Review: Canon EOS R3 by The Digital Picture

[Joules]

Another is to simply take multiple, successive exposures and combine them in Photoshop (overlay them and adjust the opacity to equally weight them (i.e. opacity = 100/image number, so the bottom image is 100% opacity, the second is 50% opacity, the 3rd is 33%, 4th is 25%, etc.).

Alternatively to changing the opacity by hand, you can also do this by selecting all your layers, turning them into a smart object and then setting the blend mode to average. Processing into a smart object usually takes quite a while though. But is does give some other useful blend modes like median or max which allow creating sort of synthetic long exposures.

 

[kaihp]

Yes, I think it's just a mistake. He must have meant to write R3 and he uses R5 in several instances.

That definitiety makes sense, but why is Bryan comparing to the 5DsR?

Time to fill out his contact form...

Edit: After searching the R5 review, I can see that the second sentence is copied verbatim from the R5 review. When reviewing the R5, comparing to the similar pixel-count 5DsR makes sense, but comparing the AF system of the R3 with the venerable 5DsR does not makes sense IMHO. Comparing to the 1DX series (Marks I, II or II) would make much more sense.

 

[entoman]

Yes, I think it's just a mistake. He must have meant to write R3 and he uses R5 in several instances.

For context, I own a R5 and a 5DMkiv, and have previously owned the 5DS.

5DS/5DSR have comparatively slow AF acquisition, especially in low light, and are both pretty hopeless at tracking. When using 100-400mm and 1.4x extender, which reduces maximum aperture to F8, only the central AF spot is active.

5DMkiv has very fast AF acquisition, even in low light, and does a reasonable job of tracking birds (and even butterflies) in flight. All AF points are active at F8.

R5 has very fast acquisition (marginally faster than 5DMkiv) but is much more accurate. It is extremely efficient at tracking faces and pretty good at tracking birds. The animal-eye AF is very sticky, even when subjects are quite small in the frame - it does sometimes jump onto another part of the bird/animal, but hardly ever jumps to the background.

Bryan at Digital-Picture knows Canon cameras inside out, so his comment was clearly a copy and paste error that got overlooked.

 

[kaihp]

Bryan at Digital-Picture knows Canon cameras inside out, so his comment was clearly a copy and paste error that got overlooked.

Looking over the R5 review and comparing it to the R3 review, I found that a very large amount of text and images were copy-pasta. Sure, the focus systems and the points he wants to make are similar, but he really shouldn't be showing R5 data (photos) to prove a point in an R3 review, when the R3 wasn't used for the particular test (e.g. eye and head tracking).

To me, this is starting to detract from the trustworthiness of the individual reviews.

 

[entoman]

Looking over the R5 review and comparing it to the R3 review, I found that a very large amount of text and images were copy-pasta. Sure, the focus systems and the points he wants to make are similar, but he really shouldn't be showing R5 data (photos) to prove a point in an R3 review, when the R3 wasn't used for the particular test (e.g. eye and head tracking).

To me, this is starting to detract from the trustworthiness of the individual reviews.

Yes I agree, it's very tempting to take short cuts and use copy/paste, then edit as necessary afterwards, but it can lead to silly mistakes and make the author look less reliable. Copy/paste is standard practice for authors writing multiple articles about similar subjects, but we need to be meticulous in proof-reading. It can also get a bit boring reading the same text, with just a few words switched.

 

[AlanF]

He has corrected the mistakes now. At the very least, his write ups are backed with evidence and useful data and genuine knowledge, unlike the garbage spewed out by so many so-called YouTube "experts". His reviews of the R3 and the R5 (and R6) don't bode well for the future of ovfs.

 

[kaihp]

Yes I agree, it's very tempting to take short cuts and use copy/paste, then edit as necessary afterwards, but it can lead to silly mistakes and make the author look less reliable.

Sir, we are in violent agreement

 

[kaihp]

He has corrected the mistakes now. At the very least, his write ups are backed with evidence and useful data and genuine knowledge, unlike the garbage spewed out by so many so-called YouTube "experts". His reviews of the R3 and the R5 (and R6) don't bode well for the future of ovfs.

Thanks Alan. Same paragraph now says:

The R3 focuses extremely fast. My perception is that the R3 focuses slightly faster than the already fast EOS R5 during side-by-side testing with the same lens. The R3's advanced AF system is suitable for nearly all pursuits.

Interesting that the R3 only focuses "slightly faster" than the R5, as I found there was a significant difference between the 5D3 and the 1DX bodies. Of course, if the R5 is already very quick, then it is going to be hard to improve, and the perception of the improvement can be minute.

Absolutely agree on Bryan' reviews being backed up with evidence and useful data (like the database of test chart shots with different lens/body combinations).

I haven't received a reply to my email question, though. But the primary point is that he has corrected the text on the page.

Edit: email from Bryan received

 

[AlanF]

Thanks Alan. Same paragraph now says:

Interesting that the R3 only focuses "slightly faster" than the R5, as I found there was a significant difference between the 5D3 and the 1DX bodies. Of course, if the R5 is already very quick, then it is going to be hard to improve, and the perception of the improvement can be minute.

Absolutely agree on Bryan' reviews being backed up with evidance and useful data (like the database of test chart shots with different lens/body combinations).

I haven't received a reply to my email question, though. But the primary point is that he has corrected the text on the page.

The only thing I don't like is his lens charts, which I'll repeat here to the boredom of some. His image quality are just a rough guide, but no more than that, and some take them too seriously.

 

[YuengLinger]

I remember reading Bryan's reviews about the first L lenses I bought, back in 2005. We've had a lot of water under the bridge since then, and I know I'm not getting any younger. Auto-pilot, phoning it in, cut-and-paste...They do become more tempting as time goes by...

 

[NKD]

Quality review & pics as usual
I just love the look of this new 1d style body, cant wait to hold one.
Hopefully one day own a copy of a newer unreleased camera with this body

 

[woodman411]

Looking over the R5 review and comparing it to the R3 review, I found that a very large amount of text and images were copy-pasta. Sure, the focus systems and the points he wants to make are similar, but he really shouldn't be showing R5 data (photos) to prove a point in an R3 review, when the R3 wasn't used for the particular test (e.g. eye and head tracking).

To me, this is starting to detract from the trustworthiness of the individual reviews.

The purpose of Bryan using some of the R5 autofocus content was to articulate the difference between modern Canon mirrorless to DSLR, and he does preface it by saying "Here is a sample showing that benefit, along with a bit of discussion from the R5 review. Know that the R3 is considerably more advanced – I'll share some of those advancements after this reversion." I have no issues with what he did here, and he does expand more on the differences between the R5 and R3 afterwards.

It is ironic you mention trustworthiness, to me, he is the most trustworthy online reviewer for Canon and Sony. He doesn't clickbait, he doesn't hype ("Canon lied! I'm switching! End of Canon!") or goes with the trendy-but-out-of-context stupidity like dynamic range doom/overheating doom/<insert latest Canon doom here>. His reviews are thorough, measured when possible, contextual, and ultimately the most informative of any reviewer out there. TDP should be considered one of the most valuable review resources available.

 

[docsmith]

His reviews are thorough, measured when possible, contextual, and ultimately the most informative of any reviewer out there. TDP should be considered a valuable Canon resource.

Also when being measured, he often hits the nail right on top of the head:

"With each new high-end camera iteration, it becomes more difficult to be happy with the newfound annoyances (additional steps or efforts required to get the same job done) of the older models, and the R3 brings out those inadequacies of the other models."

We have gone through the steps of camera development, where it is really now camera refinement. The advances are not so big. You really can still take great pictures with the xx year old camera you currently own.

But the R3, and other newer cameras....just makes doing so easier.

 

[bbasiaga]

Thanks Alan. Same paragraph now says:

Interesting that the R3 only focuses "slightly faster" than the R5, as I found there was a significant difference between the 5D3 and the 1DX bodies. Of course, if the R5 is already very quick, then it is going to be hard to improve, and the perception of the improvement can be minute.

Absolutely agree on Bryan' reviews being backed up with evidence and useful data (like the database of test chart shots with different lens/body combinations).

I haven't received a reply to my email question, though. But the primary point is that he has corrected the text on the page.

Edit: email from Bryan received

R5 and R6 got a lot closer to the 1D series AF speed than the 5D series ever got. So they closed a lot of the gap already. R3 is essentially just as fast as the 1D series, and is at least as fast as the Sony A1. Maybe an R1 will be faster, but right now it seems hard to imagine how things get better from here. I'm sure they will, but this is already uncharted territory.

Brian

 

[VegasCameraGuy]

This is a very useful, thorough review. However, still no one has done what I am most interested to see, which is an image quality comparison between the R3 and R5 where the R3 images are scaled up to R5 size. In this review, it is stated that a better comparison would be to scale the R5 images down to R3 size, and depending on how you work that may be most relevant to you, but for me I really want to know just how much quality I would be losing with the R3 vs the R5 when making a large print of a cropped image.

That's my point, I rarely use the full-frame and will sometimes crop an image to get a vertical out of a horizontal when I see that it's a better shot. Anytime you're cropping in the image resolution is a factor to maintaining a decent output. If it's a subject that you can't reshoot then you need as much flexibility as possible if you can't go back and reshoot the subject.

 

[VegasCameraGuy]

Also when being measured, he often hits the nail right on top of the head:

"With each new high-end camera iteration, it becomes more difficult to be happy with the newfound annoyances (additional steps or efforts required to get the same job done) of the older models, and the R3 brings out those inadequacies of the other models."

We have gone through the steps of camera development, where it is really now camera refinement. The advances are not so big. You really can still take great pictures with the xx year old camera you currently own.

But the R3, and other newer cameras....just makes doing so easier.

It will be interesting to see the sales comparison of the R3 to the R5 and tell if the resolution is a deciding factor for new camera purchasers or if they are willing to give away resolution to get the new features on the R3?

 

[neuroanatomist]

It will be interesting to see the sales comparison of the R3 to the R5 and tell if the resolution is a deciding factor for new camera purchasers or if they are willing to give away resolution to get the new features on the R3?

1) the >$2000 price difference is a confound that can’t be ignored, and 2) we won’t ever see those numbers anyway.

 

[kten]

I believe the idea is a single RAW file with the MSNR applied, i.e the multiple shots combined into a single RAW file with lower noise. There are dedicated astrophotography postprocessing apps that do that sort of thing (not on RAW files, though), but I'm not aware of any standard RAW converters that allow combining RAW images to reduce noise. The concept is similar to long exposure NR, where the camera takes a dark frame after the shot and subtracts it from the primary image before writing the RAW file.

Note that there are a couple of workarounds to achieve the same ends. The way to get a RAW file is to use Canon's multiple exposure feature and set the exposure control method to Average – that lets you take up to 9 exposures to combine (which would give better NR than the 4 exposures of MSNR). Another is to simply take multiple, successive exposures and combine them in Photoshop (overlay them and adjust the opacity to equally weight them (i.e. opacity = 100/image number, so the bottom image is 100% opacity, the second is 50% opacity, the 3rd is 33%, 4th is 25%, etc.).

When I know I'm going to be doing that for NR reasons I find it much quicker and easier to take multiple underexposed images where total exposure of the group adds up to same as correct exposure and use screen blend mode if that makes sense. Eg. when I've done astro or low light stuff where plan to stack for NR reasons instead of blending say 10 x 1 seconds shots with opacity method you mention I'd take 10 x 1/10th shots and blend with 100% opacity in screen.

Obviously I tend to do it for longer exposures than 1sec but it can be handy for shortish ones handheld with no stabilisation and simply crop a tiny bit after auto aligning layers with PS align layers script under pano and stackign options. Much faster and kills the same kind of random noise that opacity normal blend does with less time for things in scene to move too much (stars, trees etc etc).

 

[DBounce]

I’m starting to think that 30 fps is just marketing hype. The video below is actually shot in still mode. The stills are processed in Lightroom and batch exported. These stills were then imported into Resolve as an image sequence. However, when played back on a 30 fps timeline the video looked sped up. That means there were not enough frames to play at normal speed on the 30 fps timeline. Worst still, when played back at the video’s 24 fps frame rate, it still appears slightly sped up.
It looks like the R3 can only achieve about 22 fps, outside of paid influencers hands.

 

[neuroanatomist]

I’m starting to think that 30 fps is just marketing hype. The video below is actually shot in still mode. The stills are processed in Lightroom and batch exported. These stills were then imported into Resolve as an image sequence. However, when played back on a 30 fps timeline the video looked sped up. That means there were not enough frames to play at normal speed on the 30 fps timeline. Worst still, when played back at the video’s 24 fps frame rate, it still appears slightly sped up.
It looks like the R3 can only achieve about 22 fps, outside of paid influencers hands.

There are caveats on lens choice and camera settings that preclude achieving 30 fps. With no details on those parameters, it’s premature to draw that conclusion.

We know DPR doesn’t, but some reviewers actually RTFM.