Purple fringe and blooming again

Started Aug 27, 2004 | Discussions thread
MatW Senior Member • Posts: 1,470
Neither blooming nor bichrominance.

Let's stop theorization and start looking at some practical examples:

10D, 85mm/1.8 at F1.8

As we can see, the color of the fringes depends on the focal point. This is true for every image, no exclusions( ). Now, wasn't purple the color everybody associated with blooming? Why should it change the color in front or behind the focal point? This really shouldn't matter for the photo cells, should it? Another thing to think about: Why should it be purple anyway? Blooming per definition doesn't involve any color at all, it's pure white spreading over a certain area (just as in the D70 shot in the first post, this is real* blooming, a perfect example!). While there are theories on why it could be purple, none of them can explain the green/yellow.

10D, 85mm/1.8 at F1.8
Again, colors depending on the position relative to the focal point.

10D, 85mm/1.8 at F1.8

Note: This picture is not overexposed (which would be required for blooming), still the color artifacts are clearly visible. Plus, if one looks at the three color channels separately (in Photoshop or so), different focal plains are visible for each color (don't believe it? Try it out, it's amazing). This all very much suggests longitudinal CA.

And neither is it restricted to digital:

FILM, 100mm/2.0
=> The micro lenses and Canon's bichrominance can't be the explanation.

I think what we see here is in fact longitudinal CA by the lens . This explains why different lenses have different amounts of it, why it is most visible at large apertures (as a matter of fact, every fast prime lens shows LCA) and diminishes with stopping down. It explains the aberration's different color depending on whether it is in front (purple) or behind (green/yellow) the focal point (converging vs. diverging light rays). It explains why it is visible on film.

One thing that needs to be considered: If everything is in focus, only purple aberrations will be visible since they actually go a bit behind the focal plane (see the picture of the sheet with the lines). This is why we usually don't see much yellow/green with digicams - on most pictures, everything is in focus.

( ) The pictures shown can easily be reproduced with any fast lens at a large aperture. If you don't believe it, try it out yourself.

This topic has recently been discussed in the following thread, much that has been posted here has been found out by Michael Schaefer (though he still seems critical about the LCA theory). Might be worth a look:

Petteri Sulonen wrote:
Ron Parr wrote:

If blooming were the cause of purple fringing, changing lenses
while keeping everyting else constant wouldn't make a difference.
The phenomenon appears to be caused by the lens focusing some
frequencies of light slightly differently than others, i.e.,
chromatic aberration. Not surprisingly, it is worst at a lens's
widest aperture.

I've been puzzled about this phenomenon for a while. I don't think
the CA theory adequately explains everything about it either --
L-CA is a red-cyan "shift" or "shadow images," not purple, and A-CA
is purplish (magenta, actually), but it's symmetrical around the
highlights and drops off gradually, instead of being a sharply
delineated fringe like PF. I've done some research into it, and it
appears that the technical term for this is different --
"bichrominance" rather than CA.

The explanation that appears to me to fit the facts best has to do
with microlenses and off-axis light. A lens that's not telecentric
or near-telecentric will have an exit pupil that's fairly close to
the sensor. This means that a large part of the light is off-axis,
especially towards the corners. Open up the lens, and even more
will be off-axis. What if the off-axis light that hits the
microlenses gets scattered into neighboring photosites? This would
explain a number of phenomena:

(1) The sharp delineation. This kind of scattering would "fade"
quickly, and probably only be grabbed by neighboring photosites.

(2) The radial quality. Light would be scattered away from the
center, creating a shadow image that's larger than the "real" one.

(3) The violet/purple color. Since there are more green than red or
blue pixels, more light would get scattered away from green than
away from red or blue. Red+blue = violet.

This is all conjecture; I'm sure someone knows, but they ain't

[ http://www.prime-junta.tk ]

-- hide signature --

Keep it simple. -Alfred Eisenstaedt

 MatW's gear list:MatW's gear list
Fujifilm X-E1 Fujifilm XF 35mm F1.4 R Fujifilm XF 60mm F2.4 R Macro Fujifilm XF 14mm F2.8 R Fujifilm XF 18-55mm F2.8-4 R LM OIS
Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow