R5II dynamic range and noise at photonstophotos

[Sporgon]

Bill Claff has the analysis of the R5II up on photons to photos.

Photographic Dynamic Range versus ISO Setting

DR at base ISO fractionally lower than original but pretty insignificant in practice.

 

[AlanF]

All the points for the R5ii have downward pointing triangles, which denote noise reduction being applied to RAW throughout the whole iso range.

 

[basketballfreak6]

Does anyone know the reason for the DR bump at ISO 400? I know it's a thing with the newer Canon sensors but never really knew the reason for it.

 

[AlanF]

Does anyone know the reason for the DR bump at ISO 400? I know it's a thing with the newer Canon sensors but never really knew the reason for it.

Using different amplification.

 

[basketballfreak6]

Using different amplification.

So is it dual gain that people talk about?

 

[AlanF]

Yes

 

[Loswr]

So is it dual gain that people talk about?

People were talking about dual gain output (DGO) for the R1 (didn’t happen), which is two gains used to increase the dynamic range of a single image. This isn’t that.

The DGO sensor explained

How the EOS C300 Mark III and EOS C70 deliver 16+ stops of dynamic range by blending high and low gain images using Canon's revolutionary new DGO sensor technology.

www.canon-europe.com

Dual (and triple, in some cameras) gain amplifiers have been used for some time now, and that’s different. The latter is what accounts for the ‘DR bump’ you see from Bill Claff’s data.

 

[basketballfreak6]

People were talking about dual gain output (DGO) for the R1 (didn’t happen), which is two gains used to increase the dynamic range of a single image. This isn’t that.

The DGO sensor explained

How the EOS C300 Mark III and EOS C70 deliver 16+ stops of dynamic range by blending high and low gain images using Canon's revolutionary new DGO sensor technology.

www.canon-europe.com

Dual (and triple, in some cameras) gain amplifiers have been used for some time now, and that’s different. The latter is what accounts for the ‘DR bump’ you see from Bill Claff’s data.

Thanks for that

 

[YuengLinger]

R5 1.5?

 

[Quack]

Whew! All those who were out on the ledge can come back in now.

 

[amorse]

While seeing a bit of a dip from the R5 at ISO 100 on mechanical shutter may be a bit sour, there's a few good news stories for measurebaiting in there.
The Electronic shutter on the R5ii is notably better than on the R5 at lower ISOs, and better than the Z8 at every ISO from 100 and onward, while the mechanical shutter is better at ISO 100 than the Z8 is at its base ISO 64. For anyone actually leaning on the speed of the R5ii, that's a pretty good story - less rolling shutter, more dynamic range at high-speed, and definitely very comparable (to really any other full frame camera) dynamic range at peak with mechanical shutter. With that said, Bill Claff also noted that while there is noise reduction at all ISOs on the R5ii, there is less noise reduction on low ISO in the R5ii than in the R5. I'm not sure if that will improve shadow detail, but it may counter some of the concerns attached to noise reduction being baked in to spoof DR at low ISO.

 

[AlanF]

While seeing a bit of a dip from the R5 at ISO 100 on mechanical shutter may be a bit sour, there's a few good news stories for measurebaiting in there.
The Electronic shutter on the R5ii is notably better than on the R5 at lower ISOs, and better than the Z8 at every ISO from 100 and onward, while the mechanical shutter is better at ISO 100 than the Z8 is at its base ISO 64. For anyone actually leaning on the speed of the R5ii, that's a pretty good story - less rolling shutter, more dynamic range at high-speed, and definitely very comparable (to really any other full frame camera) dynamic range at peak with mechanical shutter. With that said, Bill Claff also noted that while there is noise reduction at all ISOs on the R5ii, there is less noise reduction on low ISO in the R5ii than in the R5. I'm not sure if that will improve shadow detail, but it may counter some of the concerns attached to noise reduction being baked in to spoof DR at low ISO.

It truly is measurebaiting. The R5ii in ES does have 0.4-0.5 more DR at iso 50-100, but from iso 200 onwards, the R5 has the edge. It's all immaterial to me as I never shoot below iso 200 and am usually up at iso 800-1600 for my nature photography.

 

[neurorx]

I shoot a lot of lower light photography at high speeds for dance and use electronic shutter with my R5. I also do senior photos outdoors which is almost the opposite end of the perspective. I am wondering how this might impact now compare to the R5m2. Any technically inclined photographers have an advice here? Thank you.

 

[amorse]

I shoot a lot of lower light photography at high speeds for dance and use electronic shutter with my R5. I also do senior photos outdoors which is almost the opposite end of the perspective. I am wondering how this might impact now compare to the R5m2. Any technically inclined photographers have an advice here? Thank you.

I’m not sure that I’m a technically inclined photographer, but I’d be pretty surprised if anyone could distinguish an image from an R5 and an R5ii in a blind test based on everything I’ve seen so far. It’s all so close, whether ahead or behind, when comparing as closely as one can with ISO and ES vs EFCS considered.

 

[Sporgon]

With that said, Bill Claff also noted that while there is noise reduction at all ISOs on the R5ii, there is less noise reduction on low ISO in the R5ii than in the R5. I'm not sure if that will improve shadow detail, but it may counter some of the concerns attached to noise reduction being baked in to spoof DR at low ISO.

I wonder if that’s why the II is slightly lower at base ISO. After his work on the original R5 Bill Claff was pressed on DPR to say how much NR was being applied. His answer was that he was not able to calculate it, but his guess was between 1/3 and 2/3 of a stop at base ISO. So if you take that amount off you get the same base ISO DR from the R5 as you do from the 5DIV, which is why I included that camera on the graph. The R5II falls between the two. As the 5DIV already had ample latitude in the shadows it makes the drop one thing less to worry about.
It’s quite fun to apply NR to just the 0-8 values in the raw file from a 5DS and see how much Sonyesque DR you can get out of it
Bill Claff says he as found evidence of NR in the lowlights of these Canon files, but (generally) not other makes. What baffles me is why we don’t see more shot (photon) noise in the pushed raws of Sony, Nikon etc.

 

[docsmith]

If the R5 II had increased performance...that would have been interesting. Providing the same excellent sensor performance as the R5 is fine, IMO. And not the type of fine where I will yell about it later.

While Measurebaiting....I hadn't taken the time to look at the Nikon Z8/9 against the R5....for people that flip out about DR...what up?

Oh, yeah, it isn't Canon....and the Z8/9 really do take excellent images.

 

[neurorx]

I’m not sure that I’m a technically inclined photographer, but I’d be pretty surprised if anyone could distinguish an image from an R5 and an R5ii in a blind test based on everything I’ve seen so far. It’s all so close, whether ahead or behind, when comparing as closely as one can with ISO and ES vs EFCS considered.

Thank you. This is bothering me a lot. It wouldnt make sense for me to get the camera if it isnt capturing ES with better or same DR.

 

[Diko]

Basically aside from faster read out speeds I see no new innovation in that new CMOS sensor. That is most unfortunate. To see no improvement in more than eight years in that direction. I try to shoot events without flash whenever I can. And I really hope that back illuminated sensor offer more.

Why really aren't camera manufacturers implementing dual ISO (GDO). If I recon correctly Magic Lantern even had it back 15 years ago.

On a side note I'm really impressed how fast the DXO came out with the results.

 

[Loswr]

Basically aside from faster read out speeds I see no new innovation in that new CMOS sensor. That is most unfortunate. To see no improvement in more than eight years in that direction.

There's really not much room for improvement in things like DR and noise. That's what happens with a mature technology. The new frontier is readout speed, but if innovation along those lines doesn't matter for what you shoot, then there's no reason to upgrade your sensor (of course, the sensor is only one part of the camera and not upgradable alone).

If you want better IQ from a CMOS sensor, you need to be looking at sensors larger than FF.

 

[Diko]

...
If you want better IQ from a CMOS sensor, you need to be looking at sensors larger than FF.

I agree with what you're saying.

But back in the day there was hopes about graphene film layer addition. And last night I met a scientist that explained me why it is not considered as an option anymore. Something between the lines of graphene lacking the density to transfer over the current from the converted signal from the photons.

Actually I don't need any more improvement of the IQ. I need flexibility in order to shoot without a flash. And as far as I know probably MF won't do the job as well. However to be honest I've never had the chance to test medium format camera let's say on a concert for obvious reasons of lacking speeds necessary for this kind of photo shootings. ;-)