Printable View
A gentleman by the name of Ed Dale from Australia has posted some production Mark IV high ISO RAW files for inspection (over on DP Review). Here's a 100% crop of an ISO 12,800 image shot at f/2 and 1/400; Lightroom adjustments only. I think the fine detail is pretty darn good considering we're viewing an ISO 12,800 image at 100%. Haven't seen the D3s at the same ISO.
Here's the entire image.
Pretty impressive. When are they shipping?
Doug,
Pretty sharp details, indeed.
I trust he states that downloading and reposting images is ok ;)
Chas
Thanks Doug I'm sure we will be seeing more in the near future !!
Not bad however this isn't very low light, D3S is a different class and will easily beat this one in noise department see samples below, low light and high ISO. Can you send a link to the original RAW file?
D3S samples:
http://www.robgalbraith.com/bins/con...=7-10045-10329
ISO 12,800
http://ftp.robgalbraith.com/public_f...12800_Trio.jpg
I'd like to see a D3S vs 1D4 using a 200/2.0 lens.
Here's what he said: "And here are the RAW files for you to play with (for non-commercial use only please!)"Quote:
Originally Posted by Charles Glatzer
I think what I did with the image is well within the 'rules.' Here's a link to the thread. And here's a link to the post where he provides links to the RAW files. I haven't had a chance to look at the ISO 6400 images yet. If ISO 3200 looks 2 stops better than this, I'll be one happy camper.
Quote:
Originally Posted by Doug Brown
Thanks Doug, will give these a try with DPP 3.7.2
IQ is not an issue for me, the only thing I care about is AF!!!!! :)
Galbraith's image has had PS smart sharpening and high ISO NR applied to it. The Mark IV image has had neither (other than default Lightroom NR). Galbraith's photo was also taken at f/2.8 and 1/1600 vs. f/2 and 1/400 ( a difference of 3 stops). All things considered I don't see that much of a difference between the two.
Quote:
Originally Posted by Doug Brown
True but I don't think there is any way you can produce samples as clean as those with the image you posted, you can try applying NR and see what you get besides LR already applies NR and you can't really turn it off (you can with DPP). The noise is very coarse and no amount of NR will get rid of those color blotches without destroying the details. Nikon shot has NR but very fine detail is preserved, Galbraith photo has deep deep shadows and bright highlights at the same time the sample above is uniformly lit so not the best case for noise comparison, any way we have to wait and see, but pixel size difference in this case is so great (more than 2 times) there is really no way for Canon to fill the gap with normal improvements in efficiency and such...MKIV sensor data is not available yet either.
For low light Nikon is the obvious choice but for telephoto reach and hopefully AF I think Canon will have an edge. To me Nikon have decided to target sports' shooters who work at close range at night and in poorly lit stadiums. Canon has concentrated on increasing resolution for telephoto reach which is best for wildlife and birds...
BTW, Galbraith seems to have switched to Nikon entirely and quietly, I doubt if he will say anything about AF at all or at least in near future. This time the burden is on the bird shooters to test the AF!!! :)
Arash -
I agree that to me AF on the new MK4 will be a key feature
But i'm not convinced that you're argument is correct
If you look at the more "sensitive" cross-type AF points on the new MK4, not much has changed from the Mk3 for those that use f4 L lenses. Here's a link that addresses the issue: http://cameragossip.blogspot.com/200...af-module.html
All the improved AF features on the Mk 4 really only apply to use with f2.8 lenses. So those using the 300mm or 400mm f2.8L will likely be very happy with the Mk4. With an f4L lens, only the centre AF point will function as a cross-type AF sensor, all remaining 44 AF points are horizontal-line AF sensors at best.
The 7D has all 19 AF points function as cross-type AF sensors for lens combo's up to f5.6 (ie- a f4L lens x 1.4 TC). I think many who have both the 7D and Mk 3 will privately admit that the 7D has superior AF for BIF with f4 or higher lenses (that was my experience last week in Florida). Most AF sensors on the high-end Nikons are also cross-type AF sensors up to at least f5.6
There are many great-looking features on the Mk4, but i'm not sure that improved AF with an f4 L lens will be one of them.
Right Peter, on paper MKIV Af points will not operate in cross type mode with popular birding lenses such as 500 f/4, 600 f/4 or 800 f/5.6 naked and with TC, but the general understanding is Canon uses faster hardware and better tracking algorithms in 1 series cameras that compensates for the lack of cross-type sensitivity and delivers better overall AF. Any ways, I am very eager too to hear about the AF performance of this camera in the field compared to both 7D and MKIII, stakes are really high for Canon and this time they only have one shot at it.
Quote:
Originally Posted by Peter Hawrylyshyn
Just upgraded to DPP 3.7.3. Here's another 100% crop processed in DPP only. Looks better than LR 2.6.
The crop processed in DPP looks really good. Seems to look promising for images at lower ISOs.
Forgot a step in the processing. Here's the complete processed image (DPP to CS4). I can tell you that the full-sized image looks very good at 13x19.
And here's a 100% crop of the processed image. Remember, this is ISO 12,800!
Doug, thanks for posting this. I just am on the edge of my seat waiting for my MK4! These images look fantastic to me. Certainly better than the Mark 2 I will be upgrading from. The 7d we bought in October will be my wife's camera. If the comments on the pre production bodies are correct the AF is equally impressive as well.
Are we calling this usable???????????
Do you feel that it's not?
I sure don't :cool: Lotsa chroma noise. Why would I want to do that anyway??
I'm not suggesting that I would want to shoot at ISO 12,800. And if I did I wouldn't print at the equivalent of 100% magnification (something like 3 feet by 4 1/2 feet). But this shot looks very good at 13x19. And I think it also looks very good sized for web use.
In all honesty I don't see myself going above ISO 3200 (I suspect I'll do 95% of my shooting at ISO 1600 or below). That's 2-3 stops less noise than this image has. I'm pretty excited about the possibilities that this image hints at. And that's the reason why I posted it.
December 07, D3, actual pixels, ISO 6400, no processing. I call this usable.
Date: 12/2/07
Time: 7:04:10 AM
Model: NIKON D3
Firmware: Ver.1.00
Frame #: 42
Lens (mm): 650
ISO: 6400
Aperture: 6.7
Shutter: 1/640
I get the feeling we sometimes forget about what ISO we are talking here....
Those 12800 are way better than a Mark IIn's ISO 3200...
I totally agree with Doug about the ISO range we really use. To me that's also <1600. If I get good IQ in this area I will be more than happy. When ISO >12800 is needed I am usually in my bed sleeping ;)
The Mark IV just cannot have a better or equal noise performance than the D3. It's 16.1 mpix on a 1.3 crop size sensor vs 12mpix on FF, we shoudln't forget that fact.
I also agree that the AF will be the main success criteria. I am quite sure it will have a pretty impressive performance.
We just have to wait till our orders arrive to evaluate the performance ourselfes. However, I have serious doubts that I will touch ISO 12800 at all.....
Shouldn't Artie be allowed now to post some images!?
@ Fabs, that a pretty impressive sample.
With all due respect the Heron image is unusable for anything other then further manipulation with filters for some outside of the box effects. There is no fine feather detail, the whites are shot, the noise in the blacks can never be suppressed with out further muddying them up and I don't see any inherent sharpness that can be improved through further editing. In addition your comparing the "usable shot" to one that was taken at twice the ISO and still blows that one away in every aspect.
As someone who shoots sporting events as much as possible I am excited to see the samples from the MKIV at those ISO's. Sports don't require the fine details that birders require and I am betting Canon and Nikons revenue from media types exceeds wildlife / bird generated sales.
You can definitely make usable images with Nikon D700/D3 let alone D3S at ISO 6400 in dark and challenging conditions, you can even process and heavily crop the photo. It is nice to have very high ISO capability for birding as well, even though you may not use it as often.
Here is an example, it is not a super flight shot but it is still a nice dive and reminds you of the moment.
http://www.stanford.edu/%7Eahazeghi/...high%20iso.jpg
D700 ISO 6400 + 1EV pushing in post f/7.1 1/2500 sec 550mm manual exposure.
To give you an idea of how dark it was and how tight the crop is
http://www.stanford.edu/%7Eahazeghi/...l%20screen.jpg
Full frame no adjustment, manual exposure at sunset.
100% crop after 1EV pushing.
http://www.stanford.edu/%7Eahazeghi/...rop%20d700.jpg
I agree 12,800 is too much but a professional $5,000 camera should IMHO deliver clean enough images at 6400. It is Canon's job to ensure it balances pixel density with noise such that high ISO performance is acceptable. When Nikon claims it has "ISO 12,800" it means business and something usable. Canon claims ISO 12,800 on 7D too but we all know what it looks like. Nikon don't even claim ISO 3200 for their flagship D3X, they call it expansion mode.
We should wait and see how the field samples look like, overall if it has good IQ up to 3200 and really improved AF it will fair well for wildlife and birding and will complement 7D nicely.
I am sure Doug's own carefully exposed and processed samples will be much better than the mediocre samples posted on the internet so far.
Let's turn this into a learning exercise instead of a Canon vs. Nikon competition - at least for me! :DQuote:
Originally Posted by Doug Brown
With the 7D I am finding that I am shooting a lot at 1600 because of low light and wanting to maintain a SS of at least 1/1600 and preferably 1/2500. Yes, they are noisy and Topaz Denoise is doing a very good job.
With the 1D4 from what we are seeing in these images 3200 definitely and 6400 when necessary are certainly achievable and the noise is certainly removable.
Doug, since I am quoting you, why not shoot at the higher ISO? Wouldn't using a higher ISO, everything else being equal, allow you to use a smaller aperture resulting in a larger DOF field? Wouldn't a larger DOF allow 1) a more forgiving capture, e.g., if you are close to the eye and not spot on the eye there is a greater chance that the eye will be tack sharp, and 2) by using a higher ISO at, e.g., 1/2500 you can use a smaller aperture resulting in more of the total bird being tack sharp.
Why the hesitation to always shoot at 1600 or 3200 simply to achieve the ability to shoot BIF at f/8 or higher so as to obtain wider DOFs at the same SS?
Jay my purpose was not Canon vs Nikon debate, I personally use Canon gear for majority of birding work and am considering a MKIV after I learn more about its performance.
The purpose was to show ISOs higher than 1600 or even 3200 are also useful for birding in some conditions and therefore nice to have.
Quote:
Originally Posted by Jay Gould
Good question Jay! I think that 'ideal' camera settings are a moving target that change as technology improves. For hand held flight photographry, where shutter speed is extremely important, I boost ISO to keep the shutter speed in my target zone. Improved high ISO performance will get me that many more keepers. I don't normally need to get much more DOF than f/8 offers, but I do understand your point. How you choose to divide up available light with a given body, and how far you're willing to push ISO are two concepts that I love to experiment with! It takes quite a bit of time with a given body to balance everything out.Quote:
Originally Posted by Jay Gould
I was just out in California photographing with Jim Salywoda. One sunny morning I suggested that we use ISO 1600 for all of our images taken over the next hour (even though ISO 400 was more than enough for what we were doing). I spent the next hour experimenting with how to use the additional stops of light. Image quality was pretty good too.
Here's another processed ISO 12,800 image. Full frame but resized for the web. This was taken at f/2, 1/200 and hand held with a 135mm lens.
I would be really interesting in seeing some ISO 800-1600 images
My understanding is that chromanance noise can be removed without affecting much detail, but luminance noise is a lot harder. Thanks for the links to the raw images- I'll have a play. I am interested mainly because the mark IV will be my first pro-level camera.Quote:
Originally Posted by arash_hazeghi
You and me both!Quote:
Originally Posted by Jan Wegener
May I add ISO 400 to the wishlist? :)Quote:
Originally Posted by Doug Brown
Mike, your opinions and mine are usually totally opposite :)Quote:
Originally Posted by Mike Tracy
Not surprised of your answer here, what else is new?
I think we need to be reminded that the 1D4 images posted here were taken in poorly lit conditions. The photos I've seen at ISO12,800 in good light are very usable.
Hi Mike, how about you download and process it? You may be surprised. You don't even need the raw file for that purpose.Quote:
Originally Posted by Mike Tracy
But if you want to showcase your camera's high ISO performance, images taken in poorly lit conditions should be the ones to show, IMO.Quote:
Originally Posted by Blake Cook
A little reality (and some simple physics):
For a n 18% gray level, normal metering, exposure at ISO 200 there are 1300 photons/square micron over an idealized green passband (less in red and blue with sunlight) at the focal plane in the camer (above the sensor and its filters). The blur, Bayer, and IR filter transmissions reduces that as does the quantum efficiency and fill factor to about 20% of that, so about 260 photons/square micron.
Next figure the ISO effect: the photons/sq micron = 260 * 200/ISO.
Thus ISO 12800 will have about 260*200/12800 = 4.1 photons/sq micron.
The Nikon D3 has 8.46 micron pixels, so would get about 4.1*8.46*8.46 = 293 photons/pixel, ISO 12800, 18% gray
The Canon 1D4 has 5.7 micron pixels, giving about 4.1*5.7*5.7 = 133 photons/pixel, ISO 12800, 18% gray.
Noise is square root the number of photons, so no matter how you cut it, larger pixels on a larger sensor will collect more photons and have better high ISO performance (given similar technology in the sensors).
So the signal-to-noise ratio (S/N), what we perceive in our images as noise, is pretty low: S/N =sqrt(293) = 17 for the D3 18% gray, and 11.5 for the 1D4 pixels. Then consider a few stops below 18% gray and the shadows get lost in noise.
Most of the recent cameras are performing very well at the high signal levels and the noise is dominated by the noise from photons. What the camera manufacturers have done to be able to achieve is not so much improvements in photon detection (it has improved a little), but reduction in fixed pattern and read noise at the low end. The images shown here with the D3 and 1D4 illustrate a remarkable lack of banding at the low level. Banding, or fixed pattern noise is very annoying to the average viewer, but random noise is more acceptable (perhaps for that films look).
All this, combined with better algorithms for processing the Bayer arrays is leading to better high ISO performance. However, except for the fixed pattern noise, throw the new algorithms at images from your older cameras and the will come out better.
The bottom line is I see a nice improvement in fixed pattern noise with this next generation of cameras, but the high ISO claims seem bogus as photon collection efficiencies have not improved by factors of 2, 4, 8+ as is implied. The basic physics still tells the real story.
Roger
Agreed Desmond. And I think these images do showcase the Mark IV's high ISO performance. I'm looking forward to getting one in my hands in the next couple of weeks and doing some testing of my own.
Moving bird, actual pixels, noise reduction (one pass) linear contrast and sharpening, captured at 1/640 of a second ISO 6400 in D3, two technology years ago.
That doesn't make me always right, you always wrong or vice versa. We just see some things differently and when I dissent it raises your ire though that's not my intention. Since you practice a art and am in a position of leadership with this fine forum you should be open to ideas that don't always concur with your own without throwing barbs.Quote:
Originally Posted by Fabs Forns
Merry Christmas to you and Mr. Forns :).
... didn't see the barb and Merry Christmas to you Mike !!!!
Me too, hopefully! Happy HolidaysQuote:
John Chardine: I am interested mainly because the mark IV will be my first pro-level camera.
True, but the images are being compared to well lit Nikon shots. I would certainly be interested in seeing the Heron processed by Fabs. It certainly looks beyond recovery by me.Quote:
Originally Posted by Desmond Chan
Oops, I see she did clean it up. I'm impressed she got that much out of it, however I still wouldn't consider it a keeper.
Aaah...which one you were referring to:Quote:
Originally Posted by Blake Cook
http://imaging.nikon.com/products/im...g/pic_006b.jpg
http://imaging.nikon.com/products/im...g/pic_001b.jpg
http://imaging.nikon.com/products/im...d3s/sample.htm
http://www.robgalbraith.com/bins/con...=7-10045-10329
There're more details yet to be recovered in the whites.Quote:
Oops, I see she did clean it up. I'm impressed she got that much out of it, however I still wouldn't consider it a keeper.
I look forward to seeing that.Quote:
Originally Posted by Desmond Chan
Impressive photos. Perhaps we'll see similar shots from the MkIV community once they've had their cameras for more than the time it takes to give the battery its first charge.
[quote=Fabs Forns;410120]December 07, D3, actual pixels, ISO 6400, no processing. I call this usable.
Fabs, Is the Great Blue Heron image at 100%?
