Home · Register · Search · View Winners · Software · Hosting · Software · Join Upload & Sell

Moderated by: Fred Miranda
Username   Password

  New fredmiranda.com Mobile Site
  New Feature: SMS Notification alert
  New Feature: Buy & Sell Watchlist
  

FM Forums | Canon Forum | Join Upload & Sell

1
       2       3       end
  

Archive 2009 · Dynamic range - Canon vs Nikon
  
 
mfurman
Offline
• • • •
Upload & Sell: Off
p.1 #1 · p.1 #1 · Dynamic range - Canon vs Nikon


Can anyone explain to me why Canon cameras have their dynamic range basically unchanged from ISO 100 up to 320 (5D mk II) and Nikon newer cameras are much better (dynamic range) at low ISO?

I am referring to DxO tests

http://www.dxomark.com/index.php/eng/DxOMark-Sensor



Jul 11, 2009 at 03:00 PM
Chris Beaumont
Offline
• • • • •
Upload & Sell: Off
p.1 #2 · p.1 #2 · Dynamic range - Canon vs Nikon


The DxO results are highly specious at times anyway, if you look at their "top rated cameras" you'll see things that run very counter to what general experience says (D40X is supposedly a superior camera to 40D, D5000 better than the original 5D.........)

That aside, DO Nikon cameras have BETTER DR at low ISO than Canon, or does Canon's DR break up later in the ISO range ?



Jul 11, 2009 at 03:06 PM
mfurman
Offline
• • • •
Upload & Sell: Off
p.1 #3 · p.1 #3 · Dynamic range - Canon vs Nikon


I could word my question differently: why does EV vs ISO characteristic curves differ so much between Canon and Nikon cameras? Canon usually has a plateau at low ISO values. Canon generally does not seem to have good dynamic range.

I would like to emphasize that I do not have experience with any digital SLRs but Canon (Canon user since 1990). It seems to me though that in certain respect Nikon cameras measure better (I specifically did not say that they are better). This shows not only in DxO but other tests as well.




Jul 11, 2009 at 03:15 PM
jscoby05
Offline
• •
Upload & Sell: Off
p.1 #4 · p.1 #4 · Dynamic range - Canon vs Nikon


I can tell you from personal experience that if you have a 5D, 1D Mark III, 1Ds Mark III and for the most part the 5D Mark II; there is nothing you will be missing by not shooting with Nikon. Those first 3 cameras produce excellent tonality and dynamic range for non-MF digital camera backs.

http://www.robgalbraith.com/bins/multi_page.asp?cid=7-8740-9068-9537



Jul 11, 2009 at 04:08 PM
stargazer78
Offline
• •
Upload & Sell: Off
p.1 #5 · p.1 #5 · Dynamic range - Canon vs Nikon


mfurman wrote:
I could word my question differently: why does EV vs ISO characteristic curves differ so much between Canon and Nikon cameras? Canon usually has a plateau at low ISO values.



As you have noticed, the DR curve for Canon reaches a plateau at the lower ISO settings. The reason for this is read noise caused by sub-standard electronic components on the sensor. There is a constant low-level noise generated by poorly designed electronics, and this low-level noise sets a hard limit on how much detail you can pull out of the shadows.

Whatever extra DR you might've gained in shooting ISO 100 instead of ISO 400 is obliterated by this blanket of constant noise. Note that this operational noise produced by electronic components is very low, but constant; so at higher ISO settings their effects are negligible compared to other, more significant sources of noise. Unfortunately, at the lowest ISO settings this operational noise is essentially the largest source of noise.

This is the reason why the 5D Mark II, for example, doesn't have as much DR as you might've expected at low ISO... given its excellent high ISO performance.

But it would be unfair to attribute this problem strictly to Canon cameras. Based on the DXO charts, it looks like (almost) all other brands out there exhibit this characteristic. But Nikon must've made major advancements in its recent batch of SLRs. They developed it for use with the D3x, then apparently passed down the tech to other cameras since (D90, D5000). That's why these newer Nikons all have DR curves whose slopes don't plateau at the lower ISO ranges. For example, the lowly D5000 boasts higher DR than the mighty D700...



Jul 11, 2009 at 04:12 PM
skid00skid00
Offline
• •
Upload & Sell: Off
p.1 #6 · p.1 #6 · Dynamic range - Canon vs Nikon


One of the reasons DXO tests have been -fooled- by the latest Nikon cameras is that Nikon is clipping the black level in the RAW file. IOW, if the pixel has a numeric value of 2 (out of say 255) Nikon changes the value to zero. If the pixel has a value of -2 (noise) it also gets clipped to zero. (Keep in mind that the pixels are -analog- and can have positive OR negative voltages).

This makes the file -look- cleaner/have more DR, but it also is wiping out shadow detail.

Canon applies a positive offset to the deepest blacks. IOW, if the pixel has a value of 2, Canon will add 5 to it, giving a final value of 7. If the pixel has a value of -2, after adding 5 to it, you have a pixel with a value of 3. Unlike in the Nikon, you have two shadow pixels with two different values.

Even though the Nikon deleted 'detail', to DXO's simplistic examination it looks cleaner, therefore DXO gives it a higher DR.

The above example is not specifically how the Nikon and Canon systems work, but is a reasonable simile.



Jul 11, 2009 at 06:48 PM
skibum5
Offline
• • • • • •
Upload & Sell: Off
p.1 #7 · p.1 #7 · Dynamic range - Canon vs Nikon


skid00skid00 wrote:
One of the reasons DXO tests have been -fooled- by the latest Nikon cameras is that Nikon is clipping the black level in the RAW file. IOW, if the pixel has a numeric value of 2 (out of say 255) Nikon changes the value to zero. If the pixel has a value of -2 (noise) it also gets clipped to zero. (Keep in mind that the pixels are -analog- and can have positive OR negative voltages).

This makes the file -look- cleaner/have more DR, but it also is wiping out shadow detail.

Canon applies a positive offset to the deepest blacks. IOW,
...Show more

but for their DR test they find the lowest distinguishable SIGNAL and use that as the base point so who cares if nikon clipped off the blackpoint I don't think their DR test has anything to do with that at all unless they did it very badly.

the one thing that seems really odd is the weird kink in the D90 graph for DR though when it gets a cusp and changes from positive to negative curvature in a way that just seems very odd





Jul 11, 2009 at 06:59 PM
keithreeder
Offline
• • • •
Upload & Sell: Off
p.1 #8 · p.1 #8 · Dynamic range - Canon vs Nikon


I'll tell you what's odder - according to DXO, the Noink D200 has a better DR than the Canon 40D, FFS!

Yes, that's the 40D that, until the release of the Nikon D5000, was acknowledged as having the best DR of any prosumer body out there.

I've owned D200s in the past, and I own a 40D now: and I can assure anyone reading this that the 40D is in another league in terms of Real-World-Useful DR compared to the Nikon body.



Jul 11, 2009 at 08:02 PM
stargazer78
Offline
• •
Upload & Sell: Off
p.1 #9 · p.1 #9 · Dynamic range - Canon vs Nikon


keithreeder wrote:
I'll tell you what's odder - according to DXO, the Noink D200 has a better DR than the Canon 40D, FFS!



According to DXO, the Nikon D200 has superior DR to the 40D when both cameras are at ISO 100. But I was under the impression that the D200 base ISO was ISO 200? Isn't ISO 100 on that camera one of those faked expansion ISO settings?



Jul 11, 2009 at 10:02 PM
brainiac
Offline
• • • • •
Account locked
p.1 #10 · p.1 #10 · Dynamic range - Canon vs Nikon


DXOmark is a really bad way of choosing a camera, particularly if you look at the default (per pixel) analysis. Ignore it.


Jul 11, 2009 at 10:26 PM
 

Search in Used Dept. 



vpk24_astro
Offline
• •
Upload & Sell: Off
p.1 #11 · p.1 #11 · Dynamic range - Canon vs Nikon


DXOmark is full of crap. You cannot make statements about populations of electronic products by testing A SINGLE SAMPLE. To make meaningful statements, you need to test lots of bodies (100+). DXOmark does nothing like that. What they have is an advertising tool. They write a RAW converter and draw attention to it by publishing these results in the hope that photographers will think that there is something superior about their software since they're a more 'scientific' company.
Electronic products typically carry huge standard deviations from published specs because the products are all manufactured identically and then simply binned based on sampled quality. Go rip an electrolytic capacitor (the ones that look like big blue drums) out of an old radio and measure the capacitance - you'll be surprised by what you find.
Out of curiosity, why is it that you think Canon and Nikon don't publish the DR and noise levels of their cameras? They aren't idiots when it comes to marketing. The truth is - it is hard (if not impossible) to come up with a few numbers to quantify all aspects of performance accurately. Furthermore, any numbers to quantify DR and noise levels would necessarily carry huge error bars on them because thats how it must be if you are to make an affordable camera - and that applies to the $8000 bodies too. In the world of high performance sensors (anything backlit for example), $8000 is barely a tenth of the entry point.



Jul 11, 2009 at 10:51 PM
mfurman
Offline
• • • •
Upload & Sell: Off
p.1 #12 · p.1 #12 · Dynamic range - Canon vs Nikon


vpk24_astro:
Out of curiosity, why is it that you think Canon and Nikon don't publish the DR and noise levels of their cameras?


I frankly cannot believe that there could be big differences in DR between different copies of any DSLR.

To all who are skeptical about DxO mark, could you point me to any more objective test?

stargazer78, it is interesting what you wrote but I could not believe that Canon would be so much behind Nikon in IC development.



Jul 12, 2009 at 12:21 AM
skid00skid00
Offline
• •
Upload & Sell: Off
p.1 #13 · p.1 #13 · Dynamic range - Canon vs Nikon


skibum5 wrote:
but for their DR test they find the lowest distinguishable SIGNAL and use that as the base point so who cares if nikon clipped off the blackpoint I don't think their DR test has anything to do with that at all unless they did it very badly.


Clipping the blacks eliminates the noise at those levels, making the file measure as if it had a higher DR.

Let's make an example of an imaginary Canon file. In my model 1skidoo, there's no clipping of blacks, so shadow pixels have enough noise so that none of them ever go below a value of 20 on a 0-255 range (8 bpc). It's "DR" is 20 to 255 or 236.

My new and improved 2skidoo model clips values of 20 and below, artificially putting them at zero. It then stretches the remaining 21-255 values into the 0-255 range so there are no gaps. This model has a "DR" of 0 to 255, or 256. It measures greater, even though the actual SNR/DR is identical to the first model.

Again, this is an example, and not exactly how things work IRL.



Jul 12, 2009 at 03:42 AM
skid00skid00
Offline
• •
Upload & Sell: Off
p.1 #14 · p.1 #14 · Dynamic range - Canon vs Nikon


mfurman, in several cases, DXO's results contradict users visual perceptions. IIRC, they claimed that one of Nikons older crop cameras had better DR than a D300, when it was visually obvious that it just wasn't so.

My two cents...



Jul 12, 2009 at 03:46 AM
skibum5
Offline
• • • • • •
Upload & Sell: Off
p.1 #15 · p.1 #15 · Dynamic range - Canon vs Nikon


skid00skid00 wrote:
Clipping the blacks eliminates the noise at those levels, making the file measure as if it had a higher DR.

Let's make an example of an imaginary Canon file. In my model 1skidoo, there's no clipping of blacks, so shadow pixels have enough noise so that none of them ever go below a value of 20 on a 0-255 range (8 bpc). It's "DR" is 20 to 255 or 236.

My new and improved 2skidoo model clips values of 20 and below, artificially putting them at zero. It then stretches the remaining 21-255 values into the 0-255 range so there are no
...Show more


that seems pretty ridiculous you can't get more DR by chopping away part of what you collected and I think DxO knows that as well as you and I don't believe the process they use would do anything like that at all, although i guess it is always possible that they do something where the tail of deviation being cut-off alters their results, but i would hope not




Edited on Jul 12, 2009 at 08:09 PM · View previous versions



Jul 12, 2009 at 04:02 AM
George.ML
Offline
• •
Upload & Sell: Off
p.1 #16 · p.1 #16 · Dynamic range - Canon vs Nikon


mfurman wrote:
stargazer78, it is interesting what you wrote but I could not believe that Canon would be so much behind Nikon in IC development.


Well, it's not exactly sub-standard electronic components on the sensor.
They were actually the best ... circa 2003.

The fact is, Canon is still using the same basic technology they had in 2003 - if not earlier.

Nikon's current technology is newer and obviously better.

In 2007 Canon announced that they were building a new CMOS plant for sensors.
There may be some truth after all behind the rumors that the 1D4 and 60D will have better ISO (and hopefully DR):
http://www.reuters.com/article/companyNewsAndPR/idUST14022720070715




Jul 12, 2009 at 07:14 AM
alundeb
Offline
• • • •
Upload & Sell: Off
p.1 #17 · p.1 #17 · Dynamic range - Canon vs Nikon


The clipping that occurs in some Nikon cameras is mainly just omitting the offset. The clipping then occurs at the true black level, and any "signal" is still perfectly measurable, as long as the "signal" is uniform light. The noise level as standard deviation becomes roughly halved, and the DR measures as roughly 1 stop higher. A perfect camera design for fooling the DxO mark DR test.

The one Nikon camera with doubtless excellent DR is the D3x. They really made it right with that one, and I'm hoping it hurts Canon's pride enough that Canon will catch up.

The noise floor that affects Canon DR is not Read Noise in the strict definition of the term. The Read Noise out of the sensor is amplified proportionally with the ISO setting, and if that was the dominant noise at low ISO, the DR curve would be linear all the way. The noise that cripples DR at low ISO, comes from the amplifiers and/or the A/D converters.

It's not true that Canon's electronics is unchanged since 2003.
Comparison of ISO100 black frames of a 450D (image stolen from theDigitalBean in the famous criss-cross thread) and my own 500D:

Image 349329 not found



Image 349330 not found





Jul 12, 2009 at 10:13 AM
keithreeder
Offline
• • • •
Upload & Sell: Off
p.1 #18 · p.1 #18 · Dynamic range - Canon vs Nikon


stargazer78 wrote:
According to DXO, the Nikon D200 has superior DR to the 40D when both cameras are at ISO 100. But I was under the impression that the D200 base ISO was ISO 200? Isn't ISO 100 on that camera one of those faked expansion ISO settings?


Yep, correct - no real 100 ISO on a D200.

That DXO stuff is just such utter bloody nonsense.



Jul 12, 2009 at 10:15 AM
vpk24_astro
Offline
• •
Upload & Sell: Off
p.1 #19 · p.1 #19 · Dynamic range - Canon vs Nikon


mfurman wrote:
I frankly cannot believe that there could be big differences in DR between different copies of any DSLR.

To all who are skeptical about DxO mark, could you point me to any more objective test?

stargazer78, it is interesting what you wrote but I could not believe that Canon would be so much behind Nikon in IC development.


I don't see any error bars in anything that DXOMark publishes. Why would you not have error bars if you're testing more than one lens or more than one sensor? All analog devices have random variations built into them. Sensors are inherently analog (sensor data becomes discretized after going through the ADC) => sensors
have variations built into them.
There are no public objective tests. To perform a statistically reliable test would require the tester to purchase the body being tested in wholesale quantities. In this respect, I agree with Luminous Landscape's policy of reviewing the construction, ergonomics and feel of a camera body/lens as opposed to the IQ. Incidentally lenses also have IQ variations from sample to sample - so don't take the IQ tests performed by the-digital-picture.com or photozone.de to heart. They're decet indicators of what IQ you might expect but they might have received a sub or superstandard sample.



Jul 13, 2009 at 02:36 PM
thw2
Offline
• • • •
Upload & Sell: Off
p.1 #20 · p.1 #20 · Dynamic range - Canon vs Nikon


Thank you mfurman for bringing up this topic. It's been disturbing me for a while... I think some of the replies in this thread have addressed my concerns.


Jul 13, 2009 at 03:48 PM
1
       2       3       end




FM Forums | Canon Forum | Join Upload & Sell

1
       2       3       end
    
 

You are not logged in. Login or Register

Username   Password    Reset password