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

jolyonralph

Game Boy Camera
CR Pro
Aug 25, 2015
1,423
944
London, UK
www.everyothershot.com
===

I highly doubt Canon itself will make an RF to EF adapter...BUT... I do think someone like Wooden Camera might sell one as a package along with one of their Camera Cages and/or other accessories. In terms of focus and iris, they could put a control ring into the adapter itself for semi-automatic and full-automatic focus and iris control protocols which already have been reverse engineered from a technical point of view by Sigma and others.

Sorry Harry, you're completely barking mad on this.

If you are desperately trying to hold onto the dream that some day maybe you'll be able to use the glorious RF glass on an EF body then by all means continue to delude yourself.

It won't happen. Canon would never do it. They want to transition everyone over to RF as soon as possible.
Wooden Camera? They almost certainly haven't got the experience to reverse engineer the RF protocol.
Sigma? They almost certainly haven't reverse engineered the RF protocol yet. Remember it took them DECADES to get the EF mount right (something that Tamron still haven't 100% figured out with some of their lenses not working on the adapted R, and they couldn't even get the EF-M protocols right).

The RF protocol is far more complex than EF. For all we know it may require signed authentication between lens and body, and/or have an encrypted communication channel.

The only way you'll ever be able to adapt an RF lens to work on an EF camera is if you stick an EOS R on the hotshoe.
 
Upvote 0
Then maybe user will just have to MANUALLY focus
Sorry Harry, you're completely barking mad on this.

If you are desperately trying to hold onto the dream that some day maybe you'll be able to use the glorious RF glass on an EF body then by all means continue to delude yourself.

It won't happen. Canon would never do it. They want to transition everyone over to RF as soon as possible.
Wooden Camera? They almost certainly haven't got the experience to reverse engineer the RF protocol.
Sigma? They almost certainly haven't reverse engineered the RF protocol yet. Remember it took them DECADES to get the EF mount right (something that Tamron still haven't 100% figured out with some of their lenses not working on the adapted R, and they couldn't even get the EF-M protocols right).

The RF protocol is far more complex than EF. For all we know it may require signed authentication between lens and body, and/or have an encrypted communication channel.

===

It might be encrypted. That wouldn't surprise me, but since SPEED is a big issue, it would have to be some FAST but cheap encryption like the original 64-bit or an extended 128-bit version of FEAL-NX (i.e. from NTT Japan) where you can put a "Man-in-in-the-Middle" chip to sign the codes with your own digital signature which would NOT violate any of Canon't patents. To figure the protocol, all you need to do is put a scope between the junctions and see what comes out! Run the captured signals through some deep learning/data mining software and PRESTO I could have a whole table of control codes in about two months. It takes four high end GPU number crunching cards on a decent PC to figure out some encrypted signals that are 128-bits wide within a few weeks and create your own digital signatures put on your own DSP chip!
 
Upvote 0
Currently I am still working with 5D Mark II bodies. Great cameras but I get it, tech has moved on.

I had the opportunity to shoot with a 5DsR for half a week for free so took it. Man it's good, the resolution and detail were great but the AF and other features were just as nice for me. I can see a 5Ds being added to the kit in the future as I don't think the 5DsR is worth it for my work.

But 75mp on a 35mm format is just too much. You have to take care shooting the 5Ds/R so with 75mp it's going to further limit its usability, and for me personally over 50mp = going medium format.
 
Upvote 0

Don Haines

Beware of cats with laser eyes!
Jun 4, 2012
8,246
1,939
Canada
Then maybe user will just have to MANUALLY focus

Strike one - what do the extra pins do on the RF Mount? These are extra to the EF mount, in other words, the EF mount is a subset of the RF Mount. You can use a subset in a superset, but not a superset in a subset

Strike two - the focal distance. EF lenses focus on a plane further away than the RF Mount does. That means, put an EF lens on a RF body and it focuses behind the sensor. Put an RF Lens on an EF body and it focuses in front of the sensor. You can easily add a spacer on a RF body to allow an EF lens to focus properly, but you can not take one away (because there is none) to allow an RF lens to focus on an EF body. To do this, you need to add a series of lens elements between the end of the RF lens and it’s focal plane, and with the tight geometry on RF lenses, this is a virtually impossible task that would at best greatly degrade the image quality.

Strike three - you still have not given me a ride in your helicopter :)
 
Upvote 0

jolyonralph

Game Boy Camera
CR Pro
Aug 25, 2015
1,423
944
London, UK
www.everyothershot.com
Then maybe user will just have to MANUALLY focus

Which, if you had been paying attention and/or actually understood what you're talking about, you would know they can't do that because RF lenses have focus-by-wire systems meaning you have to decode and understand the RF protocols even to manually focus the lens.

Lens technology has come on a long time since the days of simply pulling one line on the interface low to signal that the focus should move in a specific direction.

Give it up Harry, you are really out of your depth on this one.
 
Upvote 0
Which, if you had been paying attention and/or actually understood what you're talking about, you would know they can't do that because RF lenses have focus-by-wire systems meaning you have to decode and understand the RF protocols even to manually focus the lens.

Lens technology has come on a long time since the days of simply pulling one line on the interface low to signal that the focus should move in a specific direction.

Give it up Harry, you are really out of your depth on this one.

===

I am quite aware that RF lenses are focus-by-wire! I am ALSO WELL AWARE that I can put a bunch of probes between the pins and see what the O'Scope says. I am assuming some pulse width modulation as the underlying schema. Even IF there was encryption (probably cheap but fast FEAL-NX from NTT Japan), I can always put in my own Man-in-the-Middle chip to sign my own packets/commands with my own digital signature and send them to the lens controller chip once I figure out whether or not encryption is used. I can put an expert system on the problem using a bunch of GPU graphics cards to do the signals-oriented number crunching, since we already know the GENERAL protocols of Canon lenses and can have the expert system figure out which set of PWM signals mean what and for when. It would take our engineering team maybe three months to do, since we can go and buy 20 of the new cameras and function-by-function figure what modulated signal expresses what effect on the lens and then build up a lookup table to duplicate all the functionality and NOT fall afoul of any of Canon's Patents!

It is NOT THAT HARD !!! Spend some money! $75,000 is NOT a big deal for a company to buy some cameras and some Gigahertz scopes to decode a communications protocol!

---

In terms of lens elements, if they're high refractive index and optical-sciences classes of Acrylic, then you might get a 1/8th or less stop of light loss, so that F/1.2 50mm become F/1.3 instead! Still a bargain for the $2300 price to make your DSLR 1Dx and 5D, 6D and 7D series last a few more years until Canon comes out with their 75 megapixel RF monster ...OR...you finally capitulate and buy our high frame rate 50 megapixel 16-bits per channel Medium Format large sensor system instead!
 
Last edited:
Upvote 0
Mar 26, 2014
1,443
536
Even IF there was encryption (probably cheap but fast FEAL-NX from NTT Japan), I can always put in my own Man-in-the-Middle chip to sign my own packets/commands with my own digital signature and send them to the lens controller chip once I figure out whether or not encryption is used.

More probably a new encryption protocol, so it would be patented, and any 3rd party lens implementing it would be pulled off the market for patent violation. That would provide for 20 years of protection - about half the lifetime of the EF mount.

IIRC, I've read somewhere part of the problem is Canon cameras are "too familiar" with the lenses, e.g. how each lens's focus engine responds to commands (how far it moves the focus). This means if a 3rd party lens pretends to be a specific Canon lens, it has to imitate that lens' engine response, and apparently that's not so easy.
 
Upvote 0

Don Haines

Beware of cats with laser eyes!
Jun 4, 2012
8,246
1,939
Canada
More probably a new encryption protocol, so it would be patented, and any 3rd party lens implementing it would be pulled off the market for patent violation. That would provide for 20 years of protection - about half the lifetime of the EF mount.

IIRC, I've read somewhere part of the problem is Canon cameras are "too familiar" with the lenses, e.g. how each lens's focus engine responds to commands (how far it moves the focus). This means if a 3rd party lens pretends to be a specific Canon lens, it has to imitate that lens' engine response, and apparently that's not so easy.

It’s not a problem,it’s an advantage. By knowing how fast elements accelerate and the amount of movement needed to get to the focus position, you can accelerate the lens element right up to the point where you need to apply braking, resulting in the shortest time possible to move that element. Obviously, canon has this data for thief lenses and is not going to determine it for Tamron or Sigma lenses and load it into their firmware.

As a result, Second party lenses have to pick the closest profile. They get close to the desired performance. As well, when the camera commands the lens, it sends “move X focus units”, and the second party makers lens will interpret that number as appropriate for its lens. This mitigates most of the hunting for focus.

As to encryption, this is HIGHLY unlikely. You lock out other manufacturers and you have every store clerk on the planet telling customers not to buy Canon because they don’t work on other lenses. That would be a sales catastrophe.
 
  • Like
Reactions: 1 user
Upvote 0
Mar 26, 2014
1,443
536
It’s not a problem,it’s an advantage. By knowing how fast elements accelerate and the amount of movement needed to get to the focus position, you can accelerate the lens element right up to the point where you need to apply braking, resulting in the shortest time possible to move that element.

I did not mean this is a problem for Canon, but rather for other manufacturers who need to "get close to the desired performance".

As to encryption, this is HIGHLY unlikely. You lock out other manufacturers and you have every store clerk on the planet telling customers not to buy Canon because they don’t work on other lenses. That would be a sales catastrophe.

What I meant was, if Canon wanted to block other manufacturers from making EOS-R lenses, standard encryption wasn't the best way of doing it.

Then again, if Canon thought 3rd party lenses were good for Canon, wouldn't it make the mount specification open?
 
Upvote 0
I did not mean this is a problem for Canon, but rather for other manufacturers who need to "get close to the desired performance".



What I meant was, if Canon wanted to block other manufacturers from making EOS-R lenses, standard encryption wasn't the best way of doing it.

Then again, if Canon thought 3rd party lenses were good for Canon, wouldn't it make the mount specification open?

===

Encryption takes TIME and POWER and lenses won't have the digital signal processing horsepower to implement a higher-level AES-256 type of algorithm. Canon will use something like a bunch of internally-designed bitwise XOR-and-NOT-commands as an encryption protocol OR maybe just use FEAL-NX which is good enough AND FAST, easily workable on 16-bit controller processors! It's not that hard to implement Feal-NX and any relevant key management system to just use your own signatures to send commands.

From an engineering point of view, I would say a company like Sigma could buy 10 of the RF cameras and assign some science interns to record on a scope what each signal set means from the various cameras so they can find out the general protocol which will LIKELY be a superset of the current Canon lens communications protocols. They could do it in about six months if they hire enough interns and buy them all scopes!

In the USA, a clean-room-based, WELL DOCUMENTED reverse engineering program that implements a FUNCTIONALLY similar interface but NOT the exact design IS ALLOWED and is cannot have a patent-lawsuit applied to it. Canon CANNOT sue Sigma for implementing it's own chip that sends PWM signals to a lens. It CAN sue Sigma if the chip's software uses nearly identical microcode! This is Why you implement using a DIFFERENT chip that uses a DIFFERENT instruction set but imitates the general functionality using a different physical design. That is perfectly fine to do patent-wise! Ergo, you can't patent a PWM signal (Pulse Width Modulation) or the IDEA of using a metal tube attached to a user-adjustable ring or wheel which starts and stops an electrical motor attached to a set of gears and a software interface!
 
Upvote 0

Don Haines

Beware of cats with laser eyes!
Jun 4, 2012
8,246
1,939
Canada
Then again, if Canon thought 3rd party lenses were good for Canon, wouldn't it make the mount specification open?

The best situation for Canon is that they remain #1 in sales and that they sell all the lenses that are used on their cameras, but that is not going to happen. If they block third party lenses, then they loose camera sales, and then lens sales. As said above, if all the store clerks say not to buy Canon because they do not allow other lenses/accessories, sales plummet. And remember, it is the low end cameras and kit lenses that make up the vast bulk of sales. Most people buy a single camera and one or two bundled kit lenses. They are not going to throw that away to protect higher end lenses
 
Last edited:
Upvote 0

===

You CANNOT PATENT AN IDEA !!! Period !!! I've done enough Patent Lawsuits to know that is the case!

You CAN PATENT the specific implementation of an idea!

RSA only patented a specific implementation of Powers of X being the base calculation for a numeric result. This means that RSA specifically used the MOD (Modulo) operator after a base number value is raised by a specific power. That modulo result is used for other parts of the algorithm. That is ABSOLUTELY FINE in that a specific order of math operations that is NOT well known by the general public or by someone knowledgeable in the art can result in a difficult-to-factor result. This specific implementation of encryption is MOST DEFINITELY PATENTABLE!

However, my FUNCTIONAL duplicate of this, allows me to create some OTHER software that ENCRYPTS DATA! So WHY, as a business person would I WANT to duplicate the exact algorithm of a competitor, when I want TO IMPLEMENT AN IDEA (i.e. Encrypt Data) but be able to do it faster or with needing less hardware resources! Ergo, I create a NEW encryption algorithm (i.e. ROTATE-DATA-768) that emulates the function (i.e. Encrypt/Decrypt data) but do it FASTER and WITH LESS hardware resources!

---

In terms of hardware, the specific design and layout of the individual gates, pathways and logic of a CPU chip to obtain a specific result is patentable. What is NOT patentable is the IDEA of using junctions, pathways and storage locations to create a CPU or GPU chip of any type or size! In my case, I DO NOT WANT to emulate the pathways of an Intel i9 chip! I have a much fancier layout that spreads simple asynchronous microcores for EVERY TYPE of math and character processing function amongst an entire substrate keeping the cores closest to the cache and long-term memory as possible.

Ergo, I make a chip that allows me to work on 128 different text strings up to 65,535 characters in length AT THE SAME TIME (i.e. in Parallel). This is IDEAL for massively parallel text search algorithms! OR...I add simple microcores that do nothing but add, subtract, multiply, divide, root, power, mod and XOR, OR, AND, NOT, REVERSE BITS, SPIN BITS for an array of 4096 64-bit integer numbers in parallel so I can do a fancy convolution filter in parallel on a single 4K line of 64-bit RGBA video or still image pixels. That is a NEW TYPE of SIMD command so it's patentable! (Not anymore though, since I just disclosed this NOW as being open source under the GNU GPL3 licence!)

Remember, even IN a clean room environment, you are implementing the DUPLICATION OF A FINAL RESULT, not of the process itself! Ergo, you use as much divergence of process as you require so that ANY REASONABLE PATENT EVALUATOR would conclude that while you are implementing the IDEA of say compression or encryption, that the PROCESS of that end result is different enough that you do not infringe on another competitor's patents. They key point is, that you ARE ALLOWED to emulate or duplicate PARTS of a process that are deemed NOT NOVEL or NEW.

i.e. You can "Quote" parts (i.e. a few sentences) of your competitors processes but you CANNOT plagiarize "Entire Paragraphs and Pages" of their work!
.
 
  • Like
Reactions: 1 user
Upvote 0

dtaylor

Canon 5Ds
Jul 26, 2011
1,805
1,433
As to encryption, this is HIGHLY unlikely. You lock out other manufacturers and you have every store clerk on the planet telling customers not to buy Canon because they don’t work on other lenses. That would be a sales catastrophe.

It also won't work for the simple fact that any 3rd party lens can pretend it's an EF lens. Given the RF mount pin layout it is extremely unlikely that the RF-EF adapter does anything other than pass EF signals through while adding control ring data on the new pins. The RF protocol could be a high level, serialized and encrypted data packet protocol sent exclusively over the new pins. That still leaves the "dumb" and reverse engineered EF mode.

Canon marketing can claim what they want about the RF protocol, I doubt it makes any practical real world difference in lens performance at the end of the day. The 28-70 f/2L exists because of mount distance to the sensor, not the RF protocol. So an EF protocol RF lens would work fine.

If the RF protocol is not encrypted...and possibly even if it is...it's open to reverse engineering.

I wish Canon would just openly license EF and RF protocols and put an end to speculation, 3rd party lens glitches, etc. Actively trying to suppress 3rd party lenses via strong encryption and/or patents would probably send me to another brand, though that would be well off into the future. (My current DSLR kit exceeds what I need for years to come.)
 
Upvote 0

Don Haines

Beware of cats with laser eyes!
Jun 4, 2012
8,246
1,939
Canada
It also won't work for the simple fact that any 3rd party lens can pretend it's an EF lens. Given the RF mount pin layout it is extremely unlikely that the RF-EF adapter does anything other than pass EF signals through while adding control ring data on the new pins. The RF protocol could be a high level, serialized and encrypted data packet protocol sent exclusively over the new pins. That still leaves the "dumb" and reverse engineered EF mode.

Canon marketing can claim what they want about the RF protocol, I doubt it makes any practical real world difference in lens performance at the end of the day. The 28-70 f/2L exists because of mount distance to the sensor, not the RF protocol. So an EF protocol RF lens would work fine.

If the RF protocol is not encrypted...and possibly even if it is...it's open to reverse engineering.

I wish Canon would just openly license EF and RF protocols and put an end to speculation, 3rd party lens glitches, etc. Actively trying to suppress 3rd party lenses via strong encryption and/or patents would probably send me to another brand, though that would be well off into the future. (My current DSLR kit exceeds what I need for years to come.)
My bet is that the new pins on the mount add a high speed serial connection between camera and lens with differential inputs.

Also, odds are that all new info is through that serial link. It is highly unlikely that a pin would be dedicated to a specific function from a specific ring. The main advantage of the EF mount was getting rid of specific controls and replacing them with serial commands.
 
Upvote 0
Again, even IF Canon is using encryption, I am pretty sure it will be basic like FEAL-NX (from NTT Japan) or even a simple bitwise XOR-NOT-XOR-based function. AND I am also pretty sure they will SUPERSET the commands of the EF mount rather than code an entirely new communications protocol. A third party manufacturer like Sigma or Wooden Camera would find it it fairly easy (i.e. within 6 months) to emulate the protocol in a non-patent infringing way.

It WOULD be preferable for Canon to do an RF-to-EF adapter itself. And my suggestion is for Canon is to slowly STRANGLE the EF lenses by changing the mount to RF at each scheduled lens update cycle.

By switching each each lens (i.e. 135mm, 85mm, 50mm, etc) to RF mount and no longer offering the EF equivalent, BUT supplying an ADAPTER for EF mount users, those users are FORCED to use and adapt to the new RF lenses which in my opinion is quite superior optically to EF lens. The user will slowly get used to the fast new glass and at the NEXT purchase cycle will simply buy an RF mount body anyways. The large circle will quite surprise users in that the image brightness is noticeably BETTER on the RF 50mm than the EF 50mm. I have PERSONALLY NOTICED that the f1/2 is DEFINITELY SHARPER in the centre! (we have had multi-hour access to review units!) Even with an adapter, you're looking at only 1/8th to 1/4 stop of light loss depending upon your iris setting in our estimation. So the lens only become an f1.3 -- SO WHAT!!! -- It is NOT that big of a deal when you can get a noticeably superior lens on a EF body!

Sooooo, Wooden Camera and Sigma had better start cooking for making both RF to EF adapters and/or make some NEW large mount RF lenses!

One thing that is a MUST is focus control rings, iris control rings AND insertable or dial-selectable neutral density filers are ABSOLUTELY REQUIRED for any adapters!
 
Upvote 0