The GetDPI Photography Forum

Great to see you here. Join our insightful photographic forum today and start tapping into a huge wealth of photographic knowledge. Completing our simple registration process will allow you to gain access to exclusive content, add your own topics and posts, share your work and connect with other members through your own private inbox! And don’t forget to say hi!

A few photos from Canon 5D Mark II

Jeff Turner

Member
Hey Guy and good morning! I included a portion of the plate so that viewers could evaluate noise not only in the shadows as per Jack's request, but also in mid and lighter tones. I agree with your assessment on the noise; 1600 starts to look a little noisy and I would probably only use it if I had too. For my work, 3200 and 6400 are out of the question.

Oh...and I did forget to mention that I turned off all noise reduction in DPP, and as far as I can tell so far, there is no user adjustable NR in camera except for long exposures.
 

dseelig

Member
I will say the shadow noise in the mk 11 looks much better for me then the nikon d3 does at 6400 . I think the point is the mk11 is usable for street shooting at 6400 not ad work. I was just in New Orleans, I used my m8 for low light work and the mk11 for no light work. Will post photos in a few days. By the way for me it is not a replacement for a 1ds mk111 just another tool for other work. I.E. Focus responsiveness.
 

Jeff Turner

Member
I will say the shadow noise in the mk 11 looks much better for me then the nikon d3 does at 6400 . I think the point is the mk11 is usable for street shooting at 6400 not ad work. I was just in New Orleans, I used my m8 for low light work and the mk11 for no light work. Will post photos in a few days. By the way for me it is not a replacement for a 1ds mk111 just another tool for other work. I.E. Focus responsiveness.
I agree that the Nikon D3 is certainly no better than the 5dII, and the 5DII may have the edge, at least based on my experience with my D3, though I may do a head to head comparison with the same shot. I don't have matching 135 mm lenses for the two, but will slap the Zeiss ZF 100 Makro on both and shoot this same image again tonight. Direct comparison will be interesting.
 

Jack

Sr. Administrator
Staff member
...and finally ISO 3200 and 6400....

Oh, by-the-way, these are screen shots of 100% crops.

Cheers, Jack!
Thanks for these Jeff! I would say that 6400 is usable for when I'd want to use it -- I took the liberty of copying the 6400 shot and running DFine on it... Granted, best to run it on the actual file, but this looks very promising to me! I linked the before and after below to make them easier to compare:
 

jonboring

New member
Several of us last year on FM did some detailed raw conversion tests for the 5D and we concluded DPP 3.1 generated files were superior to ARC, C1 and just about everything else. Historically DPP sucked but with 3.1 the raw conversion improved considerably. I don't use it for anything but raw conversion, generate a 16bit TIF and pull it into CS3. Of course, with the 5DII we need to redo these tests.
 

Jack

Sr. Administrator
Staff member
Here are some new comparison crops as run on the 6400 tiff. Folks, I think we have a winner for high ISO work :)

Before and after DFine:
 

Jeff Turner

Member
Not too shabby Jack. That does look pretty good! I wish I had the rest of my software here in Port Angeles, WA....so that I could run similar tests. When I am home....or I could just keep shipping files to you....LOL

Cheers!
 

bradhusick

Active member
I just took delivery of my new 5Dm2 and I can agree that it is truly remarkable. It's easily as good as the D700 (which I owned for a short time and loved) and the files are huge and detailed, which is something the D700 didn't really give me. ISO 3200 is super and cleans and sharpens with Noise Ninja like crazy. I will try to post some samples soon.
 
A

asabet

Guest
There are 5D II vs D3/D700 super high ISO comparisons popping up all over the place, and my impression is that it's close enough that it doesn't matter which one is the high ISO "champ".

I think this was to be expected though, since the 1Ds III (when processed carefully from RAW) was also very close to the D3 at high ISO, and the 5D II has a slightly improved 1Ds III sensor. In fact, the dxomark tests showed more of an improvement than I expected from the modifications Chuck Westfall described.

One of the things that makes the 5D II high ISO ability more obvious than that of the 1Ds III is the ability to shoot 10MP sRAW with the 5D II. This means that even someone who has no processing skills can fire off an sRAW1 capture and get similar results to a native D3/D700 file. Cool stuff!
 

Jack

Sr. Administrator
Staff member
Hi Chris:

The second set I posted are not screen shots, but jpeg conversions directly off the tiff, so are the best I can show you online...

That said, yes *ALL* NR tools reduce detail to some degree. In addition they "smear" color. The bigger issue here is how much color-channel smearing goes on beforehand during raw conversion. IMO with AA-filtered DSLR's it is a lot -- and thus the additional from the NR tool is usually not all that bothersome as it is only adding to what is already there. This smearing is what renders the waxy or plastic color effect many refer to when comparing a typical Nikon or Canon shot to a MF DB or M8/DMR non-AA capture.

When you use the NR tools on a non-AA sensor, the color smearing is fairly notable (and most of us don't like it) and IMO gives an appearance of even less detail -- probably not the case in reality, but it's what it looks like to me at least. However, *most* of the time, cleaning up the noise and getting the color smearing is preferable to not cleaning up the noise at all -- though I will go on record stating that any NR routine definitely detracts from overall file detail, so finding a happy medium on the amount applied is key.

The things I like about DFine are 1) it has a very intuitive UI, 2) it allows you to easily adjust the Luminance noise reduction (detail) separate from the color noise (smearing), and 3) it has an option to output the NR as a layer over the original in CS, very beneficial for further tweaking and masking. That said, I think Noise Ninja and NeatImage are also a very nice tools -- my recommendation is to download and compare the trial versions of all three to see which ultimately suits your workflow best.
 

mark1958

Member
I have noise ninja and as soon as I have some time to try it out--- going to play with Dfine. I really have only used Noise ninja with my LX-3 or DLUX-2 in the past. I rarely have used it with my canon images because I rarely go beyond a specific high iso (which I consider a cutoff) if I can avoid it. In cases where i use the higher isos it has usually been for the kids events and they only want small prints
 

robertwright

New member
another thing about noise ninja that I am afraid to admit is that I "just" discovered that you can control the noise reduction per channel, and imo this really sets it apart, (not sure if dfine has that) which makes it really easy to dial back until you are just filtering the noise and not the detail.
 

Jack

Sr. Administrator
Staff member
Hi Robert:

Yes, in DFine you have the options to use the whole image (what I did in the examples above), individual color channels, luminance only or chrominance only. You can then set the L and C NR values manually, have DFine do it automatically (what I did for the above), and then adjust those auto settings after the fact if desired. Finally, you can "brush" the NR levels chosen to apply it only to the areas of the image you want it on.
 

ptomsu

Workshop Member
Hello Peter! Do you find this to be generally true regardless of the proprietary file format (NEF vs CR2), or is your observation apply just to Canon CR2 files?

Thanks!!
Sorry to answer after such a long time!

I find this a general observation, so for all RAW file formats I am using, which are, NEF, CRF and DNG.
 

mark1958

Member
I spent about 2 hours after I downloaded Jeff's iso6400 image comparing Noise Ninja and DFine. I compared different color vs luminance noise reduction settings to keep the detail of the end of the banana as close as possible using the two programs. I have to say that I find that DFine did a better job of reducing noise with less degradation and artifacts on the 5D file. The red noisy area of the apple and the left lower portion of the image--- especially the shadow region were less noisy with DFine while both had the same detail. I lost more detail with NoiseNinja to get the noise to a a comparable level in my comparisons. I also fine DFine a bit easier to use. I have to admit, I am very impressed. I will try some other high iso images in the near future.
 
Top