Digital artifacts

<p>Hi all,</p>

<p>I posted this to another forum, and afterwards thought that some people who read this forum

might appreciate it.</p>

<p>In my recent research one of the things I have found most distressing is the amount of artifacts

that even the high end digital cameras can produce. I recently had to deal with postprocessing of a

number of digital images from a Canon EOS-1Ds Mk II, and I was very disturbed by the amount of

digital flare visible in the image. So I looked for other examples of this. You can see very obvious

examples on the web, and below I've posted some examples from the sample image galleries at

dpreview. For all of these, make sure that you download the original image (large file). The examples

I've posted examples are from some of the more expensive cameras, but I imagine that these apply to

less expensive cameras as well.</p>

<p>Canon EOS-1Ds Mk II: <a href="http://www.dpreview.com/gallery/">http://www.dpreview.com/

gallery/</a> canoneos1dsmkii_samples/ Look at the first image of a building looking up in wide angle

("Little London Mill House"): Full res image at: <a href="http://img2.dpreview.com/ gallery/

canoneos1dsmkii_samples/originals/vi7h3998.jpg">http://img2.dpreview.com/ gallery/

canoneos1dsmkii_samples/originals/vi7h3998.jpg</a> <br />

On the full res image you will see digital artifacting in the form of red glare at the outward edge of the

windows and the underside of the overhang. In particular look at the left side of the small white sign on

the left side of one of the windows.</p>

<p>Canon ESO 5D: <a href="http://www.dpreview.com/gallery/canoneos5d_samples/">http://

www.dpreview.com/gallery/canoneos5d_samples/</a> Look at the sixth image of a bridge in wide

angle. Full res image: <a href="http:// img2.dpreview.com/gallery/canoneos5d_samples/originals/

img_6783.jpg">http:// img2.dpreview.com/gallery/canoneos5d_samples/originals/img_6783.jpg</a>

<br />

You will see very clear red ghosting of the vertical stays for the bridge on the right hand side of the

image.</p>

<p>There are more examples that you can find yourself. I'm not trying to promote film over digital -

digital is incredibly useful, fast and has many benefits over film. The point is that some of these

problems may have been caused by rather lame glass, but others may be from the digital process itself.

Some of these problems can be avoided by not shooting into glare, but the first example is clearly not

due to glare. The Markii ds image that I processed was for a wall-sized enlargement, which is

something that few people ever do. I found that equivalent 35mm slides were better for this purpose,

because there were no such artifacts to deal with. The client insisted on using the digital image, and

removing these artifacts was very time-consuming. This is an extreme example, but it is clear to me

from the examples above that even the best digital cameras can produce artifacts that would be

noticable on even a mid-sized enlargement.</p>

<p>All of that said, I still enjoy digital very much, and with the right glass it can produce stunning

results. In particular, the lack of grain in digital is wonderful. </p>

<I>but others may be from the digital process itself.</I><P>

Yes! I've also made this point here several times.<P>

The whole idea of trying to reconstruct a scene by analyzing separate, and spatially-displaced samples of Red, Green, and Blue will inevitably produce strange artifacts under the right circumstances. I've cited a number of examples in this forum that I've encountered.<P>

Some of them involve rather arcane circumstances. e.g., shooting resolution test targets comprised of closely-spaced lines of specific colors. <P>

But I've encountered others in real-world shooting situations. I had a whole set of shots with a model completely ruined last year because I was framing closup head shots of a model with a blue hairlight shot against a bright orange background. At every interface between a blue-highlighted strand of hair and the background there was a border around it. These were in the Raw files and EVERY known Raw file converter for my Canon 20D. <B>Nor</B> is this surprising - trying to "guess" the correct color of features 1-pixel wide against a background of a complementary color is mathematically impossible with a bayer array - there is just not enough information to work with.

Will the solution be a Foveon type sensor or will some one bring out a 3 chip and colour seperation prism camera (like in high end video)

Both presumably have huge potential to out strip what we are currently doing, even with the same lenses...

The prism should in theory allow us to catch more photons. In a bayer system at best we loose at least

75% blue photons

75% red photons

50% green photons

before they ever reach the sensor and that assumes perfect filters

The red "ghosting" does not appear in a similar picture of the same bridge taken with a Nikon D2x, using a 17-55/2.8 DX.

(http://www.dpreview.com/gallery/nikond2x_samples/originals/DSC-1870-raw-capture.jpg

The red ghosting is reversed on the left side of the image, appearing on the left side of the stays. This has the appearance of chromatic aberation rather than some digital artificat. This CA may originate in the sensor, rather than the excellent "L" lens used for the capture.

I have taken many similar pictures with my D2x, none of which exhibit this phenomena. I have observed chromatic aberation in some lenses, particularly the Micro-Nikkor 105/2.8 AFD, but not my f/2.8 zoom lenses, and never to this extent.

Needless to say, I'm not holding my breath over a Foveon resurgence. I suggest you look elswhere for assigning blame or achieving a solution.

"I have taken many similar pictures with my D2x, none of which exhibit this phenomena. I

have observed chromatic aberation in some lenses, particularly the Micro-Nikkor 105/2.8

AFD, but not my f/2.8 zoom lenses, and never to this extent."

This is an interesting issue, and I'm glad that some others have contributed. I agree that

this looks related to chromatic aberation, but I think that it might be a combination of that

with sensor technology. I've never seen such a pronounced effect on film images. In

particular, it is difficult to ascribe the effects seen in the first example to chromatic

aberation alone.

Glad to hear that you've never seen it on your D2x. I wonder if it has something to do with

the larger cmos sensors.