Red fringe in low contrast snow photograph

[Boatman]

Last night we had a nice snowfall and this morning I stopped for a few photographs on my way to work. I shot in RAW and took a look at them this evening. On many I noticed a strange redish fringing on the sides of the photos. It reminds me of a light leak but I know that's not the issue.

The photos were taken in very flat light and I'm not 100% sure I'm not imagining it. I think the contrast may be playing tricks on my eyes. Look at the attached images. Do you see it? I think it is worst at the lower left hand corner. What would cause this? How can it be removed (other than cropping)?

[jiro]

I can see it, too. what lens did you use? I was thinking of some light reflection from the lens.

[Ken MT]

Could it be light reflection from an object just out of the frame (car, red jacket etc)?
What about converting to B&W, gets rid of the red colour that I think is on both sides

[MiniChris]

If you go into your raw editor to the lens correction section, you'll see a slider for removing red or yellow fringe. I didn't understand what that was all about until I shot some white on white stuff for a student demo and that occured...in my case, the slider adjusted for the red and disappeared it...good luck.

[Dave Humphries]

Hi Homer,

I wasn't familiar with that camera so I looked it up, it was quite well specified for its time, if I'd been into photography back then, I could imagine being tempted by the spec and some very useful features. Even now, apart from the small LCD size, and relatively low 8MP/800iso 'max' sensor, it isn't bad, is it?

I see some CA (chromatic abberation) on edges of tree trunks (e.g. trunk, red edge, snow; on left), but I've never known it produce this kind of shading effect on the snow, which I see as a 5% (of image width) column up both vertical edges, quite clearly.

I wonder if Chris's idea will work, I can't see from EXIF what software you use, but you'll probably need something photoshop based to do what Chris suggests.

Other random thoughts:
I do slightly fear, given the age of the sensor, whether something untoward is going on inside the camera, not that you can look and see.
I wonder if temperature played a part.
The iso was low (64), so that shouldn't be a factor.
Ken's flare (of something out of shot) idea could be valid.
As the lens is fixed, you cannot swap to isolate that aspect.
It is definitely on both sides, I could convince myself I see it, to a lesser degree, across the top, but it doesn't seem to be present across the bottom edge.

Most odd,

[Black Pearl]

That's a new one on me too.

I'd also go with Dave's suggestion and say it looks like there's a problem with the sensor or imaging pipe-line.

I'd be interested to see some images taken of a plain white piece of paper or wall to see it it shows again.

[Boatman]

Dave: I'll respond to your comments first and then investigate further tonight when I can work with the images. The A2 was quite advanced for its time, but its time was 2004. So, it's more than a bit behind the curve technologically now. I didn't look closely at the tree trunks for chromatic abbreration, but I'll check that. This is not a common problem with the camera. The red fringing is something new and not present in all the images I took. The second image does not show very much of the problem in the 72dpi conversion, but the RAW image displayed in Picasa has the worst fringing of any of the photos. Go figure.

On the A2, it is a super little camera and I would recommend one as a back up or as a camera for just having on hand as it is light and small and cheap to buy used. However it has a couple of major limitations: ISO 200 or higher is very grainy, focus time is slow (though it has a good manual focus system) and, as you mentioned, 8MP is not nearly state of the art. It will produce a good 11 x 14, though, if you don't crop too much.

[Boatman]

Well, I certainly got some good suggestions from this list. Thank you.

The photos were taken without any nearby bright objects to reflect into the lens, so I don't think that is the issue. I went back and looked at a lot of earlier photographs and found a few that presented a similar problem. They are pretty rare but all seem to be in situations where the light is very flat. In the case of these snow images, I had added 0.7EV to the exposure to bump the histogram right up against the right end, but not going over. Due to the flat light this meant that the curve was essentially empty from the mid-point to the left. I think this may have something to do with the fringing or at least be a warning that it could happen. Perhaps the 0.7EV bump was not a good idea.

Taking Dave's advice, I re-opened the shot in Photoshop RAW converter. This time I viewed it at 100% and checked for the cromatic abberation that Dave saw. Good eyes, Dave. I can't believe you saw that on the 72DPI reduced image. But it was there. A distinct red chroma at the edges of the photograph. This did not appear in the middle of the image. This shot was taken at the wide end of the 8 - 50mm zoom lens so some abberation at the edges is more or less normal, I guess.

Using the lense correction tools in RAW, I made the following setting changes:
Fix Red/Cyan Fringe: 0 changed to -30
Lens Vignetting changed from: 0 to +46 (the maximum without clipping the corners)
Midpoint: 50 to 0

The changes essentially remove the problem although there is still a bit of red tinting in the brambles in the lower left. However, these brambles have brown in them and we may be seeing the actual coloration. In any event it is no longer pronounced and probably would not be noticed by anyone not looking for it.

Thanks, forum members. I'm now just a little more skilled in my post-shot skills.

Here is the corrected image:



And this is a screen shot of Picasa so that you can see the before histogram.

[Black Pearl]

Just a quick one Boatman - the dpi is only relevant when you come to print an image, it has no effect on what you view on a monitor.

I'd still be interested in seeing some test shots of plain paper to see if the colour cast is at various focal lengths.

[proseak]

What software do you have access to? If it can handled LAB colour, convert to LAB, and apply Gaussian blur to the A & B channels to suit.

HTH