Understanding Chromatic Abberation

[Eric Gulbransen]

This is my first post on dvinfo.net, but I am no stranger to these pages. I have been reading and learning here for months. I am already in debt to most of you - from the pros, to the people who simply posted their first footage. I have learned from you all. So first of all, thanks...

My question is about chromatic abberation (I think). Not so much what it is, but more, how can we do a better job (or simply CAN we do a better job) of avoiding it, or minimizing it while shooting.

I have a lot of respect for Steven Dempsey. In fact it was his Yellowstone piece that originally attracted me to this site. Obviously the man knows how to shoot with an XL H1. So when I noticed a good bit of what I "think" is chromatic abberation in the dock_lillies.m2t clip that he offers in the "XL H1 Sample Clips" section, I thought this might be another opportunity to learn.

Here is a snapshot example of the issue I'm referring to. I took the snapshot as the original clip paused in the VLC media player, however it was also visible as the clip played:
http://gotagteam.com/images/XLH1Clipissue.jpg

Here is the oriiginal clip:
http://www.realm.cc/upload/disjecta/dock_lillies.m2t (70mb)

Please understand that I am not being critical of the footage for any other reason than to be constructive, and to create an opportunity to learn, more.

My question is about shooting. What things, if anything, could the shooter have done differently with the XL H1 to avoid, or at least to reduce, the affects of this particular challenge?

Thank you again for this great site. And Happy New Year!

[Giroud Francois]

since chromatic aberration is a characteristic of the equipement (lens), there is not much you can do except purchasing good lens with low CA, or use the lens in a way the CA does not show to much (no full zoom, no full wide, no full iris opened).

[Bill Ravens]

chromatic aberration is the result of the physics of light refraction in glass. much in the way a prism bends white light into a rainbow of colors, a camera lens will do the same. This is the result of the fact that different colors(wavelengths) of light bend(refract) passing into or out of the glass surface. One reason camera lenses are composed of so many individual glass elements is the designers desire to include correction glass elements in his attempt to mitigate CA. Totally eliminating CA is pretty difficult, even with unlimited glass elements and high quality coatings on each element. Typically, CA is well controlled in expensive lenses.

Zoom lenses are an especially difficult optical design, because what can be corrected out at one focal length, can't be corrected at another. That's why CA is most apparent at extremes of a zoom lens, especially extreme zoom. Ever wonder why some motion camera lenses cost $50K? It's because of the complexity of the lens as it contains many elements and coatings to remove, among other things CA.

There's nothing, no filter or trick, one can use to eliminate CA. One can limit the lens to the middle zoom ranges, avoiding the extremes. Likewise, there is a "sweet" spot on a lens where the CA is at a minimum. The only way to find this sweet spot is to test the lens, in use. The sweet spot is at a moderate f/stop, like f/4 or f/4.6. Too far closed down and you get diffraction, too wide and CA and other resolution robbing effects dominate.

I hope this explanation has been of some help.

[Eric Gulbransen]

Wow. That's very interesting. A little disheartening, but vital to understand. Now I get why it's so helpful to use a monitor to view, well, what you're viewing..

Thanks very much for your time.

[A. J. deLange]

Amen to the comments about finding the "sweet spot" or spots, i.e. the conditions under which your lens is least subject to chromatic aberration. A couple of additional considerations: the main manifestation of lateral chromatic aberration is red, green and blue light images of different size. Thus chromatic aberration is more plainly seen at the edges of an image than at the center. One can therefore reduce the appearance of CA by trying to compose such that things with high contrast (rooflines or branches against sky, latticework etc) are away from the edges of pictures. Also there are "filters" which combat CA. They do this by separating the image into its red, green and blue parts and radially scaling the blue and red images so they align with (are the same size as) the green one and then reconstructing luma and chroma. These have been available in Photoshop and other still image processing programs for some time and they could be used on video but it was an elaborate process in which the video would be converted to a string of jpegs, the jpegs batch processed for CA and then reconverted to video.

More recently there is a package of plugin filters by Digital Film Tools (www.digitalfilmtools.com) which corrects lateral CA (fringing). The reduction and contrast and sharpness caused by longitudinal CA cannot be fixed. But be aware that the XL-H1 appears to have CA well under control. No one has done a complete set of tests but random checks at various lens settings with the 20X and the 6X both seem to indicate CA of one pixel or so at picture edge. Given a color bandwidth of four times this most of the fringing artifacts that people see are from reduced color bandwidth rather than CA. I strongly suspect that the example you posted represents such a case. To confirm this you could take the grabbed frame and try to process it in PS or some other still image program or go to DFT's website, download their trial package and see if you can remove the fringing with their lens distortion plugin.

[Steve Rosen]

If you look for it, you'll find it... if your pictures are compelling, no one else will ever see it...

I shoot at 1.6 w/ the 16x all the time in HDV. Theoretically, this should be the worst scenario with that lens for CA. I also use the 20x to the limit... (I NEVER stop down more than f4).

My filmaking partner, who is a talented director but has no technical photographic experience, has never once asked "what's that color fringing in that shot?" This is after viewing a hundred or more hours of tape from the H1...

as I said in a previous post - I forgot about it and it went away...

[Eric Gulbransen]

Thanks A.J. I followed your lead and came up with this as a result:
http://gotagteam.com/images/PshopCA.jpg

I also learned a little more about this, which was my hope, so thanks.

Steven, I appreciate the work you guys put into this site so I did search first about CA, and I do remember reading you "forgot about it and it went away." It's just that when I viewed the clip in question I did suddenly remember it again. Like anything else, understanding a problem is the first step in fixing it. Thanks to you guys I do now understand CA a whole lot more.

[Greg Boston]

Originally Posted by Bill Ravens

Zoom lenses are an especially difficult optical design, because what can be corrected out at one focal length, can't be corrected at another.

Well put, Bill. That's why I chose to do my white shading corrections on my camera/lens with the lens set to mid-range. I have been pleased with the results.

-gb-