October 25, 2014, 08:21:51 AM

Author Topic: DxOMark Sensor Performance: Nikon vs. Canon  (Read 82158 times)

Razor2012

  • 5D Mark III
  • ******
  • Posts: 639
    • View Profile
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #345 on: October 19, 2012, 03:21:32 PM »
OK, so I took a few shots with my 5D3 (with the 24-105) in a darkish room. The mode was Av at f8 with ISO set at 12k8 and 25k6 and the noise level in the raw pics are nowhere near Mikael's picture noise. Mikael, may be you need to replace your copy of the 5D3. Either it does not like you one bit {notice no fractional reference here ;) } or you just need a good copy...

Or maybe, what's with the hard-on for Canon?
Oh dear! Are you saying that you also get the same noise levels as Mikael when you take pictures? Is your camera still under warranty? I hope so.

Lol, rpt the reference wasn't for you.  ;)
5D MKIII w grip, 70-200 2.8L IS II, 24-70 2.8L II, 16-35 2.8L II, 100 2.8L IS macro, 600EX-RT

canon rumors FORUM

Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #345 on: October 19, 2012, 03:21:32 PM »

LetTheRightLensIn

  • Canon EF 300mm f/2.8L IS II
  • *******
  • Posts: 3937
    • View Profile
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #346 on: October 19, 2012, 03:24:46 PM »
But this thread is about low ISO DR.

Is it?  Why?  Is it because everyone shoots at ISO 100 all the time?  Or because DxO bases their Score on ISO 100?   ::)

Well it is 26 pages and 25.5 have so far been about ISO100 or close to that so....  ::)

And the relevant thing is someone was trying to directly compare ISO25k to ISO100 which doesn't make sense.

rpt

  • Canon EF 300mm f/2.8L IS II
  • *******
  • Posts: 2219
  • Could not wait for 7D2 so I got the 5D3
    • View Profile
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #347 on: October 19, 2012, 03:40:13 PM »
But this thread is about low ISO DR.

Is it?  Why?  Is it because everyone shoots at ISO 100 all the time?  Or because DxO bases their Score on ISO 100?   ::)

Well it is 26 pages and 25.5 have so far been about ISO100 or close to that so....  ::)

And the relevant thing is someone was trying to directly compare ISO25k to ISO100 which doesn't make sense.
Well, I did a ISO 200 shot because it went to 30 sec at f4 with EC at 0 and it came out better than the ones I posted. so I wont post it unless you think you must see it. Next stop 100 ISO... I dont see noise at 200! And I will post later to show you. But, I must be doing something different from you chaps not to see noise. What is it? Is it the exposure getting enough photons in the buckets? So I have an idea for the 100 iso. Let me get to that now.

rpt

  • Canon EF 300mm f/2.8L IS II
  • *******
  • Posts: 2219
  • Could not wait for 7D2 so I got the 5D3
    • View Profile
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #348 on: October 19, 2012, 04:07:39 PM »
OK, Here is the ISO 100 image. I is better than others so I think I am surely not doing something you guys are doing. Sorry but can you tell me the "exact" steps to repro the problem? Obviously shooting a decent exposure is not working... Here is the image:

img708.imageshack.us/i/budha100raw.jpg


The light was very low. No lighting switched on in the room where I shot. Some light coming in the room from two other rooms and none of it on the Buddha... Some light coming in from a window opposite but not enough for a human to figure the shape...

Look forward to your inputs.

Rustom


neuroanatomist

  • CR GEEK
  • ********
  • Posts: 14752
    • View Profile
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #349 on: October 19, 2012, 04:14:42 PM »
OK, Here is the ISO 100 image. I is better than others so I think I am surely not doing something you guys are doing. Sorry but can you tell me the "exact" steps to repro the problem? Obviously shooting a decent exposure is not working...

You're shooting the metered exposure?  If so, that's the problem... or rather that's why there is no problem.   If you want to create the problem, you must underexpose by four stops or so (your image should be nearly all black), then push the exposure back up by those four stops in post. Oh, and be sure to take the absolute darkest part of that image, and view that at 100%.  And that's all you have to do to see horrible noise and banding in your images. It's just that easy.
EOS 1D X, EOS M, and lots of lenses
______________________________
Flickr | TDP Profile/Gear List

rpt

  • Canon EF 300mm f/2.8L IS II
  • *******
  • Posts: 2219
  • Could not wait for 7D2 so I got the 5D3
    • View Profile
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #350 on: October 19, 2012, 04:18:56 PM »
OK, Here is the ISO 100 image. I is better than others so I think I am surely not doing something you guys are doing. Sorry but can you tell me the "exact" steps to repro the problem? Obviously shooting a decent exposure is not working...

You're shooting the metered exposure?  If so, that's the problem... or rather that's why there is no problem.   If you want to create the problem, you must underexpose by four stops or so (your image should be nearly all black), then push the exposure back up by those four stops in post. Oh, and be sure to take the absolute darkest part of that image, and view that at 100%.  And that's all you have to do to see horrible noise and banding in your images. It's just that easy.
So I should either create a situation where the DR in the scene exceeds the DR of the camera or deliberately underexpose?

Huh!

RLPhoto

  • Canon EF 300mm f/2.8L IS II
  • *******
  • Posts: 3507
  • Gear doesn't matter, Just a Matter of Convenience.
    • View Profile
    • My Portfolio
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #351 on: October 19, 2012, 04:22:57 PM »
OK, Here is the ISO 100 image. I is better than others so I think I am surely not doing something you guys are doing. Sorry but can you tell me the "exact" steps to repro the problem? Obviously shooting a decent exposure is not working... Here is the image:

The light was very low. No lighting switched on in the room where I shot. Some light coming in the room from two other rooms and none of it on the Buddha... Some light coming in from a window opposite but not enough for a human to figure the shape...

Look forward to your inputs.

Rustom

Thats because you got your exposure correct.  ;D

canon rumors FORUM

Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #351 on: October 19, 2012, 04:22:57 PM »

rpt

  • Canon EF 300mm f/2.8L IS II
  • *******
  • Posts: 2219
  • Could not wait for 7D2 so I got the 5D3
    • View Profile
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #352 on: October 19, 2012, 04:24:31 PM »
OK, Here is the ISO 100 image. I is better than others so I think I am surely not doing something you guys are doing. Sorry but can you tell me the "exact" steps to repro the problem? Obviously shooting a decent exposure is not working... Here is the image:

The light was very low. No lighting switched on in the room where I shot. Some light coming in the room from two other rooms and none of it on the Buddha... Some light coming in from a window opposite but not enough for a human to figure the shape...

Look forward to your inputs.

Rustom

Thats because you got your exposure correct.  ;D
Shoot! and I paid over $4200 for that! Bad me!

LetTheRightLensIn

  • Canon EF 300mm f/2.8L IS II
  • *******
  • Posts: 3937
    • View Profile
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #353 on: October 19, 2012, 05:22:13 PM »
But this thread is about low ISO DR.

Is it?  Why?  Is it because everyone shoots at ISO 100 all the time?  Or because DxO bases their Score on ISO 100?   ::)

Well it is 26 pages and 25.5 have so far been about ISO100 or close to that so....  ::)

And the relevant thing is someone was trying to directly compare ISO25k to ISO100 which doesn't make sense.
Well, I did a ISO 200 shot because it went to 30 sec at f4 with EC at 0 and it came out better than the ones I posted. so I wont post it unless you think you must see it. Next stop 100 ISO... I dont see noise at 200! And I will post later to show you. But, I must be doing something different from you chaps not to see noise. What is it? Is it the exposure getting enough photons in the buckets? So I have an idea for the 100 iso. Let me get to that now.

You must be shooting scenes where the DR more easily fits than the ones he was. For many scenes that will be the case, but he is talking about the ones with really large DR.


LetTheRightLensIn

  • Canon EF 300mm f/2.8L IS II
  • *******
  • Posts: 3937
    • View Profile
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #354 on: October 19, 2012, 05:25:44 PM »
That scene has low DR so it all fits well and there is no noise issue at all. Low light doesn't mean anything regarding the dynamic range of a scene, with flat lighting like that it will often be low.

You need something where you like sun shining brightly on a light object, then you expose so as to not blow that out and then if there are important details in really dark parts of the scene that is when you can get into trouble. It's when there are really bright and really dark parts in the same scene that the trouble occurs. You either expose it to get nice noise free detail in the darker parts and then the bright stuff gets blown out to nothing or you expose to save the bright parts and then the dark parts are a mess (if the range of brightness between the two is large enough).

OK, Here is the ISO 100 image. I is better than others so I think I am surely not doing something you guys are doing. Sorry but can you tell me the "exact" steps to repro the problem? Obviously shooting a decent exposure is not working... Here is the image:

img708.imageshack.us/i/budha100raw.jpg


The light was very low. No lighting switched on in the room where I shot. Some light coming in the room from two other rooms and none of it on the Buddha... Some light coming in from a window opposite but not enough for a human to figure the shape...

Look forward to your inputs.

Rustom

LetTheRightLensIn

  • Canon EF 300mm f/2.8L IS II
  • *******
  • Posts: 3937
    • View Profile
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #355 on: October 19, 2012, 05:28:59 PM »
OK, Here is the ISO 100 image. I is better than others so I think I am surely not doing something you guys are doing. Sorry but can you tell me the "exact" steps to repro the problem? Obviously shooting a decent exposure is not working...

You're shooting the metered exposure?  If so, that's the problem... or rather that's why there is no problem.   If you want to create the problem, you must underexpose by four stops or so (your image should be nearly all black), then push the exposure back up by those four stops in post. Oh, and be sure to take the absolute darkest part of that image, and view that at 100%.  And that's all you have to do to see horrible noise and banding in your images. It's just that easy.

Perhaps if you shot more landscapes and forest scenes or scenes with large DR you'd really that is really IS just that easy and maybe not be quite so flip about it all. Again you can shoot billions of types of shots without issue but you can also find millions where it would be a bit of trouble, some of them you can then save by using special filters or multiple shots or hours of post processing but there are many where that stuff won't work. So it's hardly the end of the world, but it's hardly something to just totally laugh off either and make seems the height of silliness and absurdity.



LetTheRightLensIn

  • Canon EF 300mm f/2.8L IS II
  • *******
  • Posts: 3937
    • View Profile
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #356 on: October 19, 2012, 05:31:23 PM »
OK, Here is the ISO 100 image. I is better than others so I think I am surely not doing something you guys are doing. Sorry but can you tell me the "exact" steps to repro the problem? Obviously shooting a decent exposure is not working...

You're shooting the metered exposure?  If so, that's the problem... or rather that's why there is no problem.   If you want to create the problem, you must underexpose by four stops or so (your image should be nearly all black), then push the exposure back up by those four stops in post. Oh, and be sure to take the absolute darkest part of that image, and view that at 100%.  And that's all you have to do to see horrible noise and banding in your images. It's just that easy.
So I should either create a situation where the DR in the scene exceeds the DR of the camera or deliberately underexpose?

Huh!

Yes, case 1 is the one that most are going on about, when the DR is too much for the sensor, but case 2 certainly doesn't hurt, I mean everyone has had some one of shot they can't retake and for some reason the exposure got way underexposed by accident, who wouldn't want to be able to better rescue it? But the main point is for images that are exposed as you had wanted but still won't fit. Some people will hardly ever encounter that with the current Canon level of performance and it won't matter to them at all and then some will often and some only very here and there and some somewhat more often here and there and some a real lot, it depends.

I'd certainly prefer to be able to be open to all of the extra possibilities that shooting scenes with 3 more stops of DR allows for than not have those options.

elflord

  • 5D Mark III
  • ******
  • Posts: 705
    • View Profile
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #357 on: October 19, 2012, 07:47:23 PM »
I do dispute that you actually gain anything beneficial by downscaling a high megapixel image to a smaller size, in a real-world context

Actually, we agree here.  When you downsample, you trade resolution for lower noise. You're not creating a higher quality image. Depending on the situation, it might or might not be a tradeoff you'd want to make.
Quote
Agreed. From the standpoint of normalizing noise to compare how noisy one sensor is vs. another (which is what I gather the DXO Print DR statistic is all about based on the arguments from you an LTRLI), then sure, you should be normalizing image size.

I'm glad we agree with this, because this is basically my whole point -- if you're comparing sensors, then you really should use the print measurement. If you're using the measurement for anything else, then the "screen" DR is as good (or better)

I can understand that from the perspective of having already picked the camera and wanting to measure its DR for purposes other than comparisons, parsimony rules, so I understand your preference for the per pixel score in this context.

But DxO are really in the business of publishing benchmarks that will be used for comparison purposes, which explains their emphasis.
« Last Edit: October 19, 2012, 07:49:39 PM by elflord »

canon rumors FORUM

Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #357 on: October 19, 2012, 07:47:23 PM »

jrista

  • Canon EF 400mm f/2.8L IS II
  • *******
  • Posts: 4483
  • EOL
    • View Profile
    • Nature Photography
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #358 on: October 19, 2012, 08:35:07 PM »
I do dispute that you actually gain anything beneficial by downscaling a high megapixel image to a smaller size, in a real-world context

Actually, we agree here.  When you downsample, you trade resolution for lower noise. You're not creating a higher quality image. Depending on the situation, it might or might not be a tradeoff you'd want to make.
Quote
Agreed. From the standpoint of normalizing noise to compare how noisy one sensor is vs. another (which is what I gather the DXO Print DR statistic is all about based on the arguments from you an LTRLI), then sure, you should be normalizing image size.

I'm glad we agree with this, because this is basically my whole point -- if you're comparing sensors, then you really should use the print measurement. If you're using the measurement for anything else, then the "screen" DR is as good (or better)

I can understand that from the perspective of having already picked the camera and wanting to measure its DR for purposes other than comparisons, parsimony rules, so I understand your preference for the per pixel score in this context.

But DxO are really in the business of publishing benchmarks that will be used for comparison purposes, which explains their emphasis.

Woot! We agree! We should party.

tron

  • 1D X
  • *******
  • Posts: 1865
    • View Profile
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #359 on: October 19, 2012, 08:36:41 PM »
What the above image has in common with an image like Mikael's which needs the shadows to be lifted?
The lighting was low but without heavy shadows and extreme highlights. It looks more like you deliberately avoided such a subject.

Why don't you shoot outside in the sun and include some harsh shadows and try to lift them so as to show details?

canon rumors FORUM

Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #359 on: October 19, 2012, 08:36:41 PM »