5D Mark II VS D700 High ISO noise performance

ISO25600

http://dc.watch.impres​s.co.jp/cda/st.../d700​3_09l.jpg vs http://dc.watch.impres​s.co.jp/cda/st...5dmk2​3_10l.jpg

ISO12800

http://dc.watch.impres​s.co.jp/cda/st.../d700​3_08l.jpg vs http://dc.watch.impres​s.co.jp/cda/st...5dmk2​3_09l.jpg

ISO6400

http://dc.watch.impres​s.co.jp/cda/st.../d700​3_07l.jpg vs http://dc.watch.impres​s.co.jp/cda/st...5dmk2​3_08l.jpg

The banding in 5D Mark II highISO (12800 & 25600) quite concern me. Hope it will be resolved in production units.
I personally think the fine grained noise is more pleaseful in D700. It's more like film grain.

Interesting. What causes banding anyway?

I'm not sure where the banding comes from, but I was playing with that 5DII ISO25600 image yesterday (NR, resizing, sharpening, etc), and I noticed that the visible banding changed pretty significantly when the image was resized in Gimp vs. PS. It might be worthwhile to play around with different resizing options and see if the banding can be minimized.

Looking at 12800 the 5D retains apparent color better in the color palette because the noise reduction leaves fewer dark spots in the colors, but per pixel sharpness is not as good and chroma noise is more apparent. The D700's grains add texture which seems to generate artificial detail in certain areas. The 5D is more blotchy, keeping in line with all the previous Canon noise I've seen.

ISO25600 is stifling and I don't think either camera should offer it, the 5D being somewhat worse in appearance and retaining less detail.

I don't see alot of banding in the 12800 Canon. The nikons looked alot more blotchy especially in the gray parts of the pic.

form wrote in post #6576220
Looking at 12800 the 5D retains apparent color better in the color palette because the noise reduction leaves fewer dark spots in the colors, but per pixel sharpness is not as good and chroma noise is more apparent. The D700's grains add texture which seems to generate artificial detail in certain areas. The 5D is more blotchy, keeping in line with all the previous Canon noise I've seen.

ISO25600 is stifling and I don't think either camera should offer it, the 5D being somewhat worse in appearance and retaining less detail.

It'd help if the image was in focus you know... Until someone frames and focuses EXACTLY the same image on both cameras, there is no way to compare. From what I see, 5dmkii has less noise, but what noise it does have is chroma noise. If you enable high iso NR, it'll get rid of a lot of chroma noise, and if you take it one step further and make a noise profile, it should be a very clean image from either camera.

Honestly, looking at the ISO 25600 images I think the D700 ones look better... But until someone comes out with a test that everyone agrees on I won't make up my mind whether I'm gonna by one or the other. That is why I'm waiting for the dpreview comparisons...

But the scene looks well lit, not a situation anyone would use for high ISO shooting...
Less color noise on the D700, not worth mentioning in the ISO 6400 range. More pronounced on the higher range.

I know this is Canon on the Net and I own 2 Canon cameras but I'm very happy with the images from the Canon in comparison to the Nikon, 12800 and 6400 look very usable to me.

deeeez wrote in post #6579797
But the scene looks well lit, not a situation anyone would use for high ISO shooting...

People keep saying this, but I don't think it's ever been demonstrated or explained why light level significantly effects the look of a high iso exposure. I shoot about 90% of the time at ISO1600+ and I just don't see this phenomenon. Quality of light is certainly an issue, regardless of quantity. That said, it may be true that the quality of light may often be less interesting at low light levels. However, I just haven't seen any real evidence to suggest that an evenly lit seen at a low light level will produce a worse high ISO image than an evenly lit scene at a higher light level.

That being said, I've spent more time than I care to mention looking at these samples. The interesting thing that I notice is that (to me) the canon images look quite good up to 12800, but the noise level (and banding) jumps quite a bit going to 25600. I'm thinking that at the standard jpeg NR level, we're seeing a fair amount of in Camera processing (probably an unprecedented amount by Canon standards)...but the amount of detail retained seems to be quite high. It will be interesting to see how ACR handles the RAW files.

bacchanal wrote in post #6579975
People keep saying this, but I don't think it's ever been demonstrated or explained why light level significantly effects the look of a high iso exposure.

The more information you try to drag up from darkness, the more noise comes with it...

Less light means more amplification and an increase in the noise floor. It works exactly the same as having to turn up the volume on music that was recorded at too low an input level.

You can make the music louder but you also get an increase in hiss, rumble, etc... ie, more noise.

FlyingPhotog wrote in post #6579995
The more information you try to drag up from darkness, the more noise comes with it...

Less light means more amplification and an increase in the noise floor. It works exactly the same as having to turn up the volume on music that was recorded at too low an input level.

You can make the music louder but you also get an increase in hiss, rumble, etc... ie, more noise.

Right, but if the meter reading (in camera) and the ISO is the same for 2 exposures (only the aperture or shutter speed changes), how are you increasing the amplification?

An ISO test is an ISO test as far as I'm concerned (as long as there are at least some shadows).

For instance, what is the difference between:

f/1.4 1/100 ISO1600 - meter reading 0 (low light)
f/11 1/100 ISO1600 - meter reading 0 (decent light)

non-scientific example:

1/100 f/22 ISO1600 flash (580ex bounced) 1/1 power

And what does this prove ? Looks good I Guess

john salgado wrote in post #6580339
And what does this prove ? Looks good I Guess

It doesn't prove anything. It suggests that light level is not significant when comparing images at the same ISO value. Which would suggest that an ISO test in decent light is no more or less meaningful than an ISO test in low light. This probably wouldn't hold true as the light gets very dim and shutter speeds get very slow (1 sec plus).

bacchanal wrote in post #6580133
Right, but if the meter reading (in camera) and the ISO is the same for 2 exposures (only the aperture or shutter speed changes), how are you increasing the amplification?

An ISO test is an ISO test as far as I'm concerned (as long as there are at least some shadows).

For instance, what is the difference between:

f/1.4 1/100 ISO1600 - meter reading 0 (low light)
f/11 1/100 ISO1600 - meter reading 0 (decent light)

non-scientific example:

1/100 f/22 ISO1600 flash (580ex bounced) 1/1 power

1/100 f/2.8 ISO1600 flash (580ex bounced) 1/64 power

1/100 f/22 ISO1600 flash 1/1 100% crop

1/100 f/2.8 ISO1600 flash 1/64 100% crop

I follow you but now instead of aperture as a variable, use ISO...

(Nice choice of subject...)