Canon executives say a lot more coming in 2019

Mar 2, 2012
3,187
542
(nor would one expect it to be, given that it is essentially the same sensor).
I would expect it to be somewhat different, given that the sensor is packaged in different camera, and used differently (always reading).

But otherwise, totally agree with the inadequacies in dxo’s pseudodatascience.
 
  • Like
Reactions: 1 user
Upvote 0

Michael Clark

Now we see through a glass, darkly...
Apr 5, 2016
4,722
2,655
It doesn't have to cover most extreme scenes. Again better DR simply means more scenes to cover in one shot.



I'm not shooting with Sony, and even if I were, would you require me to shoot hundreds of images a year with both Canon and Sony side by side?



All those cameras, even the old ones, are 14 bit, but it doesn't mean they all have 14-stop DR. Lower DR simply means more noise all over the image, not just shadows. Of course the noise affects the shadows the most, but in older cameras you can see it in the light shadows and even mid-tones while the cameras with better DR push the noticeable noise to the deeper shadows. 3.5ev is actually a lot, try pushing exposure by 3.5ev and the shadows will turn into a mess in old cameras.

Try exposing properly to begin with and you don't need to worry about how far you can push EV in post.
 
  • Like
Reactions: 1 users
Upvote 0
Literally all? The EOS R is worse in terms of DR? I doubt that. Oh, I know you’ll trot out DxO’s measurements, but if you’re going to claim 13.5 is worse than 13.6, you need to provide the error associated with those values (e.g., 13.5 ± 0.3) and the number of sensors tested. The latter is particularly important given your statement above about low ISO banding affecting only some cameras (my EOS R certainly doesn’t have banding with long exposures at low ISO). Of course, DxO doesn’t provide the error values associated with their measurements, which is bad science (despite the fact that their tagline was ‘image science’), and I’m fairly certain that they test only one camera of each model (if they tested several copies, as lensrentals does, I’m sure they would highlight that fact). Unless you have data other than those from DxO, you really cannot conclude that the EOS R is any different from the 5DIV in terms of DR (nor would one expect it to be, given that it is essentially the same sensor).

One thing I've never fully understood is this:

Canon's still sensors have demonstrably less DR than Sony's* but Sony's video cameras have less dynamic range than Canon's. The C300 Mk II has a bit more dynamic range than the FS7 and maybe a bit more than the EVA1.

What's going on?

*even if the 1DXII, 5D IV/EOS R, and SL2/80D/SL3/M50 seem to be closing the gap.
 
Upvote 0
Jan 29, 2011
10,675
6,121
One thing I've never fully understood is this:

Canon's still sensors have demonstrably less DR than Sony's* but Sony's video cameras have less dynamic range than Canon's. The C300 Mk II has a bit more dynamic range than the FS7 and maybe a bit more than the EVA1.

What's going on?

*even if the 1DXII, 5D IV/EOS R, and SL2/80D/SL3/M50 seem to be closing the gap.
It depends what the company decides to do with the DR they have, the 5D MkIV/1DX MkII et al split the additional stop of DR they have compared to most testing by going dual pixel, if you add that 1 stop back (which you can done if you process the dual RAW shots separately and blend them) then the Canon sensors comfortably outperform Sony stills sensors at over 95% of the iso range.

In other words Canons dual pixel stills sensors do actually have more DR than Sony sensors, they just choose to use the DR a different way.
 
Upvote 0
It depends what the company decides to do with the DR they have, the 5D MkIV/1DX MkII et al split the additional stop of DR they have compared to most testing by going dual pixel, if you add that 1 stop back (which you can done if you process the dual RAW shots separately and blend them) then the Canon sensors comfortably outperform Sony stills sensors at over 95% of the iso range.

In other words Canons dual pixel stills sensors do actually have more DR than Sony sensors, they just choose to use the DR a different way.

My question has more to do with why Canon's video sensors appear to have reached technical parity with Sony's (though I'm sure many will debate this, my experience is the C300 Mk II has more DR than the FS7, though not by much) while their still sensors are lagging. I doubt their cinema cameras have enough horsepower to blend dual pixel exposures in the manner you mention, and you'd think they could apply the same technology to their stills cameras.
 
Upvote 0
May 11, 2017
1,365
635
My question has more to do with why Canon's video sensors appear to have reached technical parity with Sony's (though I'm sure many will debate this, my experience is the C300 Mk II has more DR than the FS7, though not by much) while their still sensors are lagging. I doubt their cinema cameras have enough horsepower to blend dual pixel exposures in the manner you mention, and you'd think they could apply the same technology to their stills cameras.
It might have something to do with the software used in Sony still cameras.
 
Upvote 0
Jan 29, 2011
10,675
6,121
My question has more to do with why Canon's video sensors appear to have reached technical parity with Sony's (though I'm sure many will debate this, my experience is the C300 Mk II has more DR than the FS7, though not by much) while their still sensors are lagging. I doubt their cinema cameras have enough horsepower to blend dual pixel exposures in the manner you mention, and you'd think they could apply the same technology to their stills cameras.
And my answer was Canon actually lead in both. Their still cameras technically have more DR than Sony sensors (at all but a couple of iso's) if you measure the output of both frames from the dual pixel sensors not the single frame everybody actually gives figures on.

Canon stills sensors currently are on a level of technical parity with Sony sensors. If you want to you can take a single frame with a 5D MkIV and a SonyA7 III and get more DR from the Canon image at every Iso but 126, of course you have to futz about extracting both two RAW files from the Canon, which practically nobody does, but it is possible. If to is possible then the technology is on parity.
 
Upvote 0
Jul 21, 2010
31,088
12,851
And my answer was Canon actually lead in both. Their still cameras technically have more DR than Sony sensors (at all but a couple of iso's) if you measure the output of both frames from the dual pixel sensors not the single frame everybody actually gives figures on.

Canon stills sensors currently are on a level of technical parity with Sony sensors. If you want to you can take a single frame with a 5D MkIV and a SonyA7 III and get more DR from the Canon image at every Iso but 126, of course you have to futz about extracting both two RAW files from the Canon, which practically nobody does, but it is possible. If to is possible then the technology is on parity.
His point was that the C300II also has DPAF, but still has better DR than Sony. So how does DPAF not reduce DR of a ‘video’ sensor, but does on a ‘stills’ sensor?
 
Upvote 0
Jan 29, 2011
10,675
6,121
His point was that the C300II also has DPAF, but still has better DR than Sony. So how does DPAF not reduce DR of a ‘video’ sensor, but does on a ‘stills’ sensor?
We don't know the reason, even if we knew what Canon actually did we'd still need to know what Sony did too, and I'm pretty sure if you are an insider for one you aren't for the other!

My point was saying Canon doesn't have technical parity on their sensors buys into the now completely discredited meme about DR, it simply isn't true (though it was 5 years ago). That we get different results from two separate divisions from two separate corporations, well, the only surprise is that anybody is surprised!
 
Upvote 0
Mar 2, 2012
3,187
542
We don't know the reason, even if we knew what Canon actually did we'd still need to know what Sony did too, and I'm pretty sure if you are an insider for one you aren't for the other!

My point was saying Canon doesn't have technical parity on their sensors buys into the now completely discredited meme about DR, it simply isn't true (though it was 5 years ago). That we get different results from two separate divisions from two separate corporations, well, the only surprise is that anybody is surprised!
Additionally, people have to stop conflating sensor dynamic range with the dynamic range retained in an output file. The former relates to the latter, but they aren’t the same.

Perhaps canon more effectively cools its video cameras. Perhaps canon has better noise reduction in its video codecs.
 
  • Like
Reactions: 1 user
Upvote 0
His point was that the C300II also has DPAF, but still has better DR than Sony. So how does DPAF not reduce DR of a ‘video’ sensor, but does on a ‘stills’ sensor?

Not only that, DPAF seems to increase the dynamic range on the video sensor very very dramatically. Canon rated the C300 at 12 stops and the C300 Mk II at 15. The original C300 has a more aesthetically pleasing noise texture and less sensor banding, but otherwise that three stop increase seems about accurate to me in practice.

Like the FS7/F5/F55, the C300 Mk II's generous highlight detail comes from mapping the over/under differently from a stills camera, to favor highlights at the cost of more shadow noise. But while the Sony cinema cameras have an over (stops over 18% gray) of +6 in SLOG 2 at a base ISO of 2000, Canon's have an over of +6.3 in Canon Log 2 at a base ISO of 800, and in my experience, Canon's cinema line has a relatively clean image at 800 ISO, whereas I've found I need to overexpose the F5 variants by a stop to clean up the chroma noise.

Of course, that part is subjective, and the Canon sensors have issues with sensor banding (not noise banding, but horizontal banding induced by bright light sources) that Sony doesn't have. And I suspect the Venice more than closes the gap.

Additionally, people have to stop conflating sensor dynamic range with the dynamic range retained in an output file. The former relates to the latter, but they aren’t the same.

Perhaps canon more effectively cools its video cameras. Perhaps canon has better noise reduction in its video codecs.

My bad, I'm a technical novice with digital systems and much of my background is shooting film. I'll keep my mouth shut hereafter, I didn't mean to spread misinformation.

Canon does cool their cinema cameras better. They're big and have big fans.

I'm not sure noise reduction is a big factor; Canon recently offered an option to disable it on the C300 Mk II and I've also been working lately with C200 raw light, where it's disabled by default, and still seeing similar results where the dynamic range is just ridiculous. There might be some on-sensor noise reduction, though. Canon published a white paper, but since seems to have deleted it:


Anyhow forget I wrote anything. Just confused and looking for clarity, but it appears I'm instead adding more confusion to the mix.
 
Last edited:
Upvote 0
Mar 2, 2012
3,187
542
My bad, I'm a technical novice with digital systems and much of my background is shooting film. I'll keep my mouth shut hereafter, I didn't mean to spread misinformation.

It’s not a ‘bad,’ mine was merely a suggestion; considering the camera into which a sensor is installed can help.

Websites like DXO famously publish “sensor tests,” but that’s not at all what they are. I’m sure the semiconductor fabs have dynamic range (as well as many other) requirements which they verify in acceptance testing at the sensor level. But DXO et al. do analysis of camera files. The effect of the camera is baked in.

As for the advertised DR of the C300mkII, note canon is evaluating it differently than most. See: https://www.cinema5d.com/canon-measured-15-stops-dynamic-range-c300-mark-ii/
 
Upvote 0
It’s not a ‘bad,’ mine was merely a suggestion; considering the camera into which a sensor is installed can help.

Websites like DXO famously publish “sensor tests,” but that’s not at all what they are. I’m sure the semiconductor fabs have dynamic range (as well as many other) requirements which they verify in acceptance testing at the sensor level. But DXO et al. do analysis of camera files. The effect of the camera is baked in.

As for the advertised DR of the C300mkII, note canon is evaluating it differently than most. See: https://www.cinema5d.com/canon-measured-15-stops-dynamic-range-c300-mark-ii/

In my experience the 15 stop figure is pretty accurate and consistent with most other manufacturer's methodology, but point taken. There are certain cinema companies that (imo) highly overrate DR, and others that imo underrate it (Arri–who apparently internally rate the Mini at 15+ stops, anyway, but don't want to cause confusion over incremental sensor upgrades from the original). But Canon and Sony seem to be pretty accurate in my experience, or at least consistent with each other. If anything I think they both slightly overrate DR, but it depends on if you go by what you can measure or what looks good and by when grayscale clips or when yellow clips and how the clipping looks (usually first color to clip). I've used most of these cameras pretty extensively, many side-by-side.

Of course, DR is still MUCH simpler in the video world, since we're dealing with baked-in formats that account for the sensor and the processing. That might be where my mistake is.

CML, which is imo a much more reliable resource than Cinema5D (who seem to be using the Xyla chart wrong, but that's neither here nor there), find similarly:


The C700 FF surpasses Sony's newer Venice and somehow ties the Alexa, despite its dual gain architecture. (The Alexa is still way better imo. The Alexa 65 would trounce the C700.)

Then again, I might just prefer CML because their results correlate more closely with my own experiences. This is all sort of subjective and I'll admit I'm out of my league and probably wrong about most of it. I believe they found the Red Dragon to have "just okay" DR, despite DXO Mark hyping it up. But they find the Gemini to be excellent.

Regardless, Sony's video sensors seem to perform similarly to me to their stills sensors, at least in terms of DR, whereas Canon's don't. Somehow, the generation with the addition of DPAF has vastly increased the dynamic range in their video cameras and perhaps slightly reduced it in their still cameras? Again, this is just my experience. Simultaneously they've made other sensor changes that seem to have improved both stills and video, but the leap forward overall has been greater with video. The 5D Mark IV still has more dynamic range than I need for stills, though; that's not the issue. I'd love a 5D Mark IV...

I'm just wondering if they have a trick up their sleeve in the video line and, if so, what it is, and if it will make it to their stills line next. Granted, I don't think it's necessary, everything is more than good enough now, it's just odd. I didn't realize this was such a weighted subject, but should have as it's clearly a complex one.

Oddly the original C300 has a much nicer, more organic grain pattern than the C300 Mk II, and lacks the sensor bloom/banding. :/ Further confirming just how complex and obscure this all is. To some people, that might matter more than the DR. I can see why some DPs would prefer the C500 to the C700, for instance, despite it being technically inferior. I'm just curious what's up. Canon did publish a white paper, but I can no longer find it.
 
Last edited:
Upvote 0

Don Haines

Beware of cats with laser eyes!
Jun 4, 2012
8,246
1,939
Canada
My bad, I'm a technical novice with digital systems and much of my background is shooting film. I'll keep my mouth shut hereafter, I didn't mean to spread misinformation.

Please keep talking. Intelligent discussion is how we all learn, and not just those talking. The people in the background learn too.
 
  • Like
Reactions: 1 users
Upvote 0
It depends what the company decides to do with the DR they have, the 5D MkIV/1DX MkII et al split the additional stop of DR they have compared to most testing by going dual pixel, if you add that 1 stop back (which you can done if you process the dual RAW shots separately and blend them)

I've already mentioned it in this thread - no, in general, you can't. If you mean the DPRSplit app, it does the splitting but it's unreliable. The second dng file it produces, the one that's 1 stop down, often has unfixable green cast in the brightest highlights, sometimes there's a pink cast in not so brightest highlights. Never rely on it when shooting. Sometimes you can recover some blown out highlights, but overall it's hugely unreliable. Maybe the app doesn't work properly and Canon would know how to do it better, but for some reason Canon doesn't do it.
So there's no additional 1stop in DR.
 
Upvote 0
Jan 29, 2011
10,675
6,121
As I said, yes there is, technically, which means on a technical level the sensors have parity. That is the specific comment i was taking issue with, the wretched handwringing cry of “Why Canon's video sensors appear to have reached technical parity with Sony's while their still sensors are lagging.” My point was it isn't true, there is already technical parity.
 
Upvote 0
Exactly my point. You even did not bother to Google and find the truth about your stance, and keep chewing the same cud. You may call it filmmaker perspective but many see it as pure feeble-mindedness.

You
You should read first what is intelligence and then write about it:
". . . the capacity to reorganize one’s behavior patterns so as to act more effectively and more appropriately in novel situations . . . the ability to learn . . . the extent to which a person is educable. . . the ability to carry on abstract thinking . . . the effective use of concepts and symbols in dealing with a problem to be solved . . . ” - W. Freeman
And based on your actions and reactions, thread of reasoning and use of concepts, for sure you suffer from the lack of all those aspects. Even growing up wouldn't be of much help.

Based on my analysis of the same specs you had the extraordinary talent to Google and your prior nonsense spewed about filmmaking, I'd say you show little capacity for abstract thinking, and the effective use of concepts and symbols. You're more about

Not to mention, with your "gotcha" posts and personal attacks, you're revealing much more about your own childish state than anyone else's. People tend to cover up their own psychological deficiencies with transference, which is what your last post is all about.

Good luck with that.

And let me know when you want to carry on a more intelligent debate. More mature, thoughtful minds will be waiting.
 
Upvote 0