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

[Don Haines]

Ultimatly, there are only so many photons.

Lets use a pair of imaginary sensors to show the point with easy math. Both sensors are perfect, and both have a perfect A/D circuit to read them. Both sensors cover the same area, one is 10X10 pixels, the other one is 40X40 pixels.

Both sensors are mounted into a camera and are going to take the same picture of a white wall at the same exposure and aperture. Let's say 102,400 photons enter the lens and because our imaginary sensors are both perfect, we get 102,400 electrons. The 10X10 sensor has 100 photosites with 1024 electrons in each one, and this gives us 10 bits of colour depth. The 40x40 sensor has 1,600 photosights with 64 electrons in each one, and that gives us 6 bits of colour depth.

The 10X10 sensor has greater colour depth, but the 40X40 sensor has greater resolution.

However, you can bin the pixels of the 40X40 sensor to recreate the same image as the 10X10 sensor, and end up with the same lower res image at the same colour depth, but you can not go the other way around. The 40X40 image has more information than the 10X10 image. think of it like this, 10X10 X10 bits of depth is 1000 bits of information, while 40X40X6 bits of depth is 9,600 bits of information.

In the real world, it isn't so easy. A/Ds are not perfect, sensors do not have 100% quantum efficiency, and there is a seam on the edges of microlenses so some light is lost there, but it still comes to the same conclusion. Smaller pixels capture less individual cell info than larger pixels but because there are more of them, the overall amount of info for the entire image is greater.

 

[JoTomOz]

Surprised no one has talked about the lenses that will be needed to fully realise this kind of resolution. Does this mean all new RF L lenses will be a step above second gen EF L lenses?

 

[ethanz]

I heard IT HERE first!

Do we have a new HF?

 

[ethanz]

Ultimatly, there are only so many photons.

Lets use a pair of imaginary sensors to show the point with easy math. Both sensors are perfect, and both have a perfect A/D circuit to read them. Both sensors cover the same area, one is 10X10 pixels, the other one is 40X40 pixels.

Both sensors are mounted into a camera and are going to take the same picture of a white wall at the same exposure and aperture. Let's say 102,400 photons enter the lens and because our imaginary sensors are both perfect, we get 102,400 electrons. The 10X10 sensor has 100 photosites with 1024 electrons in each one, and this gives us 10 bits of colour depth. The 40x40 sensor has 1,600 photosights with 64 electrons in each one, and that gives us 6 bits of colour depth.

The 10X10 sensor has greater colour depth, but the 40X40 sensor has greater resolution.

However, you can bin the pixels of the 40X40 sensor to recreate the same image as the 10X10 sensor, and end up with the same lower res image at the same colour depth, but you can not go the other way around. The 40X40 image has more information than the 10X10 image. think of it like this, 10X10 X10 bits of depth is 1000 bits of information, while 40X40X6 bits of depth is 9,600 bits of information.

In the real world, it isn't so easy. A/Ds are not perfect, sensors do not have 100% quantum efficiency, and there is a seam on the edges of microlenses so some light is lost there, but it still comes to the same conclusion. Smaller pixels capture less individual cell info than larger pixels but because there are more of them, the overall amount of info for the entire image is greater.

Great explanation Don. Now lets see how many Forum Engineers rip into it.

 

[HarryFilm]

Do we have a new HF?

---

NOPE! The ORIGINAL HarryFilm is STILL here!

I should also note that it SEEMS my original notes about Canon were CORRECT in it's new focus on high megapixel counts and NEW mirrorless designs. They had better get on with it because a new kid on the block has a Medium Format MONSTER coming out soon which will make Canon's forays a bit of a moot point in the high megapixel count arena if it doesn't offer spectacular features to match!

---

NOW! To address one reason WHY we should have High Megapixel Count cameras of ANY SORT!

When you take a 75 megapixel camera (let's say 10,000 by 7500 pixels) and RESAMPLE that image using a high-quality image resizing algorithm such as Lanczos-3 or Lanczos-5 or Sin-C, you can take advantage of oversampling math as espoused within:

Nyquist-Shannon Sampling Theorem:
https://en.wikipedia.org/wiki/Nyquist–Shannon_sampling_theorem

AND

Nyquist Rate:
https://en.wikipedia.org/wiki/Nyquist_rate

which are not just used in their usual Audio Sampling/Recording domains but also used in Still Image resampling and VIDEO recording arenas.

If you average an ENTIRE STILL IMAGE or VIDEO FRAME using a resampling algorithm that samples on a 2x2, 3x3, 4x4, 5x5, 6x6, 7x7 or 8x8, etc. block-of-pixels and you ENSURE that the DESTINATION still image or video frame widths and heights DIVIDE EVENLY into the source width and height, then your final image quality will have a much higher SUBJECTIVE level of quality due to a "Natural" antialiasing effect brought about by the resampling algorithm. AND RIGHT AFTER you do the image resample of say a 10,000 x 7500 pixel image downto 5000 by 3750 pixels (i.e. exactly half the resolution on each axis), perform an UNSHARP MASK of about 1.0, 1.5 or 2 pixels radius, that function enhances ONLY object edges which now results in a PERCEIVED INCREASE of detail and overall image sharpness.

The key thing on large high megapixel count images is to DOWNSAMPLE them to exactly 1/2, 1/3 or 1/4 of their original width and height using Lanczos, Sin-C or even Bicubic image resize and then do an UNSHARP MASK filter. The final photo will look absolutely stunning. This downsampling and sharpening is usually done as the LAST STEP after any colour grading and image fx/correction tasks just before final publishing and/or printing.

So that 75 megapixel monster will make some AWESOME website front-page shots and/or bus-stop posters once printed out!

IDEALLY, Canon should INCREASE the physical size of the sensor to a full Large Sensor Medium Format (56mm by 42mm is suggested!) so that the size of each photosite in Microns is kept the same as the 1DxMk2 or at least the 5Dmk4 ...AND.... ensure each photosite is sampled at 32-bits per colour channel (96-bits per RGB pixel) which is THEN downsampled to 16-bits per channel of final colour sample resolution (i.e. 48-bits per RGB pixel). That would preserve light-gathering power and greatly reduce noise at lower light levels!

---

AND I DO RE-ITERATE, that a certain manufacturer IS coming out with just such a Combined Stills/Video camera with Nyquist Sampling/UnSharp Mask and it's coming much sooner than anyone realizes! ;-) ;-)


Cough! Cough! Look what the cat dragged in from Canon's arch-nemesis SONY !!!

Sony’s Next Full-Frame Sensor to Offer 60MP, 8K, 16-Bit RAW:

See link below:
https://petapixel.com/2018/11/26/sonys-next-full-frame-sensor-to-offer-60mp-and-8k-report/

Hmmmmm......I wonder what's coming NEXT down any given manufacturer's pipeline!

WE SHALL SEEEE !

And YES !!! You Heard it HERE FIRST !!!!!!!!!

 

[privatebydesign]

Surprised no one has talked about the lenses that will be needed to fully realise this kind of resolution. Does this mean all new RF L lenses will be a step above second gen EF L lenses?

Thats because talking about resolution in terms like that is an entirely specious argument.

Reposted yet again:-

System resolution can be broadly shorthanded down to this equation, it isn't perfect but pretty close.

tsr = 1/sqrt((1/lsr) ² + (1/ssr) ² )

Where tsr is total spatial resolution, lsr is lens spatial resolution, and ssr is sensor spatial resolution.

So if, for example, we have a sensor that can resolve 100 lppmm, and a lens that can resolve 100 lppmm we get this

1/sqrt((1/100) ² + (1/100) ² ) = tsr of 71 lppmm

Leave the same lens on, good or bad, and double the sensor resolution to 200 lppmm

1/sqrt((1/100) ² + (1/200) ² ) = tsr of 89 lppmm


You will notice that the system resolution, even in this simplified form, can never resolve 100% of the lowest performing portion of that system, so if a 24MP sensor is returning 80% of the potential of a lens then a 50MP sensor might return 90%, how useful that is in real life is a moot point, but it does illustrate that even the most modest lens will show increased resolution when put in front of a higher resolving sensor.

 

[ethanz]

---

NOPE! The ORIGINAL HarryFilm is STILL here!

Harry, do I have the magical ability to summon you? It appears that way.

 

[3kramd5]

Thats because talking about resolution in terms like that is an entirely specious argument.

Reposted yet again:-

System resolution can be broadly shorthanded down to this equation, it isn't perfect but pretty close.

tsr = 1/sqrt((1/lsr) ² + (1/ssr) ² )

Where tsr is total spatial resolution, lsr is lens spatial resolution, and ssr is sensor spatial resolution.

So if, for example, we have a sensor that can resolve 100 lppmm, and a lens that can resolve 100 lppmm we get this

1/sqrt((1/100) ² + (1/100) ² ) = tsr of 71 lppmm

Leave the same lens on, good or bad, and double the sensor resolution to 200 lppmm

1/sqrt((1/100) ² + (1/200) ² ) = tsr of 89 lppmm


You will notice that the system resolution, even in this simplified form, can never resolve 100% of the lowest performing portion of that system, so if a 24MP sensor is returning 80% of the potential of a lens then a 50MP sensor might return 90%, how useful that is in real life is a moot point, but it does illustrate that even the most modest lens will show increased resolution when put in front of a higher resolving sensor.

Another formula is the product of the transfer functions of each block in the system. The conclusion is the same: the total system resolution approaches the weakest link. Improving the strongest link will result in diminishing returns to the system, and improving the weakest link can result in the greatest returns to the system.

 

[HarryFilm]

"...Harry, do I have the magical ability to summon you? It appears that way....."

--

YES! You do....I AM HERE....i am there....I AM EVERYWHERE YOU WANT TO BE !!!!

with the LATEST in Camera Industry Rumours......and I do wish to update you all on a previous one which I think is HOT TO TROT !!!!!!

---

About 4 weeks ago, seven of my executive-suite sources that I have gotten super-duper-secret information from on a more-than-a-few-occassions, AND who are located in major business/import/export centres such as Toronto, New York, London, San Francisco/San Jose, Vancouver, Hong Kong and Berlin have ALL been telling me the same information which YOU here on CanonRumors GET TO HEAR FIRST !!!!!!!!!

Within the LAST THREE MONTHS, Apple Board Members have been stating eloquently within their meetings (with Apple CEO Tim Cook present!) that a corporate looksee and takeover is desired for both adding more technological supremecy to Apple's portfolio AND to basically boost their stock price into a sustainable MORE-THAN-ONE-TRILLION-DOLLAR valuation. With their 240 BILLION DOLLAR CASH HOARD, the THREE big companies Apple is looking at BUYING IN THEIR ENTIRETY for their EXTENSIVE Patent Portfolios IS CANON of Japan, TI (Texas Instruments) of the USA and Vodafone of the UK.

Canon of course has the SECOND LARGEST PATENT PORTFOLIO in the world after IBM ....AND....It is rumoured that a Japanese minister has given TACIT APPROVAL of APPLE buying ALL of Canon. I've heard through the grapevine that literally APPLE is the ONLY foreign company that is considered within Japanese political circles to even be ALLOWED to buy into and/or completely purchase such a major Japanese company! The current stock price premium being discussed is between 56 to 62 Billion Dollars U.S.

Apple wants Canon's Chip Making systems/expertise, Microcircuit Quality Control systems, OLED, Quantum-Dot and LED display patents, and optics patents. The grapevine is also talking about Super-SIRI enhanced 8k and even 16k television/computer displays using Canon LED/Quantum Dot techonology that will be coming in at 42, 55, 65, 83 and 110 inch sizes for Business and high-end Home use with BUILT-IN fully ENCRYPTED Facetime videophone systems! ....AND....I've heard a custom manufacturing plant is to be built in the State of Utah so those displays will be MADE IN USA --- WOW! THAT is something to write home about! So.....VERY SOON NOW...possibly by next September 2019, Canon Cameras and LENSES may become APPLE CAMERAS AND LENSES !!!!!

Texas Instruments has the 4th largest Digital Signal Processing and SoC (System-on-a-Chip) patents portfolio after NEC, Philips and Intel. That would contribute to Apple's Bionic CPU production capabilities and audio/video/still image processing patents portfolio. The grapevine is that Apple wants to get rid of Intel and COMBINE it's mobile and desktop CPU/GPU's into a SINGLE series of MADE-IN-HOUSE-BY-APPLE processors used for all watches, smartphones, tablets, desktops and OTHER devices. These NEW processors are to have built-in high end GPU's for 8k/16k video display, high-end audio processing and HIGH FRAME RATE 480 fps+ 4K VR using a custom Apple-built 480 frames per second VR visor sending 240 fps video to each eye slightly offset to give super-smooth VR experience! That frame rate (which IS doable on high end CPU's/GPU's at 4K!) would allow Apple to be the PREMIERE SUPPLIER of VR systems in the world! The stock premium that is being bandied about is $25 Billion U.S.

Vodafone Group Plc.....Hmmmmm.....that kinda surprised me when I heard that.....BUT when I did a bit of digging....they own the largest mobile provider in Germany and are HUGE players in the UK, Middle East, South America, South Asia and India giving Apple access to over 2.5 Billion NEW subscribers! NOW that finally makes sense to me! I am also betting that by buying a MAJOR mobile voice and data services provider, I am BETTING TWO BITS that Apple wants to practice in a non-American market so they can get enough experience to finally and EVENTUALLY buy up an American Major Phone Company such as Verizon! If THEY own the wireless pipes, it's only a matter of TIME that iTunes will be the ONLY content player available on those out-of-USA wireless/mobile networks which could add another 80 Billion Dollars PER YEAR in wireless phone/data services revenue to Apple's coffers! That would turn 2018's revenue from $265 Billion to over $345 Billion which is a 30% INCREASE in yearly revenue! YIKES!!!

So NOW I get WHY Vodafone is being considered as a takeover target by Apple! It would be the HARDEST ONE to bite into though as it's stock premium valuation would be on the order of about 72 to 76 Billion Dollars U.S. Apple Can do it Moneywise! And I would dare say that the extra $80 Billion U.S. in yearly revenue makes Vodafone the most attractive of the takeover targets from an INITIAL financial contribution viewpoint BUT Canon is BEST from a patents portfolio viewpoint which could have all it's technologies used as a springboard for NEW revenue-producing products from Apple.

Apple IS ALWAYS looking at companies to buy and in MY OPINION it will be Vodafone FIRST and then Texas Instruments within a SHORT period of time afterwards! In terms of Apple Buying Canon anytime soon, YES there IS a possibility it will be chosen first because of it's HUGE PATENT PORTFOLIO....but with the sheer number of employees it would have to absorb or spin off into new companies, I'm not quite sure WHAT Apple would do to handle that political hot potato of Apple buying a truly Japanese company as Canon!

Anyways...It's a TOSS-UP ---- who knows...maybe Canon will go for Sigma to get their stacked RGB photosite Foveon sensor technology ...OR... Apple buys Sony to get the Playstation Network AND gain control of major worldwide Image Sensor production! Who Knows???

WE SHALL SEE ....and..... YOU HEARD IT HERE FIRST !!!!!!

.

 

[scrup]

Da fuk, everytime I hear entry level R, I think what else can they gimp on the camera.

 

[justaCanonuser]

A higher resolution sensor will never perform worse than a lower resolution sensor due to diffraction or motion blur. Blur may stand out more when pixel peeping at 100% because the higher resolution file is enlarged more, but it's not actually any worse at the same view/print size.

Diffraction is also not a brick wall. "Diffraction Limited Aperture" is terrible terminology because it causes people to think of it as a brick wall.

That's true, and I didn't mean it that way, sorry if my posting was mistakable. In general, the image information captured by a e.g. a 35mm sensor is always the same, given the same amount of light and comparable silicon technology, and given a minimum pixel size that fits to the size of the Airy discs caused by a certain aperture. For those not knowing what an Airy disc is: That's the area of a geometrical image point washed out by diffraction, and such a disc contains no useful image information. Now, starting with an aperture in which Airy discs and pixel pitches fit together, and closing it further, those Airy discs grow and overlap more and more pixels. So, now you lose gradually more and more image information captured by a single pixel - the image get's softer and softer on the pixel level. If you e.g. close to f = 14 for classic landscape photography to gain a big depth of field, you just get fat files from your 75 MP monster that include no more useful information than smaller files from a, say, 24 MP camera. Your just need more processor power and disc space for such huge files, and emit more carbon dioxide into earth's atmosphere.

This is the reason why I always would change to medium format cameras if I wanted to make huge prints, because e.g. a 6 x 4,5 sensor with 80 or 100 MP technically makes much more sense.

 

[justaCanonuser]

"...Harry, do I have the magical ability to summon you? It appears that way....."

--

YES! You do....I AM HERE....i am there....I AM EVERYWHERE YOU WANT TO BE !!!!

with the LATEST in Camera Industry Rumours......and I do wish to update you all on a previous one which I think is HOT TO TROT !!!!!!

---

...
WE SHALL SEE ....and..... YOU HEARD IT HERE FIRST !!!!!!

.

Whoa! And I always think I tend to write too long posts, respect, HarryFilm . You should re-establish French style epic novel writing of the 19th century. Just polish a bit the dramaturgic composition of your posts...

 

[Del Paso]

Do we have a new HF?

Do you miss him too?

 

[Del Paso]

Unless it does 8K, no ?

Is proutprout the equivalent of petpet?

 

[JasonLee]

Did anyone give it a thought that it would be an iteration what Foveon used to apply on their sensor? 25MP x 3, boomed, you have 75MP.

I sees that it is possible given that Canon have filed the stacked sensor pattern previously.

 

[Kit.]

think of it like this, 10X10 X10 bits of depth is 1000 bits of information, while 40X40X6 bits of depth is 9,600 bits of information.

That's not completely true. 40X40X6 needs to transfer that much more bits of data (unless pixel binning is implemented directly between photocells), but how much extra information this extra data contains depends on how much information in the corresponding frequency range (5 to 20 cycles per sensor) actually reached the sensor.

 

[NKD]

Great news, will hold onto my 5dsR for another couple of years or until I land a few more large gigs to justify the upgrade.
Love the idea of having focus peaking & drop in rear filters in the new era for tricky lenses like the 17mm TS and Laowa 12mm

One feature I would love to see would be an updated camera remote trigger port. For a wired shutter adjustments, this would help me eliminate a laptop in the field to capture accurate light & dark frames. Bracketing shots can be hit and miss, having to shoot up to 7 to get the right one.

 

[VINESCAM1507]

I'm not tracking the people you see as offenders, meaning I don't know who is complaining both ways. But it is so easy to just lose patience and call those who don't share enthusiasm a troll. From what I understand, a troll is somebody who, perversely, posts in a forum for the sole purpose of angering reasonable forum members.

But what I'm reading in this thread, for the most part, is understandable frustration, skepticism, and concern. You might not agree with such sentiments, but people spending thousands of dollars, sometimes unwisely in terms of their own budgets, get emotional, cynical, and even bitter. If they are posting simply to express their feelings (and I'm seeing quite the surge in newer members since the EOS R was released), are we to just call them all trolls--without even engaging first to find out what their actual issues might be with a Canon releasing a 75 MP before, say, a best in class 40-50 MP? (Maybe Canon has discovered the Holy Grail of sensor tech, who knows?)

Now if I tell you that it would be nice if Canon made a dSLR as good as the Nikon D850, would you call that trolling? I don't think it is. Yes, Canon makes calculated, clear-eyed business decisions that have kept it the leader, but, as a customer, I can't help look at the competition and think to myself, I wish Canon made one like that. So, while I support the development of higher resolution bodies and lenses for those seeking such specifications, I'd also like to see Canon addressing the desires of many other photographers who, like myself, aren't seeking more MP at this time.

Is it rude to say, "Hey, what about me? My photography desires deserve priority!" while in a thread about rumored gear a member doesn't find appealing? In a way, yes, I'll grant you that, dak723. But it's an open forum with relaxed rules about staying on topic, and people react with their first thought, little filtering. I don't think it is helpful to just classify such posts as trolling, because, one, they are not, and two, it doesn't help the thread, the person posting, or the general feeling of goodwill that CR manages to maintain better than many other sites.

Just my opinion!

When you look at the "competition" and "wish" Canon made "one" like that ? Why?
I mean is your art suffering so drastically that you can't do with what you have.
I show my pics to lots of people and none of them pixel peep or say that the dynamic range could be better in that because you only used a 5D mk3. Had it been a Nikon D850 it would have sooooooo much more resolution and DR and less noise etc. If you want all the features of a Nikon then buy one and stop whinging!

 

[YuengLinger]

When you look at the "competition" and "wish" Canon made "one" like that ? Why?
I mean is your art suffering so drastically that you can't do with what you have.
I show my pics to lots of people and none of them pixel peep or say that the dynamic range could be better in that because you only used a 5D mk3. Had it been a Nikon D850 it would have sooooooo much more resolution and DR and less noise etc. If you want all the features of a Nikon then buy one and stop whinging!

I think if my desire for a better camera were unique, mine alone, we'd all still be using Canon 20D's. Or maybe film? And if customers don't want more and better, why are we even discussing 75 MP?

Yes, George Hurrell created great portraits back in the 1930's using 8x10 plates and huge, hot, and heavy lights. I suppose that's what I should be satisfied with, as they created great art.

Never "wish" for better tech? Does this apply to cars, TV's, and computers?

I prefer Canon's service and lenses, and it is in fact the lenses which tie us, to an extent, to a camera brand. So, I'd like to see best in the industry bodies too, even if they are only a little better than what Canon currently offers. I have friends with Nikon D8xx bodies, and the Nikons have numerous extra niceties for photographers in the menus, the ergonomics, and other aspects. Usability, not just image quality, is a factor in deciding whether one body or the other is more attractive to an individual.

Not sure what triggered you here, in your very "first" post (unless this is just a new screen name for you), to claim that I'm "whinging" for hoping the next release of a Canon body is even better than the current very good ones.

Happy holidays!

 

[ethanz]

Do you miss him too?

Yes, Del Paso, I do. Adds a little excitement to life.