Canon EOS R body with more than 75mp on the horizon [CR2]

[JonSnow]

Observation trumps theory. The available evidence shows large changes in DR with changes in ADC architecture (off chip vs. on chip) yet the highest DR sensors available today have relatively small pixels (D8x0 series; A7r3).

you compare different sensor technologys.
when you say something like this you better compare exactly the same sensor technology (both BSI, both the same ADC etc.) but with different photosite sizes.

i would place my bet on a sony MF sensor (42MP) with the same technology as the A7 R III, but bigger photosites, would show better DR than the A7 R III.

not that i am saying reducing the influence from the ADC has not a great impact.
but imo you can´t ignore that per pixel quality will suffer with smaller pixels.

how important that is for a "normalized " print size is a different question.

 

[4fun]

Sounds intriguing. The entry level model.... hmm that would be more my speed but what could they cut from the R?

easy! My 3 "nerfing suggestions" for entry level EOS R ... in this order:

1. No video and audio recording whatsoever
2. no M.-fn slider
3. more compact body without Top-LCD

 

[Architect1776]

So, how would they name it?

EOS RDs? RDs R? Rs? RsR? RR?

EF lenses support 250 mp sensors Canon makes so this should be pretty easy for new R lenses to support.

 

[sdz]

umm how bout 35-45MP Pro EOS R ALL AROUND MODEL with DUAL DIGICS something for us DO-ALL Shooters of fashion,portraits and sports 7-9fps with continuous AF and better full frame 4K video and AF and DUAL CARD SLOTS nobody cares about 75MP

Some care about 70 MP. And, I'm sure Canon will provide R mount cameras for them.

{snip}

Another niche camera for product photography and formal, static portrait work would certainly be useful to some, but wouldn't a Canon D850 buster, in either a dSLR or mirrorless body, be appealing to a larger number of photographers?

{snip}

By releasing a 70 MP R mount camera, Canon merely filled one slot in its expanding product portfolio. By doing so, Canon can, if it wishes and has the technology, develop cameras that exceed the specifications of Nikon's 850 camera. The former does not make the latter impossible.

 

[Act444]

I think you're correct. I was also hoping they would leave the sensor at 50 mp and work on cleaning up noise and adding a little dynamic range. At 50 mp Canon would still be leading in resolution for full frame cameras.

Basically sums up what I was going to say. Although I still prefer the colors out of the 5DsR to the newer 5D4 or R, which seem to have a flatter output and err more to the green end of the spectrum (most noticeable in portrait shots).

In many ways I think the 5DsR is the best stills camera Canon has ever made - and the only Canon camera to date with the AA filter cancelled to allow for maximum detail capture. At low ISO (800 and below) IQ blows me away. It’s purely a photographer’s camera and I love that about it.

That said, the one (other) area they could really improve the camera is vibration reduction. Despite the manufacturing precautions taken, vibrations and blurring can still be still a big issue with the 5DSR. If they decide to take its replacement ML, this is the body to debut IBIS in, no question about it. Also, if they can resolve the IQ issues with using electronic shutter curtain (silent or quiet shooting modes), that would help immensely.

 

[privatebydesign]

For all those thinking 75mp is way too many for a 135 format sensor I'd remind you all that an iPhone has an approximately 40mm² 12MP sensor, a 135 format 'FF' sensor is around 860mm². To get the pixel size and density we are currently using very happily with the most used cameras in the world that FF sensor would need to be in the 250MP range.

Now I know there are caveats with lenses, low light etc etc, but I am regularly blown away by the output from phones nowadays and most of those caveats are overcome with a much bigger sensor. Remember, again, noise is not a function of pixel size but sensor area.

I'd take a 75MP R over a 50MP 5DSR any day not least of which because I can use an adapter with my EF lenses and filter them, at this point filter solutions for my 15mm fisheye, TS-E 17 and 11-24 are not good or cheap or small, and the extra pixel density wouldn't faze me at all.

 

[stochasticmotions]

Looking forward to seeing this as my 5DS will be about 5 years old by that point. Also with the high resolution canon being at 75 MPixels maybe the faster 5DIV equivalent will be in the 45 MPixel area much like the D850. That would make for 3 pretty compelling mirrorless options on the canon side of the fence. Add in a 24-30MPixel that can do 20 frames per second for the olympics and we have the full spectrum. Really in all these cases I think the only thing holding canon back is the processing power of their chips....ok maybe getting 20 fps off the sensor quickly enough to be close to global shutter equivalent is another thing canon will need to solve but I'm really looking forward to what is going to come out in the next year or two.

 

[Jack Douglas]

Amazing how juicy speculation is - 6 pages before I even got a chance to view it! I'm predicting an implant behind the eyes and a digic brain replacement, offered by Canon at their facility. That's assuming the rich and powerful haven't turned us all into robots to bow and serve and work for nothing, sooner.

Jack

 

[ethanz]

Amazing how juicy speculation is - 6 pages before I even got a chance to view it! I'm predicting an implant behind the eyes and a digic brain replacement, offered by Canon at their facility. That's assuming the rich and powerful haven't turned us all into robots to bow and serve and work for nothing, sooner.

Jack

You are getting kind of far fetched there, Jack!

 

[Jack Douglas]

You are getting kind of far fetched there, Jack!

So you don't think my rumour will gain traction!

Jack

 

[ethanz]

So you don't think my rumour will gain traction!

Jack

Give it a CR2 and you'll gain traction.

 

[dtaylor]

you compare different sensor technologys.
when you say something like this you better compare exactly the same sensor technology (both BSI, both the same ADC etc.) but with different photosite sizes.

The Sony A7r III and A7 III have the same sensor tech and the same DR despite the latter having larger pixels.

but imo you can´t ignore that per pixel quality will suffer with smaller pixels.

Apparently not at the densities we currently see on FF and APS-C sensors.

 

[JonSnow]

That's assuming the rich and powerful haven't turned us all into robots to bow and serve and work for nothing, sooner.

you work for exposure.....

 

[JonSnow]

The Sony A7r III and A7 III have the same sensor tech and the same DR despite the latter having larger pixels.
.

you mean it the tests where they normalize the image to 8MP.

from 3200 ISO, you start to see more noise from the A7R III and it becomes more evident from 6400 ISO upwards.

again i am not saying pixel size makes a huge difference today.
but imo there should be one (difference) just from a statistical/physical point of view (well capacity, noise , SNR ratio etc.)

someone who put way more time and effort into this than i would:


http://www.clarkvision.com/articles/does.pixel.size.matter/

We do observe a difference in image quality in the images output from a camera with larger versus small pixels. But such difference can be effectively mitigated in post processing as illustrated in Figures 10 and 11.

When choosing between cameras with the same sized sensor but differing pixel counts, times have changed. A decade ago, I would have chosen the camera with larger pixels (and fewer total pixels) to get better high ISO and low light performance. Today I would choose the higher megapixel (thus smaller pixels). Modern cameras with high megapixel count, low read noise and low electronics noise allow one to trade resolution and noise. If one wants reasonable dynamic range in a high megapixel camera, the pixels must still be large enough to hold enough photoelectrons to give the dynamic range. Currently (circa 2016), that is not much smaller than 4 micron pixel spacing. For example, the 50 megapixel Canon 5DS(r) fits this criteria, and Nikon's D800 and D810 are in the same league.

Pixels smaller than 4 microns can often be limited (lower contrast and loss of fine detail) by diffraction. But now that system noise is low in modern cameras, adding pixels together to synthesize an image from a sensor with larger pixels, diffraction limited smaller pixels is less of a concern.

 

[SaP34US]

I would say the 75 mp camera will come late next year and if there a second higher next year will come out earlier in the year will be 50mp.
One of the lower end models will have 26mp or 30mp with some different specs than the EOS R that is currently out and one with 24mp or 26mp with similar specs to EOS R but something less then the new camera.

 

[dtaylor]

you mean it the tests where they normalize the image to 8MP.

Choose whatever image dimensions you like. The results will be the same.

from 3200 ISO, you start to see more noise from the A7R III and it becomes more evident from 6400 ISO upwards.

Not at the same view size. Here they are at 25,600, RAW, same view size:



The noise is sharper (vs. A7 III) but so is the whole image. And there's not more of it. Whether you print as is for maximum sharpness or apply some NR, the A7r III image will come out on top.

To be clear: all currently shipping FF bodies are excellent at high ISO. There's not even 1 stop difference between the best and the worst. But if we nitpick we find that it's not the lower resolution bodies that come out on top.

again i am not saying pixel size makes a huge difference today.
but imo there should be one (difference) just from a statistical/physical point of view (well capacity, noise , SNR ratio etc.)

Obsessing over "pixel quality" is missing the forest for the trees. If you have gapless micro-lenses the total image noise will be the same. Smaller pixels just means a higher sampling frequency resulting in higher resolution and ooc sharpness. The character of the noise is different because it is 'sharper' as well. That may annoy some people, but you have more room for NR in that case.

Well capacity is a real thing and should in theory mean that larger pixel sensors have higher DR. But for some reason it hasn't since at least 2012 (D800 release). For whatever reason, architecture has trumped this. Even with Canon's older, off-chip ADC architecture the highest DR sensor was also the highest pixel density sensor.

At this time you're not losing anything for sampling at a higher frequency (smaller pixels). I'm not saying this will always be true. Perhaps past a certain size noise scales at a rate which would make this untrue. But today? We're just not seeing a loss for having more MP.

Also: Did you actually read Clark's article? All the way to the end?

When choosing between cameras with the same sized sensor but differing pixel counts, times have changed. A decade ago, I would have chosen the camera with larger pixels (and fewer total pixels) to get better high ISO and low light performance. Today I would choose the higher megapixel (thus smaller pixels). Modern cameras with high megapixel count, low read noise and low electronics noise allow one to trade resolution and noise. If one wants reasonable dynamic range in a high megapixel camera, the pixels must still be large enough to hold enough photoelectrons to give the dynamic range. Currently (circa 2016), that is not much smaller than 4 micron pixel spacing. For example, the 50 megapixel Canon 5DS(r) fits this criteria, and Nikon's D800 and D810 are in the same league.

 

[JonSnow]

Choose whatever image dimensions you like. The results will be the same.


Also: Did you actually read Clark's article? All the way to the end?

When choosing between cameras with the same sized sensor but differing pixel counts, times have changed. A decade ago, I would have chosen the camera with larger pixels (and fewer total pixels) to get better high ISO and low light performance. Today I would choose the higher megapixel (thus smaller pixels). Modern cameras with high megapixel count, low read noise and low electronics noise allow one to trade resolution and noise. If one wants reasonable dynamic range in a high megapixel camera, the pixels must still be large enough to hold enough photoelectrons to give the dynamic range. Currently (circa 2016), that is not much smaller than 4 micron pixel spacing. For example, the 50 megapixel Canon 5DS(r) fits this criteria, and Nikon's D800 and D810 are in the same league.

i have. years ago and a few times since then. i even qouted it in my post.
so did you actually read my post?

but you are so desperate to look 100% correct that you don´t get my point.

i mentioned downsampling a few times.

when ADC conversion is near perfect downsampling is basically "creating" bigger photosites without negative effects.
the photon noise is then combined into a "bigger bucket", without the negative effects of read noise and ADC introduced noise etc. on each single pixel.

ADC converters are so good now that you won´t see a huge difference between donwsampled images and images from a sensor with bigger photosites.

but nevertheless there still is photon noise difference on a pixel level in the original images....

 

[Sharlin]

Seriously, no one seems to be able to get a joke these days.

That's because on the internet there are plenty of people who say ridiculous things in all seriousness. See Poe's law.

 

[dtaylor]

i mentioned downsampling a few times.

You mentioned it once in the relevant post, and did so in passing.

but nevertheless there still is photon noise difference on a pixel level in the original images....

I guess it's my turn to ask if you've read my post or if you merely wish to appear correct: Obsessing over "pixel quality" is missing the forest for the trees.

 

[Ozarker]

So you don't think my rumour will gain traction!

Jack

I heard IT HERE first!