M8 vs Nikon D700

[AlanG]

Originally Posted by ammitsboel

Well, I haven't tried it so I don't know how it performs.
But to me the idea does not sound very clever.
It sounds like fixing one problem by creating a new...

What new problem? One just has to look at the resulting picture and see that it is better. Isn't visual communication what photography is about?

[sdai]

I agree with Alan ... as long as the result looks better then there's no problem.

Manufacturers are keen on in camera correction because they have the optical formula and bench test results when coupled with their cameras. This will certainly make third party lenses ... eg. Sigma, Tamron, Tokina, Zeiss, CV, etc look worse because they're not supported.

That's not to say I'm against in camera correction but I do prefer computer processing for many more reasons, it's quite easy, once grasped and quickly become second nature.

On the other hands, lens artifacts such as vignetting, flare,etc can actually produce some dramatic effects and are desired from time to time. Journalists, photo editors add it via Photoshop before going to press. You could see this on New York Times, AP wire, Reuters, Getty ... and many more places.

[stunsworth]

I'm certain that the distortion corrected by DXO means that the resulting image is better than the original, but the real question is, is it better - or as good as - a photograph taken with a lens that doesn't need correcting?

I think this is the heart of the concern of those of us who've expressed doubts about software correction.

[lct]

'Better' is a matter of tastes. The character of a (good) lens lies in a subtle blend of qualities and weaknesses which tends to disappear with those softwares.

[sdai]

Originally Posted by stunsworth

but the real question is, is it better - or as good as - a photograph taken with a lens that doesn't need correcting?

That's a good question, Steve ... a sick person can't be healthier than a healthy person even when he is "completely" cured after treatment. Spending your money on healthy food/lifestyle makes more sense than spending them on pills/drugs.

[ho_co]

Originally Posted by stunsworth

... the real question is, is it better - or as good as - a photograph taken with a lens that doesn't need correcting? ...

Simon has a good analogy, but as you said, your question is at the heart of the matter.

Seems to me that you get the best lens you can and then correct what you need to.

If a lens has no distortion or lateral chromatic aberration, then you don't correct for those. If the D700 doesn't see CA, it won't correct for it.

I think we won't need worry too much about the matter until Holga introduces built-in aberration correction.

[AlanG]

Maybe some viewers wish to actually see an example. So I posted one at the site below:

PhotoShelter - Gallery Password Required

The password is :Leica

The example is a typical architectural shot photographed with a 5D, 16-35 series 1 lens at f10 and 17mm and shows why I like to use DXO. It is unretouched and unedited - a simple straight conversion.

Both were identically processed in DXO. The "corrected" version has c/a, vignetting, distortion, and lens softness correction "on" for the default values. The "Uncorrected" example has these turned off.

This example clearly shows the barrel distortion on the right side. There is not much c/a but if you look closely you'll see some in the uncrorrected version. The vignetting is pretty obvious.

You can view on screen and use the arrows to switch beteween photos. or download the hi res jpegs to examine the detail. (Almost 7 megs each) Please do not repost or use the hi-res images for anything. (I did not want to risk altering the image by adding a watermark.) The photos will only be on-line for a few days.

[sdai]

Originally Posted by ho_co

If the D700 doesn't see CA, it won't correct for it.

Howard, by default the D700 can't "see" CA by itself and the engineer has to program the camera to cope with it, this will need the exact geometric mapping of the specific lens' characteristics across the field, then the software can add differential geometric distortion into the color components of the image by canceling the differences in image heights of the red, green and blue color components in the picture.

So first of all, the lens needs to be recognized by the camera, secondly. the camera must be programmed to correct that specific lens.

This is totally contrary to the Photoshop approach, when you can actually see the nasty stuff and decide by yourself, where, how, and to what extent you want/need to correct it.

[sdai]

Alan, your sample is great because the 16-35 is a bad "example" to boot ...

For all the work you do, you deserve a better one such as the 14-24 or the Zeiss 16-35 ZA which will be introduced with the "A900" this fall.

Cheers,

[zlatkob]

Wow, 16-35 plus DxO looks excellent!

[AlanG]

Originally Posted by sdai

Alan, your sample is great because the 16-35 is a bad "example" to boot ...

For all the work you do, you deserve a better one such as the 14-24 or the Zeiss 16-35 ZA which will be introduced with the "A900" this fall.

Cheers,

Even the medium format digital cameras are very limited when it comes to wide angle. (Or I would have bought one long ago.) I am actually using stitching when I need much better quality and real wide. Here is an example. I regularly produce 200 meg 8 bit files this way but of course I could go higher.

[AlanG]

Originally Posted by zlatkob

Wow, 16-35 plus DxO looks excellent!

Thanks. I was trying to say that but a picture really is worth 1000 words isn't it?

[pierovitch]

One area that in camera processing or correction can cause problems is scientific / forensic documentation. The Nikon D3 and possibly the D700 can introduce artifacts from noise reduction algorithms which are not documented by Nikon because of their need to protect their algorithms. It is not visible in normal use but has been shown in astronomical applications. On the other hand I have no problem using CA correction on a microscope image to make the system perform better as long as the correction is accurate and documented.

[ho_co]

Originally Posted by sdai

by default the D700 can't "see" CA by itself ...
So first of all, the lens needs to be recognized by the camera, secondly. the camera must be programmed to correct that specific lens.

Interesting, Simon. Please give us a link to your source.

As I read it, the D700 brochure (available thru my link above) says precisely the opposite.

So you're saying that I would have to send all my 30-year-old Nikkors back to Nikon to have some kind of lens recognition added?

Or does the camera already have the functionality for all previous lenses, and I just need to select a particular lens from the lens menu?

Or are you saying that I need to send the camera as well as the older lenses back for installation of the firmware? And then if I buy another used lens, I have to send it and the camera in again?

This all sounds very unlike Nikon to me; and if any of that is necessary, I think they could be sued for misleading promotional material.

[ho_co]

Originally Posted by pierovitch

One area that in camera processing or correction can cause problems is scientific / forensic documentation. The Nikon D3 and possibly the D700 can introduce artifacts from noise reduction algorithms ...

Pierre--
Interesting point.

I know Nikon introduced forensic software for the D2 and D200, so I guess that the artifacts you mention still come through that firmware/software combination?

Is that also the case when using Canon's forensic software on the 5D and 1D series?

If so, does that render the cameras unusable for forensic purposes?

(I'm not sure what either company's forensic software does aside from making it possible to show whether a file was modified after it was shot.)

[mooky]

This is all so,,,so,,,so tech-weenee-ish. (nothing personal) I have to ask this; is it all worth the trouble to sit down for (how long does it take) to post process what the camera can or can't do? What happened to simple compositional image taking when the image was recorded with the confidence and skill in your ability and in the equipment?

It just doesn't seem 'fun'. M8 or D3 or D700 - they are all nice cameras. The problem I see for all the companies is building longevity into the technology and the camera so that price can be justified - I have a difficult time believing that the D700 is as good a camera as the D3 even with it's full frame and all. If Leica wants to maintain it's cut of the market, they have to build cameras that will not be obsolete even if they do cost more. What makes them special is the images still have a different 'look'.

The Leica image is what makes them special. Give me an R or M body that provides me with a digital image that will need no, or little post processing, and make it last, and easy to use. That's all I want. I want to take pictures, not play at a computer.

[sdai]

Originally Posted by ho_co

Interesting. The D700 brochure says the opposite. (See my link above.)

I'm not sure if you're referring to this paragraph on pg.6 of the brochure:

Unlike conventional correction methods that merely eliminate the colors of chromatic aberration, Nikon’s method compensates for the differences in the resolving index for each color, making it particularly effective for reducing the image distortion at the edges of a frame in addition to improving image quality throughout the entire frame. Moreover, because aberration is corrected regardless of lens type, this feature delivers sharper images whichever NIKKOR you use.

Only thing new is that besides "merely" eliminating the colors of CA, they add the compensation of resolving index for each color, but they've only said that it will work for Nikkor lenses.

I believe there's no way you could turn off the feature and since they've mapped almost all Nikon lenses, so they can boast regardless of lens type ... this is not difficult at all, Nikon did exactly the same by storing hundreds of thousands of exposure settings for the 3D matrix metering module.

The theory is about the same, the camera needs to be taught/programmed by the engineers before they could start to work.

The camera has no intelligence to compute the exposure settings for you, all it does is picking one set of preset value from its database which matches the meter reading.

It doesn't know what is CA, it tries to offset the troubled colors with an embeded profile matching the appropriate lens type.

[sdai]

Originally Posted by ho_co

So you're saying that I would have to send all my 30-year-old Nikkors back to Nikon to have some kind of lens recognition added?

Howard, you don't need to sell your Nikkors. Remember there is a lens choice menu you can let the camera remember lens #1 is what, #2 is what else?

You only need to "teach" the camera once then it'll take care of the rest for you.

[sdai]

Originally Posted by ho_co

I know Nikon introduced forensic software for the D2 and D200, so I guess that the artifacts you mention still come through that firmware/software combination?

I haven't read its manual yet but I believe the "forensic" set is to prevent alteration to the "original" picture so it can be used as proof for forensic purposes.

[sdai]

Originally Posted by mooky

I want to take pictures, not play at a computer.

I share exactly the same thought as you do. Which is why I don't even want to bother with RAW files anymore.