Hi ISO performance – question

My Question:
I wonder if the Nikons (mid range cameras) are truly better than Canon on this front or is it mainly marketing (and in camera PP)


Some background info :

Recently I shoot a lot at dusk or dawn (if I force myself to get up early)


I currently have an Xsi (and I like it !) but may consider a change if I shall get significantly better hi ISO performance

Recently I have read (e.g. on DPReview the new T1i review ; or camera lab hi iso comparison) that hi ISO performance of Canon lag behind Nikon
But I also carefully read the threat where PIXmantra clearly explains that with careful (and tedious) post processing he gets much better results from his 1DMKIII than any Nikon and he states that the Canon Cmos sensor & Canon Digic Processor combo has a much better image handling. (I do realize he referred to the 1D) see the thread (https://photography-on-the.net/forum/showthre​ad.php?t=701142&highli​ght=noise+d700 )

So what do you think ?

5DMKII, 1DMKIII, 1DsMKIII, D700, D3 are close enough in noise that I don't think it matters.

Great pictures Nicksan !!! (on your flickr)

What's wrong with a tripod? The primary way you get better ISO performance from larger sensor cameras is because they have larger apertures for the same perspective, framing, and f-ratio, which also gives you a more shallow DOF. So, unless you also want the more shallow DOF and its concomitant effects (softer corners and more vignetting), you'd have to stop the larger sensor options down, which means they'd need a higher ISO for the same shutter speed.

Joe, who was talking about larger sensors?

Omer, are you considering a change to Nikon or to a 1D MkIII to get better ISO noise handling?
Or both?

I've not worked with an Xsi file, so I'm not an expert there, but the 1D MkIII files are indeed amazing to work with.
I think Pixmantra is on to something.

joe mama wrote in post #8077690
What's wrong with a tripod? The primary way you get better ISO performance from larger sensor cameras is because they have larger apertures for the same perspective, framing, and f-ratio, which also gives you a more shallow DOF. So, unless you also want the more shallow DOF and its concomitant effects (softer corners and more vignetting), you'd have to stop the larger sensor options down, which means they'd need a higher ISO for the same shutter speed.

i got lost after "What's wrong with a tripod".

Lenses (well some) have larger apertures not sensors.

The answer is right there, in front of your eyes...

omer wrote in post #8077192
My Question:
I wonder if the Nikons (mid range cameras) are truly better than Canon on this front or is it mainly marketing (and in camera PP)

...Even without my opinion, try something simple (anyone can do it, indeed):

1. Go to the latest DPR T1i review, and jump to the ISO comparison section.
2. Go to the .JPG section (right at the top of the page), and FOCUS on the following three samples:

a. Ti1
b. D5000
c. 450D/XSi
http://www.dpreview.co​m …s/canoneos500d/​page15.asp

3. LOOK at the actual face-detail shown there at ISO100, and then quickly roll your mouse cursor over the ISO1600 label.... and you will notice (contrary to DPreview's own comments) that it is the 450D showing the overall best, most detailed image among ALL the cameras, followed by the T1i, and then following by mush-factory, Nikon D5000 (this is unquestionable, it is right there for everyone to see)...

Now, here comes the tragic part: do the exact same pan, as described above,but for the RAW samples, just scrolling down a bit on the same page, and roll over the ISo100 label, and then over the ISO1600 label... What do yo see? Exactly same results as above: 450D comes out strongly detailed, sharper (higher image integrity), then followed by T1i, and then (again) followed by even same mush-factory, Nikon D5000 (this is very, very strong indication that Nikon ***RAW*** files are all crooked/massaged with heavy-metal noise-reduction, on-board).

Answer to your question: LOTS of marketing and LOTS of on-board Noise Reduction (which is highly tuned for the camera's output, but not on par to what you can get from DiGiCIII output, and equally dedicated/tuned Noise Reduction).

omer wrote in post #8077192

But I also carefully read the threat where PIXmantra clearly explains that with careful (and tedious) post processing he gets much better results from his 1DMKIII than any Nikon and he states that the Canon Cmos sensor & Canon Digic Processor combo has a much better image handling. (I do realize he referred to the 1D) see the thread (

https://photography-on-the.net/forum/showthre​ad.php?t=701142&highli​ght=noise+d700 )


So what do you think ?

...The purpose of a tool like FlexNR is to remove the word "tedious" and replace it with "one-click" for everyone, and leave only one person with the "blunt" of the work (that is me... ).

Now, you can rest assured the 1D3's on-board output (DiGiCIII .JPGs or conversions with RIT/ZB) already allow the camera to achieve unseen results, from the camera itself, and/or any other camera currently available, in the high-ISO department (and all this, mind you, with native ISO100 support, all the way to ISO6400, which is still analog-output from the 1D3 sensor, and not pushed digitially... ):

IS6400 1D3-vs-D3, .JPGs (from Imaging Resource, which I always recommend):

(look at the red cloth, the black mug, deeper shadows, and green plastic leaves)

http://www.pbase.com …/image/11272501​6/original
http://www.pbase.com …/image/11272501​8/original

And here is a simple, ISO12800 sample from the 1D3 (via RIT/ZB but otherwise identical to .JPG output, upward compensated +1.00ec), taking advantage of the unique spectral composition of .JPGs/RIT output (DiGiCIII):

http://www.pbase.com …/image/11326282​4/original

Enjoy!

PIX

PIXmantra wrote in post #8078036
[/FONT][/SIZE]

...Even without my opinion, try something simple (anyone can do it, indeed):

1. Go to the latest DPR T1i review, and jump to the ISO comparison section.
2. Go to the .JPG section (right at the top of the page), and FOCUS on the following three samples:

a. Ti1
b. D5000
c. 450D/XSi
http://www.dpreview.co​m …s/canoneos500d/​page15.asp

3. LOOK at the actual face-detail shown there at ISO100, and then quickly roll your mouse cursor over the ISO1600 label.... and you will notice (contrary to DPreview's own comments) that it is the 450D showing the overall best, most detailed image among ALL the cameras, followed by the T1i, and then following by mush-factory, Nikon D5000 (this is unquestionable, it is right there for everyone to see)...

That is exactly what I was going to post. At some point, noise-free means WORSE image quality.

Hell, if you want perfectly noise free images, just go into photoshop and set up a gaussian blur of 100 pixels. Perfectly smooth, noise-free images every time.

CyberDyneSystems wrote in post #8077739
Joe, who was talking about larger sensors?

Omer, are you considering a change to Nikon or to a 1D MkIII to get better ISO noise handling?
Or both?

Well I rather stay with a Canon crop body

Absolutely...

tkbslc wrote in post #8078149
That is exactly what I was going to post. At some point, noise-free means WORSE image quality.

Hell, if you want perfectly noise free images, just go into photo shop and set up a gaussian blur of 100 pixels. Perfectly smooth, noise-free images every time.

...That is precisely why a clear / sound processing strategy needs to be put in place PRIOR to conducting any Noise Reduction efforts (which are, at the end, indispensable for high quality ISO work).

If the strategy is "smooth as baby's butt", so you can score the "best" in analytical/objective tests, then you get what you pay for (Nikon's 5000).

Now, if your strategy is "reduce it, and make it look organic/natural", well, we are talking about an entirely different outcome.

I have some 450D/XSi ISO1600 runs on IR's .JPGs that look as good as you can get them, and this camera could have handled (easily) ISO3200, better than pretty much any cameras in its category.

NOTE: be aware, though, that I have not been able to see a fully-disengaged, non-NR .JPG sample from any small-dSLR DiGiCIV-based camera (it seems that DiGiCIV's NR cannnot be fully neutralized, as it is mostly with DiGiCIII, unless there is a serious, major change in RAW-to-RGB conversion routines that are now "noise-aware", with respect to prior DiGiC implementations).

For me, on-board output is of SUPREME importance, because I honestly don't think that we should spend our lives behind a RAW converter for every single frame we shoot, though.

PIX

PIXmantra wrote in post #8078036
[/FONT][/SIZE]

...The purpose of a tool like FlexNR is to remove the word "tedious" and replace it with "one-click" for everyone, and leave only one person with the "blunt" of the work (that is me... ).

PIX

Thanks for the detailed post

How do i get FlexNR

mind you i do not mind tweaking and adjusting during PP
i like to have a good understanding of what i do and a solid work-flow i can reproduce

In the OP, omer mentions that he has an XSi, talks about the T1i, and then compares noise to the 1DIII, which has a different sensor size.

Yes, it is the aperture of the lens, not the size of the sensor, that matters. However, the lenses for larger sensor systems tend to have larger apertures for a given AOV than the lenses for smaller sensor systems.

Nikon is not much, if any better than Canon. At the pixel level, the D700 looks better than the 5DII, and the D300 is better than the 50D. At the same image size, the difference is not worth discussing.

For the DPReview test it seems to me that if it's noise reduction causing the loss of detail then the Nikon would have to be using tons of NR for both jpeg and RAW even at ISO 100, which I doubt is the case. The ISO 100 views for the D5000 show an image with significantly less detail than the Canon's, which could indicate a different issue with the test (optics or AF). I can't see much additional loss of detail going from ISO 100 to ISO 1600. In fact the change in detail looks to be pretty similar between the 500D, 450D and 5000D.

All of these cameras seem to produce pretty decent high ISO results IMO.

Rest assured: massive, heavy-duty noise-reduction...

Bob_A wrote in post #8081198
For the DPReview test it seems to me that if it's noise reduction causing the loss of detail then the Nikon would have to be using tons of NR for both jpeg and RAW even at ISO 100 which I doubt is the case.

...You seem to extract from the equation that inherently lower levels of sharpening applied during on-board processing (or off-board, via raw converter), do contribute with less noise and less detail. I do not see any major optical impairement on this test, though.

The hellacious amounts of noise reduction being employed, nowadays, by Nikon can also be verified in some Imaging Resource samples from Nikon D90, for instance.

The overall best performer, with best balance of detail and noise (albeit a relatively small or non-existent amount of Noise Reduction) is, without a doubt, the EOS 450D (or XSi), also shown in those crops.

PIX