't remember the last time I've had anything printed larger than 24x16. 30MP is fine for that.

I understand your needs about a vertical grip based on your shoulder. With that in mind, do you have a R3? If not, wouldn't that be a good compromise in a gripped body for you right now?

As far as image MP, I sometime make panoramas of lots of images and print them (myself) up to 8' on a side at 600 DPI (and yes, I drop the DPI lower when I have to). More frequently I just print a single R5 image at large scale and the (usually cropped to taste) 45MP file is nice, with the sharpness & big background blur of the RF 100-500L really appreciated. So I do use high MP's. Would a 100+ MP sensor really give me better results? I'm guessing that it would (based on what I've heard from AlanF), but I wouldn't look forward to more TB storage for it. But I'd make it happen somehow if I got it on a non-gripped body (since I'm OK doing the 90 degree flip).

 

[Michael Clark]

I understand your needs about a vertical grip based on your shoulder. With that in mind, do you have a R3? If not, wouldn't that be a good compromise in a gripped body for you right now?

As far as image MP, I sometime make panoramas of lots of images and print them (myself) up to 8' on a side at 600 DPI (and yes, I drop the DPI lower when I have to). More frequently I just print a single R5 image at large scale and the (usually cropped to taste) 45MP file is nice, with the sharpness & big background blur of the RF 100-500L really appreciated. So I do use high MP's. Would a 100+ MP sensor really give me better results? I'm guessing that it would (based on what I've heard from AlanF), but I wouldn't look forward to more TB storage for it. But I'd make it happen somehow if I got it on a non-gripped body (since I'm OK doing the 90 degree flip).

If I can't justify the cost/return on investment of the $3,800, 45MP R5, what do you think the numbers for the $6,000, 20 MP R3 look like?

Instead of using a 120-300mm/2.8 with either the 7D Mark II (same pixel density as a 50MP FF camera) or 30 MP 5D Mark IV to shoot sports under lights at night, I'd need to spend what a 400/2.8 costs to get roughly the same number of pixels on subject for field sports at night.

Even comparing a 32MP APS-C camera + 70-200/2.8, like the upcoming R7, to get close to the same number of pixels on subjects almost 100 yards away at the other end of a football/soccer field or from behind home plate to the warning track (or vice versa) for baseball with a 24MP FF camera I'd need a 600/2.8 (which is not available at any price...)

R7 + RF70-200/2.8 = $4,300.
R3 + RF400/2.8 = $18,000 plus the need for another body and shorter zoom lens for when the action is too close for 400mm

There's no way a $6,000 camera and a $12,000 lens makes sense for anyone shooting local youth and high school sports and needing to be, at the very worst, revenue neutral.

Stitching multiple images doesn't really work that well for sports/action.

 

[Sporgon]

Rental patterns don't necessarily reflect overall sales patterns, though. For some things, different types of users rent vs. buy. For others, people buy what they plan to use frequently and only rent what they will occasionally use.

Can’t disagree with your view on rental patterns, however, those who own the high mp versions then need to rent the “normal” mid level mp model ? That would seem odd to me. Personally I don’t think the very high mp version of a model range is anything like as popular as many of those who frequent the likes of CR would like to think.
As we inevitably go higher and higher, the R5 is after all very high res / output, the “high mp” model may become even more niche and if so I’d expect it to be sold at quite a premium.

 

[Sporgon]

Well, there's always post production....I mean, it isn't like anyone doing portraits is selling them to clients straight out of the camera, you know?


cayenne

I have yet to come across a model over the age of about 21 who wants ultra sharp, ultra high res images of themselves !! And don’t get me on to weddings of middle aged couples !

 

[Sporgon]

And too, the really FUN thing about mirrorless cameras, is that you can adapt pretty much ANY older manual lens to them...and there is some GREAT vintage glass out there that is perfect for portraiture and may soften some images if that's what you wish.

cayenne

Regarding mirrorless being suitable for adapting old glass, I agree with you 100% . I’ve created some interesting images using my collection of 1960’s era Takumar lenses adapted to an RP. In fact the user interface in terms of focus peaking and allowing practical magnification for pin point focus accuracy very quickly is actually making these lenses viable in a live shoot I think. I don’t want to bore people here on a 100mp thread with a picture, but as I know you like these old lenses I’ll pm you with one taken on a 1964 Takumar 105/2.8, which is a rather beautiful lens.

 

[usern4cr]

If I can't justify the cost/return on investment of the $3,800, 45MP R5, what do you think the numbers for the $6,000, 20 MP R3 look like?

Instead of using a 120-300mm/2.8 with either the 7D Mark II (same pixel density as a 50MP FF camera) or 30 MP 5D Mark IV to shoot sports under lights at night, I'd need to spend what a 400/2.8 costs to get roughly the same number of pixels on subject for field sports at night.

Even comparing a 32MP APS-C camera + 70-200/2.8, like the upcoming R7, to get close to the same number of pixels on subjects almost 100 yards away at the other end of a football/soccer field or from behind home plate to the warning track (or vice versa) for baseball with a 24MP FF camera I'd need a 600/2.8 (which is not available at any price...)

R7 + RF70-200/2.8 = $4,300.
R3 + RF400/2.8 = $18,000 plus the need for another body and shorter zoom lens for when the action is too close for 400mm

There's no way a $6,000 camera and a $12,000 lens makes sense for anyone shooting local youth and high school sports and needing to be, at the very worst, revenue neutral.

Stitching multiple images doesn't really work that well for sports/action.

That makes a lot of sense. If the subject you're most interested in is distant enough to be too small on the sensor, than a more affordable & compact APS sensor with more pixel density would indeed give you more pixels on the subject, and probably would give you better results. I guess you need something like an R7 in a new more compact (& affordable) gripped body?