help on chromatic abberation

Does anybody know how to fix the chromatic abberation in this photo?

Image by lukeROB (Copyright) posted at ScubaGallery

FWIW, Photoshop CS has it's own corrector for it's RAW converter. I know there are others out there. A quick web-search should show up plenty of info.

lukeROB:

Does anybody know how to fix the chromatic abberation in this photo?

Image by lukeROB (Copyright) posted at ScubaGallery

Click to expand...

I don't see any chromatic abberation. I thought chromatic abberation was a problem with how light was focused. I'm not sure of the technical explanation, but I think instead of say, white, you would see red, blue, green, etc seperated if you looked real close. This could be caused by using a flat port with a wide angle lens and the light is not "focused" properly at the edges. If I'm wrong, please tell me, as I would feel silly using this theory and being way off. By the way, I like the picture and see nothing wrong with it. I especially like the wide angle photos in your gallery. What system and lens did you use?

s****enstein:

What system and lens did you use?

Click to expand...

Oops! I just noticed you had your camera system info listed at the bottom of your post.

luke

the person above me is correct on the light placement, but i like the shot. there is no CA (purple fringe in this photo) and with CA theres really not anything you can do for it as it is a result of the lense of the camera and refractory of light. i will dig in my digital library for a good example of it to post, or go look on dpreview as there should be some good examples there.

hope this helps

tooth

You are correct the problem is casued by a lens not focusing all the colours to the same point. For digital cameras it often shows up as a 'purple fringing'. This photo has quite a bit especially around the white bars on the clowns (zoom in and you will see it). It also has other colours fringing and makes the image look less sharp because of slight colour shifts - the channels are not lying directly on top of each other (sort of). I still like it but I wanted to get it printed and in the larger image the problem is more noticable.

I have photoshop CS but my camera cannot shoot in raw so I do not have access to this powerful correcting tool - I thought there may be something similar that does not require raw files but would let me perform the shift anyway.

Scubatooth:

luke

the person above me is correct on the light placement, but i like the shot. there is no CA (purple fringe in this photo) and with CA theres really not anything you can do for it as it is a result of the lense of the camera and refractory of light. i will dig in my digital library for a good example of it to post, or go look on dpreview as there should be some good examples there.

hope this helps

tooth

Click to expand...

Yes it is a result of a lens and the refraction of light. Just like fisheye, barrel, pincushion distortion it should be able to be correct by shifting channels relative to one another though maths. It is not going to be perfect but could improve the situation - just like how the PS CS raw converter does it or the Nikon programs correct for their fisheye lenses.

This photo has plenty of it. See how the anemone tips appear slightly blurred (look like a motion blur) it isn't. It is a red/cyan abberation and it makes what should be a much sharper image look like there is some blur.

lukeROB, in that case you may find relief in a couple of tools found in a Photoshop plug-in, Panorama Tools. Not perfect results, but better than the blue fringing in your upper right and lower left corners.

Can't remember where I downloaded it from, but it's freeware.

BTW, is your photo cropped? The centroid of the fringing appears to be a bit left of the right-hand fish's nose.

All the best, James

fdog:

lukeROB, in that case you may find relief in a couple of tools found in a Photoshop plug-in, Panorama Tools. Not perfect results, but better than the blue fringing in your upper right and lower left corners.

Can't remember where I downloaded it from, but it's freeware.

BTW, is your photo cropped? The centroid of the fringing appears to be a bit left of the right-hand fish's nose.

All the best, James

Click to expand...

I agree the centroid appears off centre but this photo is no cropped at all. I have got pano-tools is this the same thing. I have tried a bit with it but it is not an easy to understand plugin - any tips on using it for this sort of correction ???

lukeROB:

PanoTools is the tag attached to Panorama Tools after it’s loaded. Since the tools are so close to the written code, they are all at once obtuse and powerful.

Try this:

Open Panorama Tools through the Filter menu. Filter>>PanoTools>>Correct
Uncheck all boxes except Radial Shift.
Click Options. You should get a window “Set Polynomial Coefficients” or some such.

Note: This window is one of the most powerful in panotools. It applies a distortion in amounts as defined by the complex equation seen in the window. Normally, if you are correcting barrel or pincushion, varying numbers are applied across from left to right. Because you want the colors to match, the numbers in each vertical column are kept identical.

The far-right column applies a simple radial pattern with little influence in the center, and most at the edges.

Simply put, this column expands (or shrinks) each channel. 1=same size; greater than 1=shrink; smaller than 1=expands. (Seems backwards, I know)

So, since the light areas exhibit cyan “bleed” to the outside, I would first try to expand the red channel. NOTE: a little bit goes a long way! I’d start with a red expansion of 0.998. No kidding.

So, getting back to the cookbook, at the “Set Polynomial Coefficients for Radial Correction” window:
Make sure all boxes are filled with 0’s (zeros).
In the far right column, set: Red=0.998, Green=1, Blue=1
Make sure “Radial” is selected.
Click OK to apply.

You can back up if it’s too much. Now that I look closer, there is also red fringing on the left of Nemo’s stripe. This kind of aberration always comes in pairs.

I mentioned the centroid because this correction is applied symmetrically; you may end up with unacceptable results in the upper left corner.

All the best, James