First downloadable RAW R5 and R6 files

[Aussie shooter]

The R5 rocks my best Canon sensor the 5DII at ISO 4000 but it isn't up to the Z6(which I think pre applies NR to the RAW). The R6 on the other hand looks really clean, but those where lower ISO samples. I am just not sure about that R5 image, it is a little soft and uninteresting, I'll wait for some more examples but especially in the eyes I was thinking to myself that I would need to do a wee bit of work to get this right.

Just now I think it is the test image itself that is making me think it is a bit meh of a performer. And it could also be that f/11 lens doesn't match up to lenses I am used to using.

Everything about that R5 image is against it. F11, high ISO, Diffraction and very low crappy light. I doubt ANY sensor would give you a good image in that situation

 

[Deleted member 381342]

Everything about that R5 image is against it. F11, high ISO, Diffraction and very low crappy light. I doubt ANY sensor would give you a good image in that situation

Yup which is why I will wait for more examples from this camera.

 

[Adam Shutter Bug]

On safari you need a telephoto zoom, it's a must. And if you take a RF 100-500 (or even an EF + adaper), a 2xTC will give you a much more useful 200-1000mm f/9-11 zoom than either of the f/11 primes. I'd sacrifice the possible hit on IQ for the advantages of a zoom.

I would too, I really wish They had gone with something along the lines of a 200-600 f/5.6. But its the only use case I can see for these is something like a Safari and a Canon rep has even given it as the example.

 

[Aussie shooter]

Yup which is why I will wait for more examples from this camera.

And I forgot to mention low shutter speed to obviously try and get as much exposure as possible. I don't care how good the stabalization is. Handholding an 800 at that shutter speed is gonna be problematic unless you are incredibly well balanced

 

[Adam Shutter Bug]

Really? If you have a look at the image of the cheetah I put up a few posts back, that is raised by about 1.5 stops. And it is still just a tad dim IMO. Maybe Topaz works differently but there is no way it would look like an ISO 100 shot with any sort of NR I could do. As for the flamingo. It was clean and sharp But yeah. looked a bit Wierd. Maybe it is more that the colour of Flamingoes can make an image a bit odd?

Ive uploaded the image finally but only a downgraded version sadly. It is very clean for what it is. I still wish they had gone another way with the lens but for what it is its not as bad as it reads on paper.

 

[Aussie shooter]

Ive uploaded the image finally but only a downgraded version sadly. It is very clean for what it is. I still wish they had gone another way with the lens but for what it is its not as bad as it reads on paper.

Defintely better than LR. And certainly waaaaaaaaaaaaaaaay better than anything I can achieve on my 7d2 at anything over 800. Still a bit darker than what I would have done but that is a matter of style, not a matter of process. I am definitely looking forward to seeing an R5 image in good circumstances though. Those R6 images were crisp AF and I can only imaging the R5 has to be far sharper. Still out of my price range though. Will def get the R6 when it comes down to a reasonable price in Australia though

 

[Adam Shutter Bug]

Defintely better than LR. And certainly waaaaaaaaaaaaaaaay better than anything I can achieve on my 7d2 at anything over 800. Still a bit darker than what I would have done but that is a matter of style, not a matter of process. I am definitely looking forward to seeing an R5 image in good circumstances though. Those R6 images were crisp AF and I can only imaging the R5 has to be far sharper. Still out of my price range though. Will def get the R6 when it comes down to a reasonable price in Australia though

A friend put me on to Topaz a while ago, it is much better I’ve found than what I can do in lightroom myself especially with masking. If its a very noisy image from say my EOSR 4000-6400 it struggles As theres just no detail to really preserve but I have gotten prints from a ISO6400 image before saying that.
It was an editing choice to go slightly darker as I found lighter showed the background a bit too much. Nothing spending more time wouldn't sort but this was only a test really. if doing a proper job I would probably go up another half stop just on the subject which is easily done and I did raise the overall exposure at one point by another two stops just to try it and it handled fine. There was no chromatic aberration or fringing or anything really in the colour that made me go B&W either, that was purely a stylistic choice based on the lines of the subject and I didn't care for the colour of the background.

 

[Deleted member 381342]

And I forgot to mention low shutter speed to obviously try and get as much exposure as possible. I don't care how good the stabalization is. Handholding an 800 at that shutter speed is gonna be problematic unless you are incredibly well balanced

I am only caring about the noise at ISO 4000. I have started to get comfortable shooting in dark environments at ISO 3200 and even ISO 10,000 which has been rather clean on newer bodies.

 

[highdesertmesa]

Lets put this to bed.

You said


Here is the R6 compared to the R
View attachment 191458

Studio shot comparison: Digital Photography Review

Expert news, reviews and videos of the latest digital cameras, lenses, accessories, and phones. Get answers to your questions in our photography forums.

www.dpreview.com

You were 100% wrong.

Not at 1:1 on each the higher you go. I predict your next move will be to say that reduced to 20mp the R will look about the same, which is a false argument since no one upgrades to higher resolution just to view images at 75% or to not print larger.

 

[highdesertmesa]

R6 sensor lab test is up at DPR. It is as good as 1Dx3 at ISO6400 and it is not much better than 5D4 or R at ISO6400. People get emotional. Sure. Let’s look at a reliable data instead.

The person I quoted via screenshot was referencing R5 samples.

 

[privatebydesign]

Not at 1:1 on each the higher you go. I predict your next move will be to say that reduced to 20mp the R will look about the same, which is a false argument since no one upgrades to higher resolution just to view images at 75% or to not print larger.

View attachment 191461

If you are comparing, that is making a valid comparison, the sensor noise you have to view at same size, the size is irrelevant but the size has to be the same. Comparing per pixel noise is basically irrelevant in this instance as the objective is to know how much noise there is in an image.

All you are doing is enlarging the noise in the R 50% more and then saying ’look there is more noise’, that is fallacious. What people want to know is how much noise will there be in an 8x10 or 12x18 print or on a full screen image. Of course they also want to know how much detail there will be and how much DR etc etc, but when you are discussing noise you have to compare noise per image or noise per sensor area, noise per pixel is a nonsense.

 

[highdesertmesa]

For somebody who throws around such a strong term as pseudo science you sure have a lot of flaws in your own testing method.

The first one obviously being that you compare shots from different resolutions (30 MP in the R, 45 MP on the R5) at 1:1 pixel ratio.

But your scene is also so different. It appears to me much darker, which is supported by your statements about exposure and even pushing the shot. Noise isn't caused by high ISO values, it is caused by low light, among other things. We don't know the settings or ambient temperature from your examples. I am not saying they definitely matter here. But a long exposure in warm conditions would also affect noise.

All fine for me if that's how you want to evaluate the things you may or may not buy. But it is hard to tell much from the example. So when you refuse to accept demonstration based on documented measurements (that you yourself can repeat if you doubt the results) and studio image comparisons that actually change only one variable between samples, and call that pseudoscience, don't be surprised to get pushback.

What I was calling pseudoscience were the recent comments about how we’ve reaching a physical limit for ISO improvement, which completely discounts new sensor design (R5) and processor and firmware improvements. Those were statements made with even less proof since no one has a future camera in front of them to test it.

I’ll post a shot later that I took yesterday with the R in brighter light, ISO 4000, f11 on the RF 70-200. I haven’t looked at it in C1 yet, but zooming in on it in camera, it didn’t look anywhere near the R5 ISO 4000 Fro shot as far as noise control.

 

[highdesertmesa]

If you are comparing, that is making a valid comparison, the sensor noise you have to view at same size, the size is irrelevant but the size has to be the same. Comparing per pixel noise is basically irrelevant in this instance as the objective is to know how much noise there is in an image.

All you are doing is enlarging the noise in the R 50% more and then saying ’look there is more noise’, that is fallacious. What people want to know is how much noise will there be in an 8x10 or 12x18 print or on a full screen image. Of course they also want to know how much detail there will be and how much DR etc etc, but when you are discussing noise you have to compare noise per image or noise per sensor area, noise per pixel is a nonsense.

Thanks for proving me right about jumping to the equivalency argument when the direct 1:1 comparison fails. So then when we get to compare an ISO 6400 shot from the R5 sized down to 20mp against the 1DX3, will you accept that as several stops of improvement if that’s what the results show? Just want to make sure we have this straight before the R5 DPR test comes out.

But back to equivalency, I understand the concept of what you’re saying, but I’m making the argument that it’s an impractical comparison for those of us changing bodies. If I changed from the R to the 1DX3, would I view all my images at 125% and would I expect to print larger and keep the same IQ? No. Same goes for moving from the R to the R5. I’m still going to be reviewing my images at 1:1, and I’m going to be putting the higher resolution to good use by cropping more often. So for practical upgrade purposes, comparing sensors at 1:1 makes more sense.

But if you insist on equivalency, then try comparing a 1DX3 shot and R5 shot at ISO 6400 printed to 40x30”. But that’s not the favored type of equivalency, is it? You’d rather size down the larger file and print within the range of the lower resolution camera. That’s only 1/2 of of the story unless you also compare them enlarged to the same size.

 

[privatebydesign]

I have been entirely consistent in my assertions, you have wibbled and wobbled all over the place.

Do you agree that in your ‘comparison’ you are enlarging the R noise 50% more than the R6, and if so what are you trying to illustrate by doing that? It is meaningless and it is not a comparison. If you don’t agree that we have nothing else to talk about because your comprehension of reality is too far off base.

You are the one that started all this by saying you believed Canon had come up with technology that would improve high iso sensor performance by two stops at least, I said that is unlikely, you took issue with that. Now you have changed what most people would think of as high iso to much lower figures, you have come up with this curious per pixel noise idea etc etc.

Let me be clear, from your comments I believe you are wrong if I can’t see at least two stops of noise/DR improvement in high iso images (not pixels, but considering the R5 has a lot more pixels than the R you’d lose on a per pixel basis too but that is a side issue). I believe most people would consider 6,400 and above high iso. To make a fair and relevant comparison both R and R5 images have to be viewed at the same size because then the noise is enlarged the same.

 

[highdesertmesa]

I have been entirely consistent in my assertions, you have wibbled and wobbled all over the place.

Do you agree that in your ‘comparison’ you are enlarging the R noise 50% more than the R6,

You were the one that posted the DPR test side-by-side of the R and R6, only for some strange reason you choose ISO 1600 instead of 6400 or higher like I did. Not sure where you're going with this.

and if so what are you trying to illustrate by doing that? It is meaningless and it is not a comparison. If you don’t agree that we have nothing else to talk about because your comprehension of reality is too far off base.

You are the one that started all this by saying you believed Canon had come up with technology that would improve high iso sensor performance by two stops at least, I said that is unlikely, you took issue with that. Now you have changed what most people would think of as high iso to much lower figures, you have come up with this curious per pixel noise idea etc etc.

For the upgrade path of R to R5, yes. Tests are not out on that. I don't give a flying rat's behind about the R6 improvements over the 1DX3 – that was you bringing that in. I didn't specify which camera model, so why do you insist on using that as an example of me being wrong? And yet you seem to think I can't compare R to R5 to prove my point when the time comes. You're the one flopping around here.

Let me be clear,

That would be a welcome change.

from your comments I believe you are wrong if I can’t see at least two stops of noise/DR improvement in high iso images (not pixels, but considering the R5 has a lot more pixels than the R you’d lose on a per pixel basis too but that is a side issue). I believe most people would consider 6,400 and above high iso. To make a fair and relevant comparison both R and R5 images have to be viewed at the same size because then the noise is enlarged the same.

Let's see how the images from the R5 stack up agains the R. And if we use your measure of equivalency, we're going to be looking at the R5 sized down to 30mp, then putting that up against the R. If I'm wrong on that, then so be it. Big damn deal, lol.

I believe most people would consider 6,400 and above high iso.

High ISO to me is 640 and up since I mainly shoot 100-400. But you were the one that posted the DPR comparison of R versus R6 using ISO 1600 as an example – because of course they look more similar at that ISO. You cherry picking again. I posted the R versus R6 at ISO 6400 just to show there was a meaningful difference – maybe 1/2 a stop, but I was just pointing out that the R6 does look a little better than the R at that ISO. But that's not me trying to prove my original point about 2 stops – that's just me swatting at your random posts like the buzzing swarm of mosquitoes that they are.

 

[AaronT]

Here is Fro's R5 photo converted in Capture One. The second is a major crop from it. For an underexposed 4000 ISO shot it looks pretty good to me. Also, at F11 diffraction is kicking in. My 5DSR definitely wouldn't look as good under those conditions.

 

[privatebydesign]

And if we use your measure of equivalency,

It's not my measure of equivalency, it is equivalent by definition!

Equal in value, amount [size]...



I keep changing metrics and iso values to try to keep up with your changes, nothing more. But 640 iso is not a high iso value by any definition.

I don't have a problem with you being right or wrong, nor myself, heck I have been wrong here before and I have admitted it and apologized. My problem is you have picked on people, including me, across various threads without being provoked and kept repeating utter nonsense despite the fact that you have been shown relevant information to the contrary.

 

[privatebydesign]

You were the one that posted the DPR test side-by-side of the R and R6, only for some strange reason you choose ISO 1600 instead of 6400 or higher like I did. Not sure where you're going with this.

When I did it I showed it with the 'print' option selected which normalizes the output, I was consistent.

For the upgrade path of R to R5, yes. Tests are not out on that. I don't give a flying rat's behind about the R6 improvements over the 1DX3 – that was you bringing that in. I didn't specify which camera model, so why do you insist on using that as an example of me being wrong? And yet you seem to think I can't compare R to R5 to prove my point when the time comes. You're the one flopping around here.

The point was I have a 1DX II and you have an R, I was trying to find a spot where we could compare our personal experiences, so that was what I used initially. The point of showing the 1DX III DR was that is shares the sensor with the R6 so raw DR is going to be VERY similar, it wasn't about the 1DX III, it was about the fact the sensor output of the R6 is already known to a very large degree.

 

[Kit Lens Jockey]

Did anyone notice the performance of the burst shooting with the electronic shutter on the R5? There wasn't much movement during the burst shooting, but the feed through the camera looked a little smoother than it would be on the EOS R. It's hard to say really, but it looks pretty good.

 

[highdesertmesa]

It's not my measure of equivalency, it is equivalent by definition!

Equal in value, amount [size]...

I keep changing metrics and iso values to try to keep up with your changes, nothing more. But 640 iso is not a high iso value by any definition.

I don't have a problem with you being right or wrong, nor myself, heck I have been wrong here before and I have admitted it and apologized. My problem is you have picked on people, including me, across various threads without being provoked and kept repeating utter nonsense despite the fact that you have been shown relevant information to the contrary.

I guess my point was I always read people who make the equivalency argument say to size down the large file to match the smaller one, which seems strange to me. Why not blow up the smaller file to match the larger one, especially if the end goal is to print large? If the end goal is always going to be a 14"x11" print, then we're talking about being limited by the noise the human eye can resolve at a given distance as being the limit for high ISO performance. But if the end goal is say a 40"x30" print that we can walk up to, then we can see more differences, if there are any.

When I did it I showed it with the 'print' option selected which normalizes the output, I was consistent.

My bad. I was looking on my mobile phone and just saw one compared against the other. So when I switch to the print option, I still see a difference at 6400 and 12K, both with daylight and lowlight simulations – R6 looks better. About 1 stop improvement from R to R6.

The point was I have a 1DX II and you have an R, I was trying to find a spot where we could compare our personal experiences, so that was what I used initially. The point of showing the 1DX III DR was that is shares the sensor with the R6 so raw DR is going to be VERY similar, it wasn't about the 1DX III, it was about the fact the sensor output of the R6 is already known to a very large degree.

Fair enough. And I'm coming from the perspective of hoping (praying maybe, lol) to see 2 stops improvement from the R to the R5. We'll have to see. And I think you were trying to say there's no way we'll see 2 stops from 1DX3 to R6.

And speaking of being wrong and admitting it, I was thinking we'd get dual gain tech on the R5, which clearly we didn't.