red edge more general?

[ho_co]

Advertisement (gone after registration)

Moderators, please delete if necessary since the images are not from Leica. Or move to M9 forum if you agree it may be of interest in regard to the M9 red edge problem.

 

 

FWIW, I hope this may be relevant to M9 red edge discussion.

 

Shot with Nikon D200 (APS-C) camera, full auto. First image is direct from RAW with JPG conversion for forum:

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

 

But after applying "Curves > Auto" in PS CS 4, thus:

 

and converting to JPG for upload, the image looks like this:

 

 

(Much of the blotchiness and the hard lines along the plane and moon edges come from the JPG conversion.)

 

In other words, the camera seems to have added a slight magenta cast along the top, and a slight cyan cast along the bottom. These are invisible until the colors are aggressively tweaked.

[jankap]

Can sharpening be the cause? Think of paintings, where edges are lifted with contrasting colors too.

Very interesting. Moderators, please do not delete this thread.

Jan

[jaapv]

The red edge is indeed not unique to the M9.

[lct]

The red edge is indeed not unique to the M9.

Which other cameras do you refer to it if i may ask?

[jaapv]

Kodak DCS 14, M8, incidental reports of other brands like Howard's here. I am begnning to suspect that we are seeing a Bayer interpolation error at the edges of the frame, which gets more visible on an AA-filterless sensor (sharper transition between the pixels, a high-frequence problem??) and is particularly visible on the M9 because of the vignetting correction.

[lct]

Never got it personally but aside from keeping an AA filter it would be interesting to know what other makers have done to fix it if any.

[mjh]

Advertisement (gone after registration)

In other words, the camera seems to have added a slight magenta cast along the top, and a slight cyan cast along the bottom.

Surprisingly the magenta stripes on the left and right don’t coincide with the edges of the image; rather there are small margins without the magenta cast at the left and right borders of the image.

[lct]

Surprisingly the magenta stripes on the left and right don’t coincide with the edges of the image...

Are they really magenta? Reducing magenta (and red) saturation doesn't seem to change anything.

[jaapv]

If I look at it again the cast on the left and the right looks more like banding. And if I look once more, the banding is visible over most of the image.

[ho_co]

Surprisingly the magenta stripes on the left and right don’t coincide with the edges of the image; rather there are small margins without the magenta cast at the left and right borders of the image.

 

Michael, that surprised me as well. It almost looks as if there were some reflection off the edges of the "film"-gate. The right-hand side seems to be more a brightening than a color cast.

 

I don't have any idea where it comes from. In fact, until one applies absurd curves to the image, it's invisible. The D200 doesn't exhibit a color edge problem in the real world--but there it is, under the surface. The image was made with a 70-200 at the maximum focal length, completely the opposite situation of the M9's anomalous edges.

 

The cause of this is likely completely different from the M9's edge colors, but it's interesting that it manifests itself similarly.

 

 

Jan, the effect may indeed have to do with sharpening, but the image is derived directly from an ACR conversion using "Auto" in the Basics panel, with no extra sharpening. BTW, I didn't know of the painterly technique you mention--quite interesting.

 

Jaap, you mention banding; could be. But the JPG conversion has produced some clumping that is less visible in the NEF file, so you may be seeing a secondary effect.

 

Effects of JPG conversion are visible in another aspect: In the tweaked-color image, there's a bird rendered as a short black line at the level of the moon a little over half-way from the moon to the left edge. The JPG conversion of the natural-color image conversion lost the bird.

 

lct, I just used convenient names when I said 'magenta' and 'cyan.' I'll be glad to use a better name if you've got it.

[mjh]

I don't have any idea where it comes from. In fact, until one applies absurd curves to the image, it's invisible. The D200 doesn't exhibit a color edge problem in the real world--but there it is, under the surface. The image was made with a 70-200 at the maximum focal length, completely the opposite situation of the M9's anomalous edges.

 

The cause of this is likely completely different from the M9's edge colors, but it's interesting that it manifests itself similarly.

While I can only speculate, my bet would be on non-uniformities of the sensor or the read-out process. If so, the effect should be independent of lens or focal length used, but appear more pronounced at higher ISO settings.

[AlanG]

It seems to me that all you are getting is some kind of anomaly from the Auto setting in Photoshop. (Which is not a good choice to use on this image.) I manually overlaid a contrasty curve and got this:

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[mjh]

You have to apply individual curves to the red, green, and blue channel, as Howard did. It doesn’t depend on Photoshop; I’ve got similar results with Pixelmator under Mac OS X. The effect is quite real.

[AlanG]

You have to apply individual curves to the red, green, and blue channel, as Howard did. It doesn’t depend on Photoshop; I’ve got similar results with Pixelmator under Mac OS X. The effect is quite real.

 

I know. He also reduced a low contrast 8 bit image to a posterized image of just a few tones. Why would you do that and what does it tell you about actual photography?

[lct]

Hard to make a Nikon behave like a Leica apparently.

[ho_co]

Alan, thanks. I guess I didn't make myself clear. I wasn't asking how to get an acceptable image from the file, but attempting to point out a similarity in the appearance of cyan and magenta tint between the two cameras, and wondering if that might point to a cause.

 

But I do thank you for your processing guidance!

 

 

lct, as for getting Leica color from Nikon files, that has already been covered online at KammaGamma .

[lct]

Interesting Howard thanks. Do you know if it works with Lightroom 2?

[AlanG]

Alan, thanks. I guess I didn't make myself clear. I wasn't asking how to get an acceptable image from the file, but attempting to point out a similarity in the appearance of cyan and magenta tint between the two cameras, and wondering if that might point to a cause.

 

 

Why not shoot a grey card then?

[ho_co]

... my bet would be on non-uniformities of the sensor or the read-out process. If so, the effect should be independent of lens or focal length used, but appear more pronounced at higher ISO settings.

 

I think you're right about the sensor read-out--delivering minor non-uniformities that are of no real photographic importance.

 

I may check further in regard to ISO-dependancy. This shot was made at the camera's base ISO 100.

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[ho_co]

lct--I'm afraid I don't understand which question you're asking:

1) If you're asking whether the KammaGamma suggestions can be applied with Lightroom, that's their ball of wax. I have no idea.

2) I don't know whether Lightroom allows such extreme control of the color channels. It probably does, but at least it won't do it automatically.

 

 

I think we're getting a little off topic. I was surprised when an occasionally helpful Photoshop auto-tool produced such a ghastly result, and doubly surprised that the image anomalies seemed similar to some produced by some M9s with some lenses.

 

Any image can be destroyed by stupidity, and this auto-tool works stupidly.

 

But my point was just that something in the digital processes of two very dissimilar cameras produced a similarly 'wrong' output despite completely different conditions.

 

I was hopeful that that fact might lead someone to some insight in understanding the M9's 'red-edge syndrome.'