Canon EOS R5 Specifications

[privatebydesign]

You're still comparing a 24mm prime shot at 1/125 with a 12-24mm zoomed all the way in at 24mm (Please see Roger's Law of Wide Zoom Relativity) shot at 1/60.

Pretty much all of the differences I see in your two crops can be accounted for by camera motion and differences in focus distance as well as lens performance. I see nothing that can be attributed to differences in DR. That's not to say that the 1D X Mark II doesn't allow more pushing of shadows than the 1Ds Mark III, but these shots don't seem to demonstrate that. Neither one is remotely close to showing either camera's DR limits.

I wasn't saying your other points weren't valid, I was correcting the factual errors you made.

I wasn't comparing lens sharpness either, I took the pictures specifically to look in the shadows, relative sharpness was not a consideration of mine. That others throw up a straw-man of they are framed differently, they aren't the same sharpness, dof etc etc while not looking solely at the metric the comparison was shot for is comical.

 

[dtaylor]

So as in my message above, the argument was about DxO and PTP measurements; how do you use say the DxO value of 13.5 stops for the EOS R, based on viewing at a print, when any real print has 7 stops max? What does this value tell you?

No offense, but it tells me you don't know enough about this topic to be discussing it. Mapping captured DR to a print is part of the art and science of photography which was described in technical detail at least as far back as Ansel Adam's books.

 

[dtaylor]

And that is why it is pointless posting here.

It is pointless to post two pictures which do not stress, test, or illustrate DR differences then get upset when that is mentioned by multiple people. I'm not saying you're wrong in general about 1DX2 vs 1Ds3 DR. I'm simply pointing out that these images do not illustrate the difference.

The difference between usable shadow lifting ability in that 1.2 stops of DR is dramatic and has proven worth it to me.

There's no real read noise in either one and no shadow zone difference. In other words you didn't push one to a higher zone than the other (illustrating more room to push) nor did one show a weakness not found in the other (noise).

Unless I'm mistaken and you eliminated read noise in the 1Ds3 image with heavy NR, the softness has nothing to do with DR.

 

[dtaylor]

I wasn't comparing lens sharpness either, I took the pictures specifically to look in the shadows, relative sharpness was not a consideration of mine.

Perhaps you should post the RAWs because I don't see much in the way of deep shadows there. Unless you severely underexposed these shots, pushed 3-4ev, and then applied heavy NR to the 1Ds3 shot to mask read noise, these shots didn't test DR.

The DR differences we are debating occur at zones 0 and 1 in an unprocessed RAW. Basically in shadow detail that's not apparent...areas that are black or nearly so...before the push. If you're at ISO 100 and taking zone 3 to zone 5 read noise is a non issue as far back as a 10D.

That others throw up a straw-man of they are framed differently, they aren't the same sharpness, dof etc etc while not looking solely at the metric the comparison was shot for is comical.

The only two differences between your crops are sharpness and color balance. You posted them saying "this is the difference" leaving others to assume that somehow DR resulted in a sharper photo. If that was not your intent then what is the difference we're looking for in your crops?

 

[dtaylor]

In one of the previous similar arguments with dtaylor, I've already referred to DPReview, but he wasn't convinced.

I think DPReview's lab tests are a great way to examine sensor DR if you have enough experience to know how that maps to the real world.

They've done a couple real world comparisons which are decent if you download the RAWs and give them proper processing.

Neither he was happy with my sample images from 5DIV alone.

You mean the 1.5ev push that any DSLR ever made could handle?

 

[epic.one]

These specs don't look realistic to me. The 4K 120fps is not like Canon at all, same for the 12/20fps.

I thought about this, but it could be possible with the 1D's processor.

 

[PureClassA]

Canon Thoughts - The internet loses its mind

It seems the internet has weighed in with the Canon EOS R5. With some interesting, and even some amusing takes. Let's be real here. The EOS R1 (or whatever it will be called - I think if it's an R5 and R6, odds are very good it's an R1) is coming in 2021. Canon is going to leave things off...

www.canonnews.com

Great piece from CanonNews

 

[jam05]

Hardly. If it is any 5D mirrorless then it is on PCP, coke and steroids all at once. The 5D doesn't do this kinda of video stuff (8k and 4k 120,or even 4k full width). The 5D line doesn't do anywhere close 12fps m-shutter, or 20 fps e-shutter. etc etc etc.

If anything this is a high MP 1DXIII mirrorless, cause 20 fpse-shutter is A9 territory. Just from that alone I would guess this will be around 4-5k price point. Knowing canon maybe even close to 6k.

In my opinion, RP = 6D, R = 5D line, and this thing would be 1D line. As close to an analogy as it can be, cause of the R5 is more in the form factor of the R it is 5D form factor but 1D specs...

You mean the 5D of yesteryear that is. It's 2020 not 2017. Heck smartphones will be pushing 8k 30fps this year and 108 and 200mp on that Snapdragon 865. Many of us in the tech industry aren't in awe of.

 

[Kit.]

No offense, but it tells me you don't know enough about this topic to be discussing it. Mapping captured DR to a print is part of the art and science of photography which was described in technical detail at least as far back as Ansel Adam's books.

What is especially hilarious is that the classic B&W silver halide film (and paper) is a "1-bit" sensor. Each single grain after processing becomes either fully transparent or fully opaque.

 

[AEWest]

Canon Thoughts - The internet loses its mind

It seems the internet has weighed in with the Canon EOS R5. With some interesting, and even some amusing takes. Let's be real here. The EOS R1 (or whatever it will be called - I think if it's an R5 and R6, odds are very good it's an R1) is coming in 2021. Canon is going to leave things off...

www.canonnews.com

Great piece from CanonNews

Thanks. That is quite a detailed explanation as to why 8K is possible.

Biggest question mark then is heat dissipation...how do you shoot 8K for any length of time without the camera becoming a toaster? I guess we'll find out soon enough!

 

[Quarkcharmed]

You mean the 1.5ev push that any DSLR ever made could handle?

No, not any. Old Canon sensors don't handle such a push well, deep shadows become unacceptably messy. Even not so deep shadows are noisy to start with.

 

[Quarkcharmed]

No offense, but it tells me you don't know enough about this topic to be discussing it. Mapping captured DR to a print is part of the art and science of photography which was described in technical detail at least as far back as Ansel Adam's books.

You simply didn't answer the question. The question was, how exactly do you use the DR value of say 13.5 stops for EOS R from DxO?

In their calculations, as far as I can see, they don't actually do any DR compression to match the DR of a paper print, which would mean some mapping of the 14 bit image to 6 or 7 bits.

 

[Quarkcharmed]

What is especially hilarious is that the classic B&W silver halide film (and paper) is a "1-bit" sensor. Each single grain after processing becomes either fully transparent or fully opaque.

Only that is totally irrelevant to what we were talking about, that is DxO or PTP method of calculating the DR.

 

[Kit.]

Only that is totally irrelevant to what we were talking about,

You were wondering how a 14-bit sensor could give more than 14 stops of DR.

B&W film was a "1-bit sensor". What DR do you think it would be limited to?

 

[PureClassA]

Thanks. That is quite a detailed explanation as to why 8K is possible.

Biggest question mark then is heat dissipation...how do you shoot 8K for any length of time without the camera becoming a toaster? I guess we'll find out soon enough!

Possibly externally. But I dont know if HDMI can support that data rate. But I think external
Recorders woth hdmi top out at 4k60. Beyond that, its either internal or SDI and there’s no way you’ll get sdi on a small body like this.

Data Rate:
8k@30fps = 4k@120 (roughly) assuming each readout is 1:1 pixels. Hence a HUGE crop in 4k120. Like 2.5x to 3x (someone do the math for me)

hence I guarantee CFExpress card slot on this camera with the probably second slot cfast2.0 or SDXC (or whatever the fastest SD card is now)

 

[Dequals61]

In economics, there is the concept of Supply and Demand. As long as customers are willing to pay for EF lenses, Canon will produce and sell them. Canon makes a profit from the sales of EF lenses. Canon's plan is not to force people into switching to RF. They are producing RF lenses that are superior to their EF counterparts so folks will want to switch...

Supply and demand is fine and true but there will come a time when EF lenses will no longer be produced. Now in Canon speak we are probably talking years. This would not happen until Canons full camera line up was completely RF compatible mounts, that includes the Cine line stuff as well. After all Canon's flagship 1DxIII which even isn't available yet for purchase is only EF mount compatible.

 

[PureClassA]

Supply and demand is fine and true but there will come a time when EF lenses will no longer be produced. Now in Canon speak we are probably talking years. This would not happen until Canons full camera line up was completely RF compatible mounts, that includes the Cine line stuff as well. After all Canon's flagship 1DxIII which even isn't available yet for purchase is only EF mount compatible.

Best to Clarify whether we mean Canon will cease to make NEW EF glass vs. Canon will cease all existing production. Both will happen. But two phases.

Canon has already signaled clearly that few if any NEW EF glass in development. Maybe a few tweaks on some of the Great Whites as EF bodies will long be in use for a big pro segment. How do we know? Brand new 50L came out after many many years of people waiting for an upgrade. It came out for RF. Hope no one is holding their breath for an EF counterpart. What about a 24-70f2.8 L IS for EF? Nope. Not gonna happen.

Canon will finish cranking out their current EF production schedule and let the stock slowly dwindle down to nothing and move full tilt into RF. And yes this will take several years.

And as soon as the RF Crop bodies start emerging, the EFM line will be killed off in the same fashion. As keeping the M line makes no sense with RF out there. Notice how few M lenses Canon has bothered to develop? Once the FF RF lens family gets fleshed out, the RF-S (like EF-S crop) will start up.

 

[slclick]

Supply and demand is fine and true but there will come a time when EF lenses will no longer be produced. Now in Canon speak we are probably talking years. This would not happen until Canons full camera line up was completely RF compatible mounts, that includes the Cine line stuff as well. After all Canon's flagship 1DxIII which even isn't available yet for purchase is only EF mount compatible.

Therefore I conclude that EF will have a production or at least healthy stock numbers for the typical 1D series cycle. i.e. EF lenses for 4-6+ years? I mean it's not such a hard thing to wrap your head around or necessary of 138 pages of CR dialog....They made the R series compatible in the best way with EF. Folks here love to complicate EVERYTHING (see current sensor jabber *Yawn*) Must be fun at parties!

 

[Quarkcharmed]

You were wondering how a 14-bit sensor could give more than 14 stops of DR.

During the last few pages we didn't even touch that topic. Current discussion is about relevance of absolute values of DR from DxO and PTP, at least for me. The conversation is sidetracking all the time though.

And no, 14-bit sensor can't produce a raw file with more than 14 stops of DR. Moreover you don't get additional DR by simple downsampling. You need to convert to 16 or 32 bits, then downsample.
But that's not the original file and most importantly not the original resolution is it? I think the DR at native resolution is also important, or per-pixel DR before any conversions.

B&W film was a "1-bit sensor". What DR do you think it would be limited to?

If you convert it to digital (by scanning), you'll get a 16-bit image and will be able to calculate the DR. If you don't want to convert it, you use analog methods to calculate S/N. You'd probably need to normalise both film and digital in order to compare, if you want to compare film and digital. Neither of imaginable comparison methods will involve converting film crystals into 1-bit pixels. Most likely you'll figure out the resolution of the film first and then you'll get a number of "1-bit crystals" per resolving unit area, not just one bit. But the crystals in film are not laid out in a nice regular 2D grid, they're stacked and overlap in 3D. Overall this comparison of film to 1-bit sensor is fallacious.

 

[Ozarker]

Folks here love to complicate EVERYTHING (see current sensor jabber *Yawn*) Must be fun at parties!

I think there may be more Dynamic Range talk than there is about the R5 itself. I'd go back and count, but I could go back to school and get a Dr. quicker. Or an art degree so I'd know what art is.