5D MKIII vs D800

tccin3D wrote in post #14294616
Canon powershot s100 is very close to 5D3 with DR, based on DxO.

wow, either the procedures and rankings are really wacko .... or maybe I'll pick up an S100

False Alarm, I've checked some flickr shots and s100 looks terrible in low light. For a moment I wanted to have s100 and save money.

tccin3D wrote in post #14294616
Canon powershot s100 is very close to 5D3 with DR, based on DxO.

Whch isn't unreasonable - it very well could have such a dynamic range. After all, they're based on the same sensor technology. But the 5D3's sensor is ten times the size (or whatever) so it will have better IQ, despite having similar DR.

tccin3D wrote in post #14294772
False Alarm, I've checked some flickr shots and s100 looks terrible in low light. For a moment I wanted to have s100 and
save money.

OK. I'll cancel my S100 order - thx

Shadowblade wrote in post #14294413
Their ISO test makes perfect sense - it's the level at which you still retain 9 stops of dynamic range. This provides a level playing field - after all, it's easy enough for any given manufacturer to push 'native' ISO higher and higher, without regard for what sort of IQ this actually gives us. You could potentially push the original 5D to ISO 102400, but you'd get hardly any useable DR. Also, it accounts for differences between the stated ISO and the true, measured ISO - a camera manufacturer may state ISO '6400', when the actual ISO is closer to 4900, for instance, while another manufacturer's ISO '6400' may be really 5800. All else being equal, the former manufacturer's '6400' will look better than the latter's '6400', because one is actually ISO 4900, while the other is ISO 5800.

How does all this work with the fact that the 5DIII has a useable 25,000 ISO and the the D800 really doesn't? I've seen raw shots from both at 12,000 and I thought they were pretty close, but at 25,000 the 5DIII was better, imo. This coming from a D800 fan btw, haha.

It's fine to say the D800 has cleaner ISO from 100-800. I think that's awesome because we rarely go above 1600, but I still don't think the DXO tests appreciates Canon's superiority at 25,000. But I'm pretty ignorant on the subject obviously

snaphappyphotography wrote in post #14294849
How does all this work with the fact that the 5DIII has a useable 25,000 ISO and the the D800 really doesn't? I've seen raw shots from both at 12,000 and I thought they were pretty close, but at 25,000 the 5DIII was better, imo. This coming from a D800 fan btw, haha.

It's fine to say the D800 has cleaner ISO from 100-800. I think that's awesome because we rarely go above 1600, but I still don't think the DXO tests appreciates Canon's superiority at 25,000. But I'm pretty ignorant on the subject obviously

The DxOMark for ISO measures the point of (actual) ISO where the image's dynamic range drops below 9 stops. If the D800 scores 2800 and the 5D3 scores 2300, it just means that the D800 has 9 stops of DR at ISO 2800 (actual, not manufacturer-claimed) and the 5D3 has 9 stops of DR at ISO 2300 (actual).

This could be due to one of two things (and probably some of both):

1) At some point around the ISO 12000 mark, the D800's DR drops below that of the 5D2. At this stage, however, both sensors have below 9 stops of DR, so it doesn't register on the score. This would be somewhat expected, since the 5D3's ISO 25600 is native, whereas the D800's ISO 25600 is really ISO 6400 pushed two stops (thus sacrificing the two brightest stops at ISO 6400).

2) The images you're looking at are based on *claimed* ISOs of 25600, not actual ISO. It's possible - indeed, likely, given past record - that the Canon's ISO 25600 is actually around ISO 16000, while the Nikon's ISO 25600 is actually around ISO 22000. In that case, the Canon images, since they are taken at lower actual (not claimed) ISO, may look cleaner.

Also, it's interesting how that Nikon forums were all bashing DxOMark in the pre-Exmor era, but, since the Exmor, it's been the Canon forums doing it. The 'neutral' forums, meanwhile, have done little bashing, and just used the results. Fanboyism?

snaphappyphotography wrote in post #14294849
How does all this work with the fact that the 5DIII has a useable 25,000 ISO and the the D800 really doesn't? I've seen raw shots from both at 12,000 and I thought they were pretty close, but at 25,000 the 5DIII was better, imo. This coming from a D800 fan btw, haha.

right, the higher ISO tests favor Canon 5D3 but the fanboys will always jump in

watt100 wrote in post #14295028
right, the higher ISO tests favor Canon 5D3 but the fanboys will always jump in

That's because DxOMark doesn't measure ISO 25600, or 12800, or any other specific ISO number.

It sets a certain benchmark for IQ (in this case, dynamic range of 9 stops) and measures the *actual* (not claimed) ISO value at which each sensor dips below that threshold.

The D800 drops below the 9-stop threshold at ISO 2853, while the 5D3 drops below the same threshold at ISO 2293. Therefore, if 9 stops is your acceptable limit of S/N ratio, the D800 does better in that regard.

What it does not measure is the ISOs for other thresholds - say, 7 stops, or 6 stops, or 5 stops. It may well be that the D800 still beats the 5D3 at the 7-stop threshold (say, ISO 5000 vs ISO 4500) but the 5D3 jumps ahead at the 6-stop threshold (say, ISO 10000 vs ISO 10200) and further ahead at the 5-stop threshold (say, D800 dropping below 5 stops at ISO 16000, vs ISO 18000 for the 5D3). With these values (and I'm just using them as an example) it would indicate that the 5D3 may pull ahead at the very highest ISOs - but, if your limit of acceptable quality is 9 stops, the D800 allows you to shoot at a higher ISO while maintaining acceptable quality. If your limit is 7 stops, or 6 stops, or 5 stops, the margin may be less, eventually swinging in favour of the Canon. It would be useful, though, to have ISO scores at a number of different DR levels to compare sensors - I guess that's what the graphs are for.

watt100 wrote in post #14294667
wow, either the procedures and rankings are really wacko .... or maybe I'll pick up an S100

Interesting how every other engineer who has tested and published dynamic range ends up with the same relative measurements as DxOMark - the numbers themselves may differ (due to different thresholds being used) but the rankings and the differences between different cameras are consistent between testers.

Or is Nikon/Sony buying them all off, after Canon ran out of money to do so (coincidentally, on the same day as the Exmor's debut)?

Shadowblade wrote in post #14295000
the D800 has 9 stops of DR at ISO 2800 (actual, not manufacturer-claimed) and the 5D3 has 9 stops of DR at ISO 2300 (actual).

I can wrap my head around that, very impressive!

I actually just ordered from Amazon today, and since we have all of July off this year, I'm hoping it will arrive before August 1st. But I'm not entirely sure it will, haha, we'll have to wait and see

Shadowblade wrote in post #14295169
but the rankings and the differences between different cameras are consistent between testers.

interesting how the real world looks - have you ever been on a photowalk and compared pics?

notice how Canon images are sharper, colors, etc. look better?

watt100 wrote in post #14295202
interesting how the real world looks - have you ever been on a photowalk and compared pics?

Yes.

No.

Shadowblade wrote in post #14295211
Yes.

No.

well check it out!

the truth is out there!

Remember those old Nikon D90's ?

AngryCorgi wrote in post #14294322
Which is ALSO soft in the corners.

Touche