Canon EOS R5 Mark II going to 60mp? [CR1]

[neuroanatomist]

"It can" doesn't mean "it will". I'd argue that percentage of well composed shots is much higher from the Phase One cameras than from the phones.

Yes, and if you carefully read the sentence that I wrote and you quoted, you’ll see that I stated, “…can still have,” not ‘will have’.

Once again, the point (that you’re persistently ignoring) is that high technical quality is meaningless unless combined with reasonable artistic quality, whereas high artistic quality can positively distinguish an image regardless of the technical quality.

But I won't be buying a 60Mp full frame camera to mount a 100x 1.4 NA oil objective on it.

Of course not, which is why I also referenced a use case more relevant to ILCs. If you’d like a more specific example, how much will the extra MP of your new 60 MP FF camera increase the resolution in the corners of a shot with the RF 16/2.8?

 

[neuroanatomist]

12-bit images from the R5 are noticeable noisier and poorer quality than the 14-bit ones.
The difference becomes less prominent only from ISO 400-800.

Is C-RAW 12-bit (with mechanical shutter on the R5)?

 

[john1970]

Is C-RAW 12-bit (with mechanical shutter on the R5)?

I believe that C-RAW is a lossy compressed version of RAW, but I do not think it alters the bit depth. For the R5 I believe that the mechanical shutter is 14 bit and the electronic shutter is 12-bit. At least that is what I could find looking a various info from Canon.

 

[neuroanatomist]

I believe that C-RAW is a lossy compressed version of RAW, but I do not think it alters the bit depth. For the R5 I believe that the mechanical shutter is 14 bit and the electronic shutter is 12-bit. At least that is what I could find looking a various info from Canon.

That was my understanding as well. Probably @Quarkcharmed is conflating C-RAW with the bit depth loss from using other than the full mechanical shutter on the R5 (1-bit for EFCS, 2-bits for full electronic).

One thing I really like about the R3 is that it delivers the full 14-bits in all shutter modes.

 

[David - Sydney]

However, I remember from the DRone Wars of past years how important it was (to some) to be able to lift shadows with impunity, and Canon was criticized (mostly by SoNikon trolls) for noise and lost detail when pushing the shadows slider to the right. You probably remember when DPR came up with the up to 6-stops shadow lifting tool for their studio scene (presumably because they wanted a new hammer for Canon-bashing). Eventually, Canon's updated sensor designs addressed the 'problem', and Canon users could routinely lift shadows by a few stops when needed.

I routinely (and deliberately) underexpose indoors sports to have a fast enough shutter speed with good depth of field (martial arts sparring) at reasonable ISO and then push in post. Pushing a least a stop and commonly a couple of stops is normal in this use case especially when the lighting is uneven on people's faces. Metering gets a little confused as the gi is solid white.
I would never want or need to push 6 stops though.

 

[David - Sydney]

RED CEO Jim Jannard Speaks About Their Lawsuit Against Sony Regarding RAW Compression

I mentioned that we might see a response on REDUser from the CEO of RED regarding the recent lawsuit against Sony, and that's exactly what we got earlier this morning. In the lawsuit, RED is claiming that Sony is infringing on patents related to REDCODE RAW compression, since Sony is also using...

nofilmschool.com

"In the lawsuit, RED is claiming that Sony is infringing on patents related to REDCODE RAW compression"
Clearly, I said raw is raw rather than raw compression codec/algorithms for either video or stills.
I can't imagine a scenario where recording a pure raw file could be subject to a patent.

 

[john1970]

That was my understanding as well. Probably @Quarkcharmed is conflating C-RAW with the bit depth loss from using other than the full mechanical shutter on the R5 (1-bit for EFCS, 2-bits for full electronic).

One thing I really like about the R3 is that it delivers the full 14-bits in all shutter modes.

I also like the fact that the R3 is 14 bit in all shutter modes as well and I expect the same for the upcoming R1.

 

[Quarkcharmed]

That was my understanding as well. Probably @Quarkcharmed is conflating C-RAW with the bit depth loss from using other than the full mechanical shutter on the R5 (1-bit for EFCS, 2-bits for full electronic).

One thing I really like about the R3 is that it delivers the full 14-bits in all shutter modes.

Probably yes, I thought it was about 12-bit electronic shutter.
If it's about CRAW, I'm not sure if we know what lossy compression in CRAW does. If it's lossy, it doesn't mean it replaces 'natural' shot noise in lower bits with some generated noise. Why would it and why would it help compression (well it might help compression but at a cost of losing dynamic range).

 

[Quarkcharmed]

Yes, and if you carefully read the sentence that I wrote and you quoted, you’ll see that I stated, “…can still have,” not ‘will have’.

Once again, the point (that you’re persistently ignoring) is that high technical quality is meaningless unless combined with reasonable artistic quality, whereas high artistic quality can positively distinguish an image regardless of the technical quality.

But how does that prove that 60Mp is too much?
Also, bad technical quality can ruin a good artistic image.

Of course not, which is why I also referenced a use case more relevant to ILCs. If you’d like a more specific example, how much will the extra MP of your new 60 MP FF camera increase the resolution in the corners of a shot with the RF 16/2.8?

Tbh I have no idea, we'd need to wait for the 60Mp camera and tests to be done. However, higher frequency sampling (higher pixel density) generally increases the resolution. Your example that the increase will be 1-2% was not based on actual measurements.

 

[Kit.]

the bit depth loss from using other than the full mechanical shutter on the R5 (1-bit for EFCS, 2-bits for full electronic).

It's 1-bit for H+ drive mode. It has nothing to do with EFCS itself.

If it's about CRAW, I'm not sure if we know what lossy compression in CRAW does. If it's lossy, it doesn't mean it replaces 'natural' shot noise in lower bits with some generated noise. Why would it and why would it help compression (well it might help compression but at a cost of losing dynamic range).

Due to fundamental physical limitations, the square root of the (linear) signal that carries photon count is shot noise.

So, when you get a signal sample and find its most significant non-zero bit, about half of the less significant bits are noise that conveys no information from the scene and thus just doesn't need to be compressed.

If you throw these bits out, you don't lose dynamic range, but you will be throwing out less bits from the samples near the bottom of the dynamic range and more bits from the samples near the top.

 

[koenkooi]

That was my understanding as well. Probably @Quarkcharmed is conflating C-RAW with the bit depth loss from using other than the full mechanical shutter on the R5 (1-bit for EFCS, 2-bits for full electronic).

One thing I really like about the R3 is that it delivers the full 14-bits in all shutter modes.

As @Kit. remarked, it's not about EFCS per se, but drive speed. If you go over 8 fps, Canon has the R5 start dropping bits. I'm not sure if it's a sensor or Digic limitation, at the highest modes, it all seems to multiply out to 900MP*12bits, even on the R3/R6II.

 

[neuroanatomist]

It's 1-bit for H+ drive mode. It has nothing to do with EFCS itself.

Thanks for the correction!

 

[Quarkcharmed]

So, when you get a signal sample and find its most significant non-zero bit, about half of the less significant bits are noise that conveys no information from the scene and thus just doesn't need to be compressed.

It would have been a very simple and fast noise reduction algorithm if it were so.
But unfortunately, you can't strip the noise out by discarding a half of lower bits.

 

[Kit.]

It would have been a very simple and fast noise reduction algorithm if it were so.

I don't see how this would follow from what I said.

But unfortunately, you can't strip the noise out by discarding a half of lower bits.

Removing the bits of the signal that contain nothing but noise is not "stripping the noise out".

 

[Quarkcharmed]

I don't see how this would follow from what I said.

You said the noise is in lower bits. So let's just strip them out and we have easy cheap noise reduction.

Removing the bits of the signal that contain nothing but noise is not "stripping the noise out".

You can't have bits of the signal that contain nothing but noise, if you're talking about the shot noise.

 

[Kit.]

You said the noise is in lower bits.

No, I said the lower bits are noise. I didn't say that there were no noise at all in the higher bits.

So let's just strip them out and we have easy cheap noise reduction.

Won't work, unless by "noise reduction" you mean something other than a photographer would mean.

You can't have bits of the signal that contain nothing but noise, if you're talking about the shot noise.

What else would you expect to extract from those bits, and how?

 

[cooldood]

Is it just me or is this thread giving you a headache?

 

[Adrianf]

Sadly the headline spec of cameras is always the pixel count. IMHO, it's never been about the number of pixels, it's the quality of the pixels that counts. How many lenses can even resolve an image sufficiently to get the full benefit of 60MPx anyway?

 

[AlanF]

Sadly the headline spec of cameras is always the pixel count. IMHO, it's never been about the number of pixels, it's the quality of the pixels that counts. How many lenses can even resolve an image sufficiently to get the full benefit of 60MPx anyway?

I don't know what you mean by "get the full benefit"? What I do know is that my RF 100-500mm at 500mm f/7.1 on my R7 (equivalent to 82 Mpx FF) gives me 40% more resolution than on my 45 Mpx R5 - it's the same as having a 1.4x TC on the R5. The point is that even with a relatively narrow lens, 60 Mpx is a real plus for resolution, and even 88 Mpx sees a serious benefit.

 

[Sporgon]

Sadly the headline spec of cameras is always the pixel count. IMHO, it's never been about the number of pixels, it's the quality of the pixels that counts. How many lenses can even resolve an image sufficiently to get the full benefit of 60MPx anyway?

Quite a few, but 60 mega pixels will enhance the output of any lens. For instance one vintage lens that I’m familiar with and has a strangely strong reputation is the 1964 era Takumar 35mm f/3.5. In fact its performance on film that is scanned to a high level is sadly lacking, producing an unsatisfactory result, yet on a 50mp digital FF camera it can produce quite good images.
Whether 60mp is required on a ‘do it all’ camera like the 5 series is another matter though.