Canon EOS R5 Specifications

[Bahrd]

[...]Regardless, when you scale the image down you no longer measure the sensor performance. Instead, you measure how good the image is for digital noise reduction [through downsampling]. All absolute values of measured DR become totally meaningless. [...]

When you are "in the field" do you take pictures or do you measure the 'per pixel' DR?

@Kit. gave you a nice analogy (perhaps originated from audio DSP): you can have a 1-bit sampler which is equivalent to (if not better than) a 14-bit ADC. You just need to sample faster.

 

[Gazwas]

The R6 which will likely cost a bit more than the R5, and I guess they will have a similar body, same batteries with optional battery grip.
It will be competing against the A7SIII which will also move to the ~4000$ price level where the S1H currently sits.

Thats a really interesting concept that I’ve never considered. I’d assume they would follow historic naming conventions going all the way back to the FD mount with the 1 series being the daddy but that could change with their modern mirrorless range.

 

[Michael Clark]

Cute

Sorry, it is a pet peeve of mine. Every time I see it it is like nails scraping a chalkboard.

 

[Michael Clark]

I really wonder about the 300mm f/2.8. The price of used ones has nosedived. There was a huge take up by birders and nature photographers when the II was released because it was the way to get to a hand held 600mm with a 2xTC and and f/4 420mm with good IS, and the only alternatives the antique 400/5.6 without IS, the older 100-400mm with two stops only or the dismal 400mm DO. The introduction of the 400mm DO II effectively killed it at the top end, as did the 100-400mm II at the affordable.

Most of the cheaper EF 300mm f/2.8 used lenses I see are either the pre-IS or first IS version. Where are you seeing bargains on the EF 300mm f/2.8 L IS II? What few I can even find available in the U.S. at 9 or 9+ grade are all over $5K, with a new one selling for $6100 (and probably cheaper via Canon Price Watch).

 

[Mark3794]

No new informations? I don't need sleep, i need answers!

 

[Quarkcharmed]

When you are "in the field" do you take pictures or do you measure the 'per pixel' DR?

I actually do 'measure' it, only implicitly. In the field I'm limited by exactly the per-pixel DR. I usually do ETTR in landscapes and I roughly know that a certain range from the left of the histogram is not recoverable. If I can't push it to the right any more, I know I have to do exposure bracketing/HDR.

@Kit. gave you a nice analogy (perhaps originated from audio DSP): you can have a 1-bit sampler which is equivalent to (if not better than) a 14-bit ADC. You just need to sample faster.

I responded above already. Yes you can improve the DR by digital processing, noise reduction, exposure blending, HDR etc. All these methods have very little to do with the sensor performance as you either lose information (by noise reduction) or need to capture multiple images. We need better sensors to reduce the need of HDR. Stacking/blending isn't always possible.

 

[Michael Clark]

Excuse me? A99 II is a formidable camera, albeit a bit nutty -- but there is a mirror.

- A

Can you still buy a new α99 II via an authorized Sony dealer?

 

[Michael Clark]

I have an economics background and having two separate lens lineups for essentially the same camera (one being mirrorless, the other DSLR) makes no economic sense.

Canon is finally coming out with the mirrorless body worthy of their professional grade R lenses. At some point (when they judge that there is enough market penetration for the cameras) they will cease production of the EF version of the R lenses.

Here's the thing. Canon does not yet have two separate (full) lens lineups in both mounts yet, and won't for quite some time.

There are over 80 different EF lenses in their current catalog. The RF stable is not nearly that broad yet.

 

[padam]

No new informations? I don't need sleep, i need answers!

It is going to be a pre-announcement, probably with the camera on display just like with the 1DX III or the RF holy trinity of lenses, and the full details will have to wait until July.

So they might not actually reveal a lot more than what has been leaked already.

 

[Joules]

It is going to be a pre-announcement, probably with the camera on display just like with the 1DX III or the RF holy trinity of lenses, and the full details will have to wait until July.

So they might not actually reveal a lot more than what has been leaked already.

The 1DX III is the only camera in recent Canon history that got a development announcement before the proper announcement which revealed all the details. I don't think any other bodies will get the same treatment. So we can expect to know most of the details about the next R in the upcoming weeks.

I mean, it also wouldn't really make sense to have a development announcement for a camera that has no predecessor and therefore only becomes interesting to people once they know enough of the specs.

 

[Michael Clark]

Almost 20.

And you really think the difference between DSLRs and MILCs is anywhere as revolutionary as the difference between film and digital?

 

[Bahrd]

[...] All these methods have very little to do with the sensor performance as you either lose information (by noise reduction) or need to capture multiple images. [...]

The mistake in you inference is an assumption that the after-sampling/quantization DSP always means a loss of information. It doesn't.
♫ All in all it's all just [a counting photons from] bricks in the wall ♫ - whether you sum them all at once (a big pixel with a great full well capacity) or in smaller chunks (with smaller pixels with shallower wells) and just add them up.

The Eric Fossum's Quanta Image Sensor is an example of a 1-bit sensor with, in principle, an unlimited DR.

 

[padam]

The 1DX III is the only camera in recent Canon history that got a development announcement before the proper announcement which revealed all the details. I don't think any other bodies will get the same treatment. So we can expect to know most of the details about the next R in the upcoming weeks.

I mean, it also wouldn't really make sense to have a development announcement for a camera that has no predecessor and therefore only becomes interesting to people once they know enough of the specs.

Some sources claimed it doesn't look too different from the R besides the joystick, scroll wheel and photo/video switch, so in that regards it does have a predecessor, just not a direct one, since it is a higher end model and not a R Mark II.
That's the point, if the leaked specs are mostly true, then the headline features are already there to grab the attention. Besides that, they just need to show how it looks like. That's what Panasonic started with the GH5 if I remember correctly.

It also doesn't make sense to fully announce a camera and show how it exactly works, etc. when it is only going to be available in the summer anyway.

 

[Joules]

So it's on your side now to explain how you're going to have more than 14 stops in a 14-bit file.

By not encoding the numbers linearly

Are you guys not just arguing if absolute or relative measures are more relevant at this point? In which case the answer should be simply subjective and dependent on a specific use case.

It also doesn't make sense to fully announce a camera and show how it exactly works, etc. when it is only going to be available in the summer anyway.

Yeah, I'm simply sceptical of the July part for that reason. But maybe the strain on their production caused by launching so many bodies with new technologies in such short time (90D and M6 II, 1DX III and supposedly 3 more FF bodies just this year) is bigger than anticipated and they require greater lead times ro get stocks to the level they are confident with.

 

[Gazwas]

Some sources claimed it doesn't look too different from the R besides the joystick, scroll wheel and photo/video switch, so in that regards it does have a predecessor, just not a direct one, since it is a higher end model and not a R Mark II.
That's the point, if the leaked specs are mostly true, then the headline features are already there to grab the attention. Besides that, they just need to show how it looks like. That's what Panasonic started with the GH5 if I remember correctly.

It also doesn't make sense to fully announce a camera and show how it exactly works, etc. when it is only going to be available in the summer anyway.

The July release time frame is a strange one but suppose it gives Canon something to talk about at the photo shows they mentioned in their financial report before release. If they don't open orders until the end of July then customers won't get cameras until late August or early September which I'd say was practically autum.

 

[padam]

Yeah, I'm simply sceptical of the July part for that reason. But maybe the strain on their production caused by launching so many bodies with new technologies in such short time (90D and M6 II, 1DX III and supposedly 3 more FF bodies just this year) is bigger than anticipated and they require greater lead times ro get stocks to the level they are confident with.

It is simply in the final stage of development with the hardware being final, but the product itself not quite production ready yet, that will only start later and they will make sure that it works reliably. Photokina looks like a good event to present it in full production form, quote: With Canon suggesting it is ‘eager’ to launch new products at the show.

Just they did with the RF holy trinity, they want to assure users or potential future buyers that is it definitely coming and reveal some things that confirms their claims about being fully committed to mirrorless. If they wouldn't do this, it would just make people even more impatient and unsure about what is coming and when. After these two cameras, they probably won't need to do this again for a while.

 

[Yasko]

Is the person behind SonyAlphaRumors ghost-writing a CR post? This has got to be a contender for the biggest load of "wishful thinking rumor" I've ever read.

The more shocking thing is that half the commenters here aren't even questioning it. Did it not raise any eyebrows that the rumor has a bunch of insane specs, paired with something as silly and minor as "no touch bar"? Red flag!

Speaking of "red", ...you do realize that 8K RAW is usually reserved for $15-30K cameras, right?

Now with more time having passed and some of these (if not all) speccs more or less „confirmed“:
The EOS R5 may well step into the footsteps of the EOS 5D mk II back in 2008!

 

[Quarkcharmed]

The mistake in you inference is an assumption that the after-sampling/quantization DSP always means a loss of information. It doesn't.
♫ All in all it's all just [a counting photons from] bricks in the wall ♫ - whether you sum them all at once (a big pixel with a great full well capacity) or in smaller chunks (with smaller pixels with shallower wells) and just add them up.

Down-sizing is loss of information, noise reduction is loss of information. You can combine several pixels into one but you lose resolution and colour data. Btw when one of the pixels gets saturated, you can't combine properly anymore. Say one large pixel has a well capacity of 8 electrons, two smaller ones 4 each. Now the two small pixels receive 8 photons, 5 + 3. The resulting value will be 7 while a larger pixel would have gotten all 8. Smaller well capacity is still a limiting factor.

The Eric Fossum's Quanta Image Sensor is an example of a 1-bit sensor with, in principle, an unlimited DR.

I'm looking forward to seeing it implemented in a real camera, but I'm not sure what you're trying to prove here. Basically my point was, you can't make 7D's sensor perform on par with A7RIV by applying noise reduction in post.
How does Quanta Image Sensor prove or disprove my point?

 

[Kit.]

I'm not trying it. You are

Not me. You. I'm asking why you think the information you are trying to record exists in the first place.

So it's on your side now to explain how you're going to have more than 14 stops in a 14-bit file.

Easily. By not trying to extract 14 stops of DR at the spatial frequencies you cannot sample from a lower resolution sensor anyway.

Or should I explain you what spatial frequencies are and how a higher resolution sensor has more of them?

 

[Kit.]

Down-sizing is loss of information, noise reduction is loss of information. You can combine several pixels into one but you lose resolution and colour data. Btw when one of the pixels gets saturated, you can't combine properly anymore. Say one large pixel has a well capacity of 8 electrons, two smaller ones 4 each. Now the two small pixels receive 8 photons, 5 + 3. The resulting value will be 7 while a larger pixel would have gotten all 8.

Actually, it's the other way around. On a sensor with smaller pixels in your example, you know that you are near the saturation. On a sensor with bigger pixels, this information is lost.