Getting to like my D3S....

[anglefire]

Jonathan, I'm not chicken - and the sample was one I found quickly - the thread I found it on (not forgetting that the sample have been taken through NR reduction in CS4 - so is not SOOC).

I've posted a shot on here before that I took at ISO6400 that was pretty clean.

But really, is it worth it? The 1DMkIV which I don't have is pretty close to the latest Nikons. Look, I've already said the Nikons are ahead of Canon? Which is what most pundits believe. So what is the point!

Though it would be an interesting exercise to compare like with like. I.e. same venue same subject. That would be a true test.

[rksmith51]

The samples I found it looks like the D3s is still miles ahead at the high end, problem is you don't know how the test was done.

http://www.photographybay.com/2010/02/10/nikon-d3s-vs-canon-1d-mark-iv-iso-comparison-part-ii/

Bob

[Jonathan]

Dunno.  Testing method seems pretty clear to me.

Shows that Canon's 102K is really pretty silly.  In fact I can't see an ISO setting where the Nikon doesn't outperform the Canon in this test as the files are presented.

HOWEVER, the Canon has more pixels so viewing at 100% makes for invalid comparisons.  The only real test would be to target output at a particular size and either uprez or downrez.  But the more s/w you get in the way the less scientific it gets.

>> But really, is it worth it? The 1DMkIV which I don't have is pretty close to the latest Nikons. Look, I've already said the Nikons are ahead of Canon? Which is what most pundits believe. So what is the point!

None really.  I just find it interesting all the people really keen to say "oh they are all pretty similar" and shy away from head to heads.  Nobody really does proper tests any more.  I would seriously love to know how a D3000 stacks up against a D3 (and especially against a D2).

[Graham]

   Well folks. I just got a call to go and photograph this lot! (In about an hour from now. )

             http://www.youtube.com/watch?v=Gd-Tq6eVhV8&feature=related

      I'll let you know how the D3s performs.
      Good job I like "Da Blooze"!
                                                                   Graham.

[anglefire]

JR, the Canon 102K is pretty crap. But the differences in size does make it hard to compare properly. (not the last one, the canon just looks bad!)

If I had the time, I'd download the full size files and compare. But at the moment i don't.

[anglefire]

I've found some time.

I've loaded both images at ISO12800 and converted by DPP and NX. I've cropped the Nikon file to roughly the same size as the Canon.

I was going to say the noise is similar, but the Nikon is clearly more pleasing to the eye. That was at full screen, with the images side by side at about the same size.

Zooming in so that the church tower, for example is a similar size on the screen and the noise on the Canon is less pronounced. And more detail - which is not surprising given the additional pixes covering the area.

But the biggest thing that I notice is that the Nikon image seems a lot sharper. That is true of all of them from scanning the thumbnails. But again, at the same size on screen, the Canon is actually the better image (Ignoring colour). But more pixels explains it.

Colour. Which one is the right colour? The Canon is muted in colour - particularly the building adjacent the church tower. The Nikon is much redder and brighter. Its as if the metering is not the same - which is probably true. Theoretically time is time, so the shutter speeds (1/5" in each case) should be the same, but the aperture may be actually different. Both say f8.

So. My conclusion. Un-scientific that it is, is that The Canon is better if the image is looked at, at a given size. At 100%, then the Nikon is better. 

So if printed out at the same size, the Canon would probably look better WRT noise. WRT colour. Hard call. They are different. What is true is not known.

But since I shoot Canon, I say the Canon wins.

I fully expect to be told I'm talking B *&*& ks, but I can live with that.

Of course, I don't have a MkIV, JR has a D3S. So, who can actually get the lower noise image is not me. 

[Jonathan]

Sorry.....meant to reply to this since you went to a lot of effort but other stuff got in the way....

Interesting reading and it seems pretty fair.  Actually on a second look, I'd say that the ISOs are actually slightly different.  With the same exposure etc the Canon files tend to be a little lighter.  This could be a bias in the tone curve or that Canon are slightly more generous than Nikon with the ISO.  DSLRs often disagree with studio meters.

Anyway, on to more unscientific testing.  I was at a party last night in a DARK warehouse in Fulham.  For complicated reasons involving a h-u-g-e wedding my assistant Mike went along for an hour before me and then we shot for a couple of hours side by side.  Mike has a 5DMkII and I was using my D3S.  Since they were real pictures for a real client we couldn't go crazy.  I shot a couple at 8K, 5 at 12,800 and all the rest at 6,400 or below.  Mike shot mostly at 2,500.  Most of my shots and all of Mike's were with flash.

Things I learnt:
(Remember the 5D MkII is prosumer and ready for refresh - D3S is state of the art pro level)

1. Both cameras were amazingly good in very difficult situations.
2. On the LCD on the back at 100%, the Canon files looked better 90% of the time (I stopped showing Mike 100% zooms of mine after a bit....)
3. Technique plays a bit part because the position on the histogram is crucial.  You really really want to expose as far to the right as you can because noise in the top 10% is significantly lower than in the next 10%.
4. Shooting in tough conditions after a 12 hour shoot and 3 cans of Red Bull is not recommended.
5. The Canon is really amazingly surprisingly good.  At 2,500 with flash you can see every hair and more facial details than most ladies would like.  However as soon as you roll off into even very light shadow (edge of jaw line) the noise ramps up very quickly.
6. The D3S just kills the Canon.  It's not even close (though see the bit at the top about last year's prosumer vs this year's pro).  My files are cleaner in the subject area at 6,400 than Mikes at 2,500.  In the shadows it's not even funny.

[BTW the reason I was running higher ISO was not just because I could - I like to let my backgrounds bleed a little light in dance shots - Mike's are more "dancer + disco light + black".  Both look pretty nice.]

[BTW2 shooting on a D2X I'd have done this very differently and the pics would look different.  That would have been very very hard after a long day.]

[rksmith51]

It will be interesting to see some of those shots Jonathan, I took an 8 second shot of the stars at the side of the house tonight and it came out quite well after a few little tweeks

http://www.flickr.com/photos/rkspics/5027457444/sizes/o/in/photostream/

[anglefire]

An interesting and fair comparison Jonathan. Funnily enough, the rumours are suggesting that the 5D will be refreshed before the 1DsMkIv is released - hopefully with a decent AF system and not one that is at least 5 years old.

I went to Quarry Bank Mill near Manchester Airport about a month ago. Some parts of the Mill were pretty dark - not as dark as it sounds you were in JR, as you could see - but I was shooting ISO2500 and 1/10" @ f/5.6. And my MkIII wasn't bad. I was pleased. I'd prefer a MkIv - but you can't have it all - As my wife keeps telling me.