Dpreview of the 80D

[rishi_sanyal]

Yes you can lift shadows four stops on the 80D but six stops on the Sonikon whatever.

That is something that is bemusing me at the moment. More and more reviews/comments talk about things like 'ah, but with the XXX camera you can lift shadows 6 stops with no noise but with YYY it is only 3 stops'.
My questions are:

- why would you ever want to (other than rectifying a complete penis-up at the time of the picture)
- if it is not a case of 'wanting to' then what does that comparison tell you about photography in the real world ?
- If it does tell you something, how often is it a real-world advantage ?

Or is it another case of desperately trying to find a difference so they have something to report on?

I used to think like that, until I started looking into it.

Here is my understanding.

Let's say you are shooting a basketball game. You know you need to shoot at 1/800 of a second to stop any action. Your lens has a maximum aperture of f2.8. The traditional approach is to ratchet up the ISO to get to your minimum required shutter speed and aperture. But, when you do so, you know you are introducing noise into the image and eventually the noise will make the image unusable.

With these sensors, the idea is that you can set your shutter speed and aperture but instead of changing the ISO, you leave it where it is and just underexpose. Then, in post, you can raise the exposure with less noise than if you'd increased the ISO.

Now, that does sound brilliant and very valuable.

It is also why I'd like to see some comparisons between the two methods to see if the promised advantages really exist or not and if they are sufficient to make a difference in the real world. If it does work, it really would change the way we shoot and process our images.

Almost. It's a common misconception that it's the ISO amplification that adds noise, but the reverse is true: ISO amplification actually helps reduce the noise that might result if you push in post. It's the decreased light levels available to the sensor when you deprive it of light (in low light, using high shutter speeds, e.g.) that lead to the higher noise levels, because of shot noise (http://bit.ly/shotnoise).

The advantage of ISO-invariant cameras is exactly as you describe: you can keep your focal plane exposure, determined by shutter speed and aperture, as required, but instead of using the high ISO required for a 'proper' exposure, dial the ISO right back down instead.

The advantage of doing this isn't that brightening in-post leads to lower noise levels than brightening in-camera, it's that brighter tones aren't amplified by the camera and clipped right out of the Raw file. In other words, decreasing your ISO setting by -X EV affords you X EV more highlight range. For an ISO-invariant camera, that comes at no additional noise cost in the shadows/midtones that you have to selectively brighten later (compared to if the camera had brightened it via ISO amplification).

So, in the end, it's about saving highlights under high ISO conditions. And the ability to do so all comes down ISO-invariance, which is a direct result of low read noise, which is correlated with high base ISO dynamic range. Note - we're not ever encouraging one to underexpose indiscriminately. You only need to underexpose enough to keep highlights from clipping. But under high contrast scenes, and with the limited dynamic range high ISOs demonstrate (because of the shrinking effective full-well capacity from the amplification), those extra +X EV of highlights can mean the difference between blown and preserved highlights.

The information you're looking for - the noise cost to doing this amplification in-post vs. in-camera, is exactly what the ISO-invariance test I implemented shortly after joining DPReview is designed to test. We've been doign it for over a year now, so you should have a number of cameras you can compare. Here's a link to the widget: http://bit.ly/1QOqxHg

What's funny is that in the time since I developed that test, we got dual-gain (Aptina) architecture built-in to Sony cameras like the a7R, a7S II, a7S, a6300, etc. These cameras have a particular ISO where they switch the amplification at the pixel-level, which helps to decrease noise even if the camera itself doesn't have much downstream read noise (our typical metric for an ISO-invariant camera: less downstream read noise means less benefit to in-camera ISO amplification vs. brightening in-post). This is why the a7R II doesn't appear ISO-invariant in our test (that link I provided above), but that doesn't mean it has poor base ISO dynamic range. It just means that the in-pixel amplification switch at ISO 640 helps eke out that last bit of extra signal by elevating it very early on in the signal processing (to make it more immune to noise). It's like ISO amplification just earlier on up in the imaging chain.

Apologies if that was a poor explanation - it's a little hard to wrap your head around at first! But it's the reason cameras like the a6300 are leading in terms of high ISO performance, and why the 42MP a7R II shows almost similar high ISO performance to even the Nikon D5 at high ISOs when normalized to equivalent size. The lack of dual-gain architecture is part of the reason the 80D falls well behind the a6300 at high ISO, though there must be another component, since it even falls behind the D7200 (the D7200 achieves a6300-esque high ISO performance without the dual-gain architecture, which is probably due to lower upstream read noise). We'll have 1DX II results up soon - we're curious to see how it performs. But understand that at this point, we're talking 1/3EV or less increases in performance at best. And after a certain point, there won't be any more to gain from decreasing noise, because we already have such low levels of noise to begin with (input-referred read noise is down to like 1 electron or less for some of these sensors - effectively nil). The only way we'll see ISO performance increase in the future is via increases in actual efficiency (or sensor size, of course), because at this point we're limited by how much light you're capturing.

That's why the D5 and D500 barely showed any gains in high ISO performance.

Here's a little more info on the a7R II's ISO-invariance; in the footnotes, I also explain this idea of dual-gain architecture: http://www.dpreview.com/articles/7450523388/sony-alpha-7r-ii-real-world-iso-invariance-study.

Have a read of our read noise treatise here to get a better idea of what 'downstream read noise' is: http://bit.ly/readnoise.

Also, thanks jebrady03 - that's appreciated.

 

[3kramd5]

Perhaps the biggest potential advantage is that if you underexpose in camera and lift in post, that push can be local (versus an ISO setting which is global). It may facilitate finer gradations within the highlights.

Would drastically underexposing then locally pushing in post be advantageous over properly exposing (by which I mean not blowing highlights you want), then locally pulling in post?

I would not think there to be a difference between ETTR+pull and ETTL+push, provided you don't blow the highlights (which is substantially less likely with ETTL).

Either way is too much work for me. I meter to the scene.

 

[Loswr]

Perhaps the biggest potential advantage is that if you underexpose in camera and lift in post, that push can be local (versus an ISO setting which is global). It may facilitate finer gradations within the highlights.

Would drastically underexposing then locally pushing in post be advantageous over properly exposing (by which I mean not blowing highlights you want), then locally pulling in post?

I would not think there to be a difference between ETTR+pull and ETTL+push, provided you don't blow the highlights (which is substantially less likely with ETTL).

Either way is too much work for me. I meter to the scene.

Pragmatism – how refreshing!

 

[rishi_sanyal]

There is no excuse for 80D to perform worst in focusing test through view finder which is a core DSLR strength.
I would like to see another source (Michael) to do the test and confirm this. If it is really the case with 80d, then no excuse of Canon. It is worst DSLR camera Canon made ( 1 frame in focus out of 16 frames). I remember DPR review on d7100 and 70d, Where they went with similar tangent like d7100 being class leading in AF test and 70d is not up to the task (not those blurry pics issue with center point). Then Michael did a epic comparison and showed us 70d actually did better than d7100 in AF test ;D ;D.

Checking out the summary of review on main page, feels like it is an advertisement for a6300. Don't get me wrong, I am also liking what I see from a6300. As and when Sony updates their kit lens (16-50 is actually very wide for kit lens) and throw in some cheap glass (looking at sigma to do those 17-50/70 lens), I am going switch to Sony. But that summary page is written by Sony marketing department.

Come on, it is not completely inferior to the a6300 on paper except for 4k video. 80d live view performance is good. 80d can focus better in low light in live view compared to a6300 ;D ;D. Regarding D7200 and 80D are both starting to look a bit old fashioned compared to to the current crop of 4K-capable mirrorless APS-C cameras (like the Sony a6300), who else is doing 4k other than Sony a6300??

d7200 is almost same as a6300 and d500 in terms of very high Iso performance. 80d is definitely not up to a6300 at very high iso performance. 80d is actually worst in high iso performance during video. In terms of video, Canon did worst with 80d.

Our AF tests are always done in triplicate, to rule out a one-off error. We used the same 70-200 F2.8L IS II that we use in all our Canon body tests, a lens that has performed just fine with the 5DS R, for example.

Also, I think you're looking at the rollover that conflates both subject tracking and depth-tracking. The single-point AF-C results showed approximately 50% of the frames in focus. Which is still a low hit-rate for a Canon DSLR in terms of AI Servo single-point.

So we're just as concerned as you are. We've just gotten a 2nd 80D in the office and will be repeating the tests to see if it's a copy thing, which'd be odd. We usually only test one body but given the odd result, we'll be following up.

Cheers,
Rishi

 

[Loswr]

Also, I think you're looking at the rollover that conflates both subject tracking and depth-tracking. The single-point AF-C results showed approximately 50% of the frames in focus. Which is still a low hit-rate for a Canon DSLR in terms of AI Servo single-point.

Thus, the earlier comment that DPR reads like the reviewers don't actually use Canon gear routinely, but rather use Canon gear solely for the purpose of reviewing it, is completely borne out. DPR natively speaks Nikon, I wonder what else is 'lost in translation'?

 

[rishi_sanyal]

Also, I think you're looking at the rollover that conflates both subject tracking and depth-tracking. The single-point AF-C results showed approximately 50% of the frames in focus. Which is still a low hit-rate for a Canon DSLR in terms of AI Servo single-point.

Thus, the earlier comment that DPR reads like the reviewers don't actually use Canon gear routinely, but rather use Canon gear solely for the purpose of reviewing it, is completely borne out. DPR natively speaks Nikon, I wonder what else is 'lost in translation'?

Except that AF-C, or some variant thereof, is the terminology used by Nikon, Sony, Pentax, Panasonic, and Olympus. Just not Canon.

Because we write reviews for everyone, not just Canon users, we tend to use the term 'continuous AF' or 'AF-C'. This is actually something we've internally discussed.

I even did a 50/50, using both 'AF-C' and 'AI Servo' in my post. Contrary to your assertion, I don't see anything 'borne out' other than an attempt to be as unbiased as possible.

 

[3kramd5]

Perhaps the biggest potential advantage is that if you underexpose in camera and lift in post, that push can be local (versus an ISO setting which is global). It may facilitate finer gradations within the highlights.

Would drastically underexposing then locally pushing in post be advantageous over properly exposing (by which I mean not blowing highlights you want), then locally pulling in post?

I would not think there to be a difference between ETTR+pull and ETTL+push, provided you don't blow the highlights (which is substantially less likely with ETTL).

Either way is too much work for me. I meter to the scene.

Rishi - has your experimentation been sufficient to answer the question?

Is there a difference between shooting -X ISO and pushing it back up X, and metering to the tones you want to preserve and pulling back down (other than direction and magnitude in post)?

 

[3kramd5]

...
Given all the effort expended to expand ISO ranges up, but not down, it seems very likely they are meant to perform well, if not best, at high ISO, even if physics preclude them from for example having their widest DR (which is again only one measure of performance) when set that way.

Improving IQ at high ISO requires IQ at every ISO between base ISO and the higher ISO ranges to be improved.

Or in other words, improvement at high ISO comes as a result of improving the IQ of the camera sensor generally.

Imagine you have a line that starts at (100,2) and finishes at (3200,12.) The values for y at x=400 and x=800 will fall on that line somewhere and will between 2 & 12. If the line goes from (100,2) to (3200,10) then the values for y at x=400 and x=800 will also be lower.

It seems that most of the gains (no pun intended) at high ISO are coming from NR and tech like aptina's DR-PIX, not merely lowering RN.

In any case, you are avoiding the question, which was if you can support the assertion that cameras are meant to work better at base ISO, rather than merely explain why they do as a consequence of the single measure of performance you're considering.

 

[Loswr]

Also, I think you're looking at the rollover that conflates both subject tracking and depth-tracking. The single-point AF-C results showed approximately 50% of the frames in focus. Which is still a low hit-rate for a Canon DSLR in terms of AI Servo single-point.

Thus, the earlier comment that DPR reads like the reviewers don't actually use Canon gear routinely, but rather use Canon gear solely for the purpose of reviewing it, is completely borne out. DPR natively speaks Nikon, I wonder what else is 'lost in translation'?

Except that AF-C, or some variant thereof, is the terminology used by Nikon, Sony, Pentax, Panasonic, and Olympus. Just not Canon.

Because we write reviews for everyone, not just Canon users, we tend to use the term 'continuous AF' or 'AF-C'. This is actually something we've internally discussed.

I even did a 50/50, using both 'AF-C' and 'AI Servo' in my post. Contrary to your assertion, I don't see anything 'borne out' other than an attempt to be as unbiased as possible.

When discussing the performance of a Canon dSLR, using the term AF-C to refer to a particular focus mode is simply incorrect. So you're '50/50' is half correct, and half wrong. I could also point out that you are posting on a Canon forum, where I suspect many people do not have a clue what AF-C means.

Many of my friends and colleagues speak German, so in an effort to appear linguistically unbiased, I shall refer to DPR as voreingenommen.

 

[3kramd5]

...
Given all the effort expended to expand ISO ranges up, but not down, it seems very likely they are meant to perform well, if not best, at high ISO, even if physics preclude them from for example having their widest DR (which is again only one measure of performance) when set that way.

Improving IQ at high ISO requires IQ at every ISO between base ISO and the higher ISO ranges to be improved.

Or in other words, improvement at high ISO comes as a result of improving the IQ of the camera sensor generally.

Imagine you have a line that starts at (100,2) and finishes at (3200,12.) The values for y at x=400 and x=800 will fall on that line somewhere and will between 2 & 12. If the line goes from (100,2) to (3200,10) then the values for y at x=400 and x=800 will also be lower.

It seems that most of the gains (no pun intended) at high ISO are coming from NR and tech like aptina's DR-PIX, not merely lowering RN.

In any case, you are avoiding the question, which was if you can support the assertion that cameras are meant to work better at base ISO, rather than merely explain why they do as a consequence of the single measure of performance you're considering.

What is "work better" to you? Lets understand what that should be before arguing about "better".

"Work" better should have been "perform" better per the first quoted post, but I'll use them interchangeably: facilitate a desired result.

So, say my desired result is a properly exposed image out of camera in low light and a stop motion shutter speed for a fast subject; clearly high ISO will perform better than low ISO. If on the other hand my desired result is to capture the greatest possible dynamic range out of camera, low ISO will perform better than high ISO.

By it's not my yardstick which is at question, it's the engineering team's yardstick, since they are the ones with the design intent ("meant to"). I bet a lot of them would say high ISO is their performance priority.

 

[brad-man]

Also, I think you're looking at the rollover that conflates both subject tracking and depth-tracking. The single-point AF-C results showed approximately 50% of the frames in focus. Which is still a low hit-rate for a Canon DSLR in terms of AI Servo single-point.

Thus, the earlier comment that DPR reads like the reviewers don't actually use Canon gear routinely, but rather use Canon gear solely for the purpose of reviewing it, is completely borne out. DPR natively speaks Nikon, I wonder what else is 'lost in translation'?

Except that AF-C, or some variant thereof, is the terminology used by Nikon, Sony, Pentax, Panasonic, and Olympus. Just not Canon.

Because we write reviews for everyone, not just Canon users, we tend to use the term 'continuous AF' or 'AF-C'. This is actually something we've internally discussed.

I even did a 50/50, using both 'AF-C' and 'AI Servo' in my post. Contrary to your assertion, I don't see anything 'borne out' other than an attempt to be as unbiased as possible.

When discussing the performance of a Canon dSLR, using the term AF-C to refer to a particular focus mode is simply incorrect. So you're '50/50' is half correct, and half wrong. I could also point out that you are posting on a Canon forum, where I suspect many people do not have a clue what AF-C means.

Many of my friends and colleagues speak German, so in an effort to appear linguistically unbiased, I shall refer to DPR as voreingenommen.

Whoever is without bias among you, let him be the first to cast a stone...

 

[3kramd5]

....
By it's not my yardstick which is at question, it's the engineering team's yardstick, since they are the ones with the design intent ("meant to"). I bet a lot of them would say high ISO is their performance priority.

We can argue about that forever and a day because neither of us are engineers on said teams (to provide a definitive answer) so all that we can do is present different perspectives on observed outcomes.

Eureka! That's why I phrased my initial question how I did.

 

[bdunbar79]

Also, I think you're looking at the rollover that conflates both subject tracking and depth-tracking. The single-point AF-C results showed approximately 50% of the frames in focus. Which is still a low hit-rate for a Canon DSLR in terms of AI Servo single-point.

Thus, the earlier comment that DPR reads like the reviewers don't actually use Canon gear routinely, but rather use Canon gear solely for the purpose of reviewing it, is completely borne out. DPR natively speaks Nikon, I wonder what else is 'lost in translation'?

Except that AF-C, or some variant thereof, is the terminology used by Nikon, Sony, Pentax, Panasonic, and Olympus. Just not Canon.

Because we write reviews for everyone, not just Canon users, we tend to use the term 'continuous AF' or 'AF-C'. This is actually something we've internally discussed.

I even did a 50/50, using both 'AF-C' and 'AI Servo' in my post. Contrary to your assertion, I don't see anything 'borne out' other than an attempt to be as unbiased as possible.

When discussing the performance of a Canon dSLR, using the term AF-C to refer to a particular focus mode is simply incorrect. So you're '50/50' is half correct, and half wrong. I could also point out that you are posting on a Canon forum, where I suspect many people do not have a clue what AF-C means.

Many of my friends and colleagues speak German, so in an effort to appear linguistically unbiased, I shall refer to DPR as voreingenommen.

Whoever is without bias among you, let him be the first to cast a stone...

Problem with that is neuro's not running a big website claiming absolute scientific testing. I've stayed silent so far but I can't take anymore. I've worked as a physicist and as a chemist and this garbage is far from scientific and it is clearly biased. That's the only scientific aspect of the testing. Some of it was even done incorrectly or seemingly in a way to obtain predetermined desired results. Even if that weren't the intent, if a third party feels that way or gets that impression, then there is something wrong. Personally I ignore the BS and let the camera performance in real life do the talking. That's why I have 2 1Dx's and my AF never misses.

 

[Loswr]

Also, I think you're looking at the rollover that conflates both subject tracking and depth-tracking. The single-point AF-C results showed approximately 50% of the frames in focus. Which is still a low hit-rate for a Canon DSLR in terms of AI Servo single-point.

Thus, the earlier comment that DPR reads like the reviewers don't actually use Canon gear routinely, but rather use Canon gear solely for the purpose of reviewing it, is completely borne out. DPR natively speaks Nikon, I wonder what else is 'lost in translation'?

Except that AF-C, or some variant thereof, is the terminology used by Nikon, Sony, Pentax, Panasonic, and Olympus. Just not Canon.

Because we write reviews for everyone, not just Canon users, we tend to use the term 'continuous AF' or 'AF-C'. This is actually something we've internally discussed.

I even did a 50/50, using both 'AF-C' and 'AI Servo' in my post. Contrary to your assertion, I don't see anything 'borne out' other than an attempt to be as unbiased as possible.

When discussing the performance of a Canon dSLR, using the term AF-C to refer to a particular focus mode is simply incorrect. So you're '50/50' is half correct, and half wrong. I could also point out that you are posting on a Canon forum, where I suspect many people do not have a clue what AF-C means.

Many of my friends and colleagues speak German, so in an effort to appear linguistically unbiased, I shall refer to DPR as voreingenommen.

Whoever is without bias among you, let him be the first to cast a stone...

Whosoever claimeth to be without bias, yet have proved that claim false, yea even before utterance, shall be condemned as a biaser and painted with a yellow N, that all may know and shunneth him. Verily.

 

[brad-man]

Also, I think you're looking at the rollover that conflates both subject tracking and depth-tracking. The single-point AF-C results showed approximately 50% of the frames in focus. Which is still a low hit-rate for a Canon DSLR in terms of AI Servo single-point.

Thus, the earlier comment that DPR reads like the reviewers don't actually use Canon gear routinely, but rather use Canon gear solely for the purpose of reviewing it, is completely borne out. DPR natively speaks Nikon, I wonder what else is 'lost in translation'?

Except that AF-C, or some variant thereof, is the terminology used by Nikon, Sony, Pentax, Panasonic, and Olympus. Just not Canon.

Because we write reviews for everyone, not just Canon users, we tend to use the term 'continuous AF' or 'AF-C'. This is actually something we've internally discussed.

I even did a 50/50, using both 'AF-C' and 'AI Servo' in my post. Contrary to your assertion, I don't see anything 'borne out' other than an attempt to be as unbiased as possible.

When discussing the performance of a Canon dSLR, using the term AF-C to refer to a particular focus mode is simply incorrect. So you're '50/50' is half correct, and half wrong. I could also point out that you are posting on a Canon forum, where I suspect many people do not have a clue what AF-C means.

Many of my friends and colleagues speak German, so in an effort to appear linguistically unbiased, I shall refer to DPR as voreingenommen.

Whoever is without bias among you, let him be the first to cast a stone...

Whosoever claimeth to be without bias, yet have proved that claim false, yea even before utterance, shall be condemned as a biaser and painted with a yellow N, that all may know and shunneth him. Verily.

OK. I just about shot my wad on verses. I'm afraid the "yellow N" reference went right over my head. Verily. Nice retort though.

 

[Loswr]

OK. I just about shot my wad on verses. I'm afraid the "yellow N" reference went right over my head. Verily. Nice retort though.

A play on Nathaniel Hawthorne. In retrospect, painted with a scarlet-orange alpha may have been a more obvious reference, albeit less fitting.

 

[Mr. Low Notes]

It looks like in the review the shooter doesn't get an initial lock...so the camera and lens are then struggling to re-adjust. This shouts of a poor technique from the reviewer...he's only a press hack I guess. To be honest...I steer clear of DPR....the forum isn't a nice place to hang out, there's a lot of weird and frankly ruthless attitudes there and their reviews are just glorified eye candy. There's nothing in those reviews which do much for me..except maybe comparing iso noise between similar cameras. The original owner was very biased towards Canon and very open about his opinion. I think you'll find there's very few 1D series reviews, certainly no 1DX but you'll find every Nikon Dx series reviewed...all with glowing reviews...like the D4, which they failed to mention the numerous lockups and AF issues which plagued every pro photographer that I met using one. Which is why so many pros got frustrated with Noik and traded in for Canon and the (working) 1DX.
[/quote]

I now refer to DP Review as DR Review since that is all they concentrate on. I'm surprised they haven't banned me yet because they don't like me saying that. I'm with you. I'm about done with DR Review. It's a nasty place in the comments section and getting worse. I do admit to throwing my two cents in but I try to be positive and not nasty. No matter. Negativity is king there as is Canon hate. Sony shooters seem to have a chip on their shoulder and Canon shooters a bulls eye on their back. I use a 60D, 70D and EOS M and I'm happy with them as are my customers with my results and that is what matters.

As for the 80D, it looks to be a fine camera. As for Sony I have nothing against them. Glad they are putting pressure on the other companies to improve. Competition is good. Being overly biased is not.

 

[rishi_sanyal]

It looks like in the review the shooter doesn't get an initial lock...so the camera and lens are then struggling to re-adjust. This shouts of a poor technique from the reviewer...he's only a press hack I guess.

Leaving aside the fact that that's simply insulting to our reviewer (not me), we repeat every test at least 3x. We always initiate focus on the biker while he's static, then ask him to start coming towards the camera. If it didn't get an 'initial lock' on 3 tries on a static biker, you might say there's a problem. And that was with single-point 'One Shot'.

I'm an uncertain of how that is an 'Idiot Method'. Also, claiming intended bias here would require every single one of our reviewers happens to be biased, or is forced to have a brand-specific bias by management. Readers of this site (and ours) should ask themselves how likely that would be.

To be honest...I steer clear of DPR....the forum isn't a nice place to hang out, there's a lot of weird and frankly ruthless attitudes there and their reviews are just glorified eye candy.

If tireless testing and re-testing and thoughts of 'did I miss anything in our tests?' that keep us up at night is 'glorified eye candy', then, yes, our controlled side-by-sides and meticulously controlled tests are really just 'glorified eye candy'.

Again, this is very insulting considering that in reality, we're constantly refining our methodologies - we now carefully control all lighting and exposure down to < 1/3 EV error to make sure our high ISO comparisons correlate with real-world performance, for example. Why would we do that if we were just in this for the 'eye candy'? I'm not actually even sure what that means - long wordy reviews aren't really eye-candy to me.

There's nothing in those reviews which do much for me..except maybe comparing iso noise between similar cameras. The original owner was very biased towards Canon and very open about his opinion. I think you'll find there's very few 1D series reviews, certainly no 1DX but you'll find every Nikon Dx series reviewed...all with glowing reviews...like the D4, which they failed to mention the numerous lockups and AF issues which plagued every pro photographer that I met using one. Which is why so many pros got frustrated with Noik and traded in for Canon and the (working) 1DX.

Like the D4S, which we didn't review?

I now refer to DP Review as DR Review since that is all they concentrate on.

1 page of 16 page reviews is 'all they concentrate on'? 1/16 = 6.25%.

You could say the same about our focus on high ISO performance and analysis via our studio scene, JPEG noise reduction and detail retention, color (where we typically state that Canon is our 'benchmark'), and, pun intended, our focus on focus, which typically takes up multiple pages.

In fact, our body and menus and ergonomics pages take up more pages than our DR pages. Dynamic range is a but a small percentage of our final weighted score.

'All they concentrate on'?

I'm surprised they haven't banned me yet because they don't like me saying that. I'm with you. I'm about done with DR Review. It's a nasty place in the comments section and getting worse. I do admit to throwing my two cents in but I try to be positive and not nasty. No matter. Negativity is king there as is Canon hate.

There's just as much Nikon hate. Every time we publish a Nikon interview, I feel sorry for Nikon Japan if they're reading our site. Which they are.

You're right about one thing though - the negativity. Which is evidently not limited to DPR.

Sony shooters seem to have a chip on their shoulder and Canon shooters a bulls eye on their back. I use a 60D, 70D and EOS M and I'm happy with them as are my customers with my results and that is what matters.

That is not only what matters when you're running a review site that tries to cover all technologies and all cameras relative to one another. I'm happy that in isolation you're happy with your gear - that doesn't mean there isn't other gear that performs better along certain axes, and it's our job to find those things and talk about them.

-Rishi

 

[ritholtz]

There is no excuse for 80D to perform worst in focusing test through view finder which is a core DSLR strength.
I would like to see another source (Michael) to do the test and confirm this. If it is really the case with 80d, then no excuse of Canon. It is worst DSLR camera Canon made ( 1 frame in focus out of 16 frames). I remember DPR review on d7100 and 70d, Where they went with similar tangent like d7100 being class leading in AF test and 70d is not up to the task (not those blurry pics issue with center point). Then Michael did a epic comparison and showed us 70d actually did better than d7100 in AF test ;D ;D.

Checking out the summary of review on main page, feels like it is an advertisement for a6300. Don't get me wrong, I am also liking what I see from a6300. As and when Sony updates their kit lens (16-50 is actually very wide for kit lens) and throw in some cheap glass (looking at sigma to do those 17-50/70 lens), I am going switch to Sony. But that summary page is written by Sony marketing department.

Come on, it is not completely inferior to the a6300 on paper except for 4k video. 80d live view performance is good. 80d can focus better in low light in live view compared to a6300 ;D ;D. Regarding D7200 and 80D are both starting to look a bit old fashioned compared to to the current crop of 4K-capable mirrorless APS-C cameras (like the Sony a6300), who else is doing 4k other than Sony a6300??

d7200 is almost same as a6300 and d500 in terms of very high Iso performance. 80d is definitely not up to a6300 at very high iso performance. 80d is actually worst in high iso performance during video. In terms of video, Canon did worst with 80d.

Our AF tests are always done in triplicate, to rule out a one-off error. We used the same 70-200 F2.8L IS II that we use in all our Canon body tests, a lens that has performed just fine with the 5DS R, for example.

Also, I think you're looking at the rollover that conflates both subject tracking and depth-tracking. The single-point AF-C results showed approximately 50% of the frames in focus. Which is still a low hit-rate for a Canon DSLR in terms of AI Servo single-point.

So we're just as concerned as you are. We've just gotten a 2nd 80D in the office and will be repeating the tests to see if it's a copy thing, which'd be odd. We usually only test one body but given the odd result, we'll be following up.

Cheers,
Rishi

Thanks Rishi. I will wait for your retest and see how it does. If results are same, you should reduce final score a lot. 84% for a camera which can't even do simple servo focus tracking (no eye/face/color tracking) is very high.
So, AF-C is same as AI-Servo single point tracking in Canon. When you refer to subject tracking, it is iTR tracking and depth tracking is simple AI servo tracking right? Looks like there is no improvement for 80d in terms of iTR which is kinda strange.

Thanks

 

[rishi_sanyal]

Some of it was even done incorrectly or seemingly in a way to obtain predetermined desired results.

That's a bold claim - do you have actual evidence for it?

Speaking of evidence-based methods, we explain all our testing methodologies, then present you with the data, typically in the form of rollovers for AF tests, and images, including Raw files, when it comes to image quality. So you don't just have to take our word for it. If you ask us for our testing methodology, we explain it, and are working on even more thorough documentation so you know exactly our testing conditions.

How does this jive with 'done in a way to obtain predetermined desired results'? The only 'desired results' or bias are the biases of the tests themselves - which you are perfectly in your right to claim as 'biased'. Our studio tests are 'biased' towards better image quality, and our AF tests are 'biased' towards better AF performance.

If that's a bad thing, please feel free to choose a site that is more brand-specific as opposed to test-specific.

Even if that weren't the intent, if a third party feels that way or gets that impression, then there is something wrong.

There is something wrong: a 3rd party who hasn't verified or repeated the tests themselves somehow claiming that, absolutely, we're the ones who are wrong. Despite never testing what the competition offers himself side-by-side in an unbiased manner.

Any third party who hasn't tested all systems in question side-by-side labeling us as wrong or biased based on his/her own limited experience is suspect. You can't only test camera X in situation A, and be confident that our claim that camera Y in situation A or B is invalid because you're just sure that camera X is great in situation A.

Personally I ignore the BS and let the camera performance in real life do the talking. That's why I have 2 1Dx's and my AF never misses.

If you have two 1DX's that 'never' miss AF, then your ability to provide unbiased reviews should probably be in question. No offense intended.

Also, could you please explain how our 80D results have anything to do with your 1DX results? Keep in mind that our 5DS single-point continuous AF results were exactly as expected: nearly perfect. I suppose those positive results were acceptable, though, right?

Also, if you 'let the camera performance in real life do the talking', please show us your comparative results vs. other systems, or your results from your 1DX in every photographic scenario ever imaginable.

Else, your frame of reference is limited by your small sample size. You're taking your limited experience, and claiming we're false despite our testing a far larger sample set than you've tested. You're claiming our 80D results are flawed because of your 1DX results, the latter which we (1) didn't show results for or talk about in this review, and which (2) when we have talked about, have generally talked about its phenomenal hit-rate in continuous drive.

That's not just bad science, that's illogical.