Canon EOS R1 Specifications [CR2]

[DrD]

A 30 MP resolution on the new R1, if accurate, would equate to each pixel being 5.4 microns.

 

[neuroanatomist]

Pixel size determines the number of photons of light that each pixel on the FF (36 x 24 mm) sensor catches. The Sony A7R5 sensor has 9504 x 6336 pixels, thus each pixel is 3.8 microns (also known as micrometer, 1,000 of which = 1 mm). The Canon R3 sensor is 6000 x 4000 pixels, and each is pixel is 6 microns wide, or 1.58x bigger than the individual pixels on the Sony A7RV -> hence its much better low-light (ISO noise) performance. The Canon R5, Nikon Z9/Z8 all have a 45 MP sensor, Nikon's sensor is 8256 x 5504 and the Canon R5 has 8192 x 5464. The pixel size on the R5 is 4.4 microns, thus each pixel is 16 percent larger than any pixel on the Sony A7R5, whilst the R5 pixels are 73 percent of the size of the R3's pixels (put another way the R3 pixels are 1.36x the R5's). The old Canon 1DX III had 5472 x 3648 or 20 MP resolution where each pixel was 6.6 µm, around 10 percent larger than the current R3. However, the R3 has a new back-illuminated stacked sensor, thus technology improvements have likely more than offset any deficits due to slightly smaller pixels.

The real question is not "How many pixels should the new R1 have?", the correct question is: "What pixel size on a standard 36 mm x 34 mm full frame sensor does a professional photographer require to have the ability to shoot high quality images in different lighting circumstances?"

For reference; the latest iPhone 15 Pro has 1.22 microns pixel size for 48 MP images and 2.44 for 12 MP. Many smartphone camera sensors have 0.9 µm pixels and are hopeless in low-light conditions.

So, before demanding 60 MP or 80 MP (10960 x 7306 with each pixel 3.25 microns) for the new R1, please say what pixel size you are prepared to pay 5x the cost of a flagship smartphone? An 80 MP sensor would have pixels just 1.5 times larger than an iPhone! Creates a problem as if FF MILC camera sensors start to resemble the pixel density of a smartphone, then why buy such a camera (if not for the better low-light capability and shallow DoF, the latter can be simulated now by computational photography)?

A situation similar to Moore's Law for integrated circuit chips, is occurring also for camera sensor resolution. Both computer chips, with smaller and smaller lithography (wafer diameter), and camera sensors with continually increasing resolutions, are hitting diminishing returns.

Personally, I look at pictures…not pixels. The noise from a pixel is inversely proportional to its size. The noise in a picture is inversely proportional to the size of the sensor. In other words, when comparing FF sensors of a similar generation, if you downsample the image output of a 60 MP sensor to 24 MP, the noise will be similar to that from a native 24 MP image, despite the higher pixel-level noise.

Try this – look at images from the Phase One XF IQ4 150MP sensor and the Sony a7R IV. Both have the same 3.76 μm pixels, but at the same ISO the image from the medium format will be less noisy than the FF image. Less noise than image from the big R3 pixels, too.

 

[John Wilde]

1-series bodies have a level of functionality and available customization that the R3 simply lacks. It’s a great camera, but it’s not a 1-series camera. The R1 will be, obviously.

Aren't 1 series cameras also designed to be "super durable", and withstand less than gentle treatment without breaking?

 

[Dragon]

Pixel size determines the number of photons of light that each pixel on the FF (36 x 24 mm) sensor catches. The Sony A7R5 sensor has 9504 x 6336 pixels, thus each pixel is 3.8 microns (also known as micrometer, 1,000 of which = 1 mm). The Canon R3 sensor is 6000 x 4000 pixels, and each is pixel is 6 microns wide, or 1.58x bigger than the individual pixels on the Sony A7RV -> hence its much better low-light (ISO noise) performance. The Canon R5, Nikon Z9/Z8 all have a 45 MP sensor, Nikon's sensor is 8256 x 5504 and the Canon R5 has 8192 x 5464. The pixel size on the R5 is 4.4 microns, thus each pixel is 16 percent larger than any pixel on the Sony A7R5, whilst the R5 pixels are 73 percent of the size of the R3's pixels (put another way the R3 pixels are 1.36x the R5's). The old Canon 1DX III had 5472 x 3648 or 20 MP resolution where each pixel was 6.6 µm, around 10 percent larger than the current R3. However, the R3 has a new back-illuminated stacked sensor, thus technology improvements have likely more than offset any deficits due to slightly smaller pixels.

The real question is not "How many pixels should the new R1 have?", the correct question is: "What pixel size on a standard 36 mm x 34 mm full frame sensor does a professional photographer require to have the ability to shoot high quality images in different lighting circumstances?"

For reference; the latest iPhone 15 Pro has 1.22 microns pixel size for 48 MP images and 2.44 for 12 MP. Many smartphone camera sensors have 0.9 µm pixels and are hopeless in low-light conditions.

So, before demanding 60 MP or 80 MP (10960 x 7306 with each pixel 3.25 microns) for the new R1, please say what pixel size you are prepared to pay 5x the cost of a flagship smartphone? An 80 MP sensor would have pixels just 1.5 times larger than an iPhone! Creates a problem as if FF MILC camera sensors start to resemble the pixel density of a smartphone, then why buy such a camera (if not for the better low-light capability and shallow DoF, the latter can be simulated now by computational photography)?

A situation similar to Moore's Law for integrated circuit chips, is occurring also for camera sensor resolution. Both computer chips, with smaller and smaller lithography (wafer diameter), and camera sensors with continually increasing resolutions, are hitting diminishing returns.

All good information, but note that the light gathering ability of a pixel is a function of its area and therefore a function of the square of its linear dimension, so the effect you refer to is magnified from what you suggest. A simple way of looking at the problem is if you double the number of pixels for a given sensor size, you will lose one stop of sensitivity at the pixel level. This does not mean the the overall image is poorer, but rather that a pixel level crop with the same pixel dimensions will be 1 stop noisier. For the whole image, the eye and particularly clever software will make the image very similar due to averaging effects. In the end, the more important trade-off is for speed and storage space and simply technological feasibility. If the new R1 actually has DGO as suggested, then based on the description of the C70 sensor, every pixel is split again after the split for DPAF, so 30 MP translates to 120 MP that need to be read out, which is double what the competition is doing. At this point, we really don't know how this new sensor is designed, so all analysis is rife with speculation.

 

[David - Sydney]

Pixel size determines the number of photons of light that each pixel on the FF (36 x 24 mm) sensor catches. The Sony A7R5 sensor has 9504 x 6336 pixels, thus each pixel is 3.8 microns (also known as micrometer, 1,000 of which = 1 mm). The Canon R3 sensor is 6000 x 4000 pixels, and each is pixel is 6 microns wide, or 1.58x bigger than the individual pixels on the Sony A7RV -> hence its much better low-light (ISO noise) performance. The Canon R5, Nikon Z9/Z8 all have a 45 MP sensor, Nikon's sensor is 8256 x 5504 and the Canon R5 has 8192 x 5464. The pixel size on the R5 is 4.4 microns, thus each pixel is 16 percent larger than any pixel on the Sony A7R5, whilst the R5 pixels are 73 percent of the size of the R3's pixels (put another way the R3 pixels are 1.36x the R5's). The old Canon 1DX III had 5472 x 3648 or 20 MP resolution where each pixel was 6.6 µm, around 10 percent larger than the current R3. However, the R3 has a new back-illuminated stacked sensor, thus technology improvements have likely more than offset any deficits due to slightly smaller pixels.

The real question is not "How many pixels should the new R1 have?", the correct question is: "What pixel size on a standard 36 mm x 34 mm full frame sensor does a professional photographer require to have the ability to shoot high quality images in different lighting circumstances?"

For reference; the latest iPhone 15 Pro has 1.22 microns pixel size for 48 MP images and 2.44 for 12 MP. Many smartphone camera sensors have 0.9 µm pixels and are hopeless in low-light conditions.

So, before demanding 60 MP or 80 MP (10960 x 7306 with each pixel 3.25 microns) for the new R1, please say what pixel size you are prepared to pay 5x the cost of a flagship smartphone? An 80 MP sensor would have pixels just 1.5 times larger than an iPhone! Creates a problem as if FF MILC camera sensors start to resemble the pixel density of a smartphone, then why buy such a camera (if not for the better low-light capability and shallow DoF, the latter can be simulated now by computational photography)?

The "raw" dimension of full frame sensor divided by the megapixels is not the full story.
I'm definitely not an expert but doesn't the microlens and well depth make a difference?
BSI wil reduce the circuitry on the front of the sensor making the light gathering size different as well.
Dynamic range should improve using BSI vs FSI although Canon did a pretty good job with the R5. Given that Canon already splits the pixel into 2 for DPAF, that already reduces the available area slightly vs a single pixel.
At the end of the day, it isn't a spec sheet mp number but the images produced

A situation similar to Moore's Law for integrated circuit chips, is occurring also for camera sensor resolution. Both computer chips, with smaller and smaller lithography (wafer diameter), and camera sensors with continually increasing resolutions, are hitting diminishing returns.

Not so fast... smaller lithography allows shorter distances/smaller transistors so power consumption goes down. Larger wafer diameter is independent but is more efficient for rectangular chips within a wafer circle. Harder/expensive for optics with bigger diameters as well.
I don't see diminishing returns though. Definitely exponentially higher costs for new fabs but the resulting (eventual) cost benefit does come with it with volume.

 

[koenkooi]

The rumoured specs state "unlimited shots at 40fps with 1:3 CRAW, up to 120fps". Does that mean that when shooting at 120fps and the buffer fills, it will drop down to a consistent 40fps or will it the usual Canon stop-and-start stutter pattern where it will refuse to shoot for seconds-long stretches?

 

[Deleted member 383640]

Dynamic range should improve using BSI vs FSI although Canon did a pretty good job with the R5.

Canon has not only done a pretty good job with the FSI sensor in the R5 but has proven that FSI can outperform the best BSI sensor in terms of dynamic range. For example, the R5 has a better dynamic range (albeit only marginally, but still...) than the A1.

 

[Deleted member 383640]

The R1 could teach the competition a lesson with two things that I think they will have implemented in it:

1. heptic feedback of the electronic shutter (we have already seen the patents),
2. FPS selection based on the force with which the shutter button is pressed (I haven't seen or missed the patents for this, but I think it would be, as people like to say, a "game changer" of epic proportions). This would make the way to access the "turbo bust" FPS mode on the A9III ridiculous.

 

[sanj]

Aren't 1 series cameras also designed to be "super durable", and withstand less than gentle treatment without breaking?

Yesss!!! That is one of the main premium feature.

 

[Deleted member 383640]

The R1, like almost all series 1 cameras before it, will have an absolutely unsurpassed sum of technical, functional, ergonomic and other characteristics that will leave the competition scratching their heads - I have no doubt about that.

Competitors may have a higher number of MP or some similar "advantage" (sometimes an apparent advantage can easily become a disadvantage), but the total sum of real utility values will remain unattainable for them.

As for the story about the R3 being a one-off, I think that in 2-3 years the R3 Mark II will be the first Canon camera equipped with a 24MP sensor with a global shutter, but without the compromises of a semi-finished product like the A9III.

 

[Anon]

Honestly I am happy that we have healthy competition also in this segment of the market. Z9, A1, A9III and now the R1 are all great cameras which push the manufacturers to not sleep on their previous achievements, but to strive for more. Good for the consumer.

PS: And I am also happy to see Nikon back on their feet. Not so long ago there were talks about them exiting the market, but the Z8 / Z9 / Zf have shown they are here to stay. I have used (and still use) the D700 for quite some time, so they have a special place in my heart.

 

[sanj]

I am a little confused after reading the specs. How many fps will the camera shoot in full frame and what will be the buffer? Or this information is yet not known? And no news on the second card slot yet, right?

 

[sanj]

The R1 could teach the competition a lesson with two things that I think they will have implemented in it:

1. heptic feedback of the electronic shutter (we have already seen the patents),
2. FPS selection based on the force with which the shutter button is pressed (I haven't seen or missed the patents for this, but I think it would be, as people like to say, a "game changer" of epic proportions). This would make the way to access the "turbo bust" FPS mode on the A9III ridiculous.

This would be great!

 

[sanj]

The rumoured specs state "unlimited shots at 40fps with 1:3 CRAW, up to 120fps". Does that mean that when shooting at 120fps and the buffer fills, it will drop down to a consistent 40fps or will it the usual Canon stop-and-start stutter pattern where it will refuse to shoot for seconds-long stretches?

I am confused too

 

[sanj]

A 30 MP resolution on the new R1, if accurate, would equate to each pixel being 5.4 microns.

Please explain the significance of this post. Please. I want to learn. Thx.

 

[koenkooi]

I am a little confused after reading the specs. How many fps will the camera shoot in full frame and what will be the buffer? Or this information is yet not known? And no news on the second card slot yet, right?

The way I’m reading these rumoured specs is that you can do 120fps with AF and AE, but your buffer will fill quickly.
The R1 might also get a burst mode with AF and AE locked, like the 195fps mode in the R3.

 

[sanj]

Three is overlap between AI and ML (think Venn diagram), both are based upon algorithms. Deep learning is part of AI but is also a subfield of ML. AI seeks to use fuzzy logic to build logic-based expert systems and can involve Bayesian networks (relying on anterior or assumed probabilities, later revised with more data), whereas ML uses clustering, neural networks, various linear, nonlinear and symbolic approaches to learning. I have been teaching PhD students this stuff for the last 7 years.

From a photography perspective AI algorithms and chips like those used by some Sony cameras, use deep learning techniques to develop the algorithms -> essentially a training exercise. They train the logic system to recognise objects, for instance; birds, vehicles, planes, humans, other animals etc. The more examples the AI algorithm is exposed to, the better it gets at recognising the differences between say a racoon and a honey badger (or could be between different types of birds). You can also train the system to select specific objects against a particular background -> think white bird against a pale sky, or dark sky, or dark bird against a pale/dark sky, not just for focus tracking but for correct exposure also.

These computational models that are developed can become quite large, hence the move to dedicated chips to run these algorithms, alongside the main camera processor. It makes sense to hive them off to a dedicated chip. Whether the R1 will have such a separate chip is not yet known, perhaps it will not need one if the new Digic X2S (whatever it will be called) is powerful enough to run these algorithms whilst doing all the other things it needs to do. My R5 has some AI capabilities - you set the AF method to Servo/People/Eye-detect and you are essentially telling the camera the anterior probability (look for a person or group of persons), once the camera focuses on human(s) it then knows that they have faces at the top of their bodies, what the ratios are between chin/mouth/nose/eyes etc and can select the eyes (usually the nearest to the lens). But what if a person closes their eyes? This would be an illustration where more sophisticated AI would know if the eye was opened or closed, or if they wore spectacles covering the eye, or if those spectacles were actually sunglasses, and so on.

So, AI is real in photography, and not just a marketing term to hype a new model. Do we want more and improved AI in our cameras? Yes!

Thank you!

 

[jd7]

Answer one:
You get what you pay for: NOTHING!

Answer two:
I can live with it!
If you can't, go buy something more "professional" and useful for you!
Do Nikon, Sony, etc. offer anything better for free?

Sorry, but DPP is not free. It is bundled with a camera purchase. Saying DPP is free is like saying you are paying for a sensor and the rest of the camera is free. If DPP was not included, potentially a buyer would have no way of processing images. Obviously Canon could sell a camera without including that functionality, and for many of us who already have our preferred software for that task we may not care, but it is not difficult to understand why Canon wants to supply a package which provides at least some ability to process images.

 

[Maximilian]

Sorry, but DPP is not free. It is bundled with a camera purchase.

Fully accepted, So somehow you sponsor the development of DPP with what you pay for camera or lens.
I wanted to be a little bit provocative.

and for many of us who already have our preferred software for that task we may not care, but it is not difficult to understand why Canon wants to supply a package which provides at least some ability to process images.

My point is, that you pay a lot for that SW, developed by independent companies.
And then people complain that DPP is not delivering the same functionality, features or performance.
It is just a basic image editing program (almost) for free.
If one wants more one has to pay extra.

On the other hand I repeat my question:
Do Nikon, Sony, etc. offer anything better (for free)?

 

[Anon]

I think Sony included a simplified version of Capture One with their cameras (at least some of them). While C1 for sure beats the Nikon and Canon software it seems that C1 has discontinued the Express version of their software when restructuring their licensing model and it will stop working end of this month. I wonder if a new agreement will be found.

IMO that is a good way for both camera and software manufacturers: include a simpler version of a good commercial product. The camera manufacturer can leave the software development to the pros and these guys in turn might get a few users upgrading to the "better" paid version.

This is common practice for example for DJ equipment: DJ Midi controller are shipped with a simple version of Traktor Pro or Serato or something similar. If one needs more features than included in the base version, these can be unlocked for a fee.