Reading mode: Light Dark

Purple Fringe Explained : IRCA -- Infra Red Chromatic Aberration

Go to last post
I bought a Tiffen hot filter for my 707 thinking it may help in some situations. Test of the internal IR filter showed that it was much less than perfect. I think it does help a little sometimes but I haven't ever really tested in a controlled fashion. Lens flair is worse so there is a trade off.

The Tiffen gives a much better IR cutoff ... If anyone has an 828 in the Bay area and wants to test it with the 58mm Tiffen I'd be willing to help.

Good point about IR. If the 828 uses the same IR filter, significant IR is reaching the sensor.

Geoff
AlanF said:
http://forums.dpreview.com/forums/read.asp?forum=1009&message=7334378
828, PF, and miracle cures: a word of advice... Petteri Sulonen

but it's of general interest, so I'll continue here ]

Purple Fringe usually occurs when there is a sharp boundary between
the sky and an object silhouetted against it.

For example, on my Sharp VE-CG30 [don't laugh!!! While saving up
for .. a ? Minolta A1 ? I've taken over 14,000 pictures with it --
the best $188 I ever spent] :



http://www.well.com/~af/shot_today/PICT1493A_LEAVES_5X.JPG
[I've intentionally put the URs in twice ... so I can
cut-and-paste from here]

I've researched the subject through google, and dismissed the
following:

a) Blooming (pixel-to-adjacent-pixel charge transfer)
b) Smear in a CCD : pixel-to-pixel charge transfer during the readout
c) De-mosaicing [interpolation] algorithms used to convert the RGB
Bayer pattern into an RGB value for each pixel

because none of those explain why the result is PURPLE.

d) General chromatic aberration -- mainly because the purple fringe
is a digital camera effect.

Then I came across:

http://www.botzilla.com/photo/G1chroma.html

Digital cameras are very sensitive to Infra-Red. Film isn't.
AlanF said:
While optical lenses perform well in the visible spectrum, some distortions can appear outside that color range. The UV/IR light can focus slightly off-center from the place where the visible light focuses -- when this happens, the distorted IR caused a "lateral aberration" just past the highlight.
Click to expand...
AlanF said:
Lateral aberration is related to the wavelength of the light, and so when aberration occurs, longer wavelengths (reds) will tend to be away from the center of the optical axis, while shorter wavelengths (blues) will tend to be toward the center. This points even more convincingly to IR as the hidden source of purple fringe, because it appears only of the "outside" of highlights.
Click to expand...
AlanF said:
If the aberration is infrared, why magenta? How does the blue get in there, rather than the aberration being strictly red?
Click to expand...
AlanF said:
The IR can show up in both red & blue channels, because those colors are interpolated across pixels -- different pixels on the CCD are masked with different colors, and those masks are apparently transparent to IR. DPreview has published a small guide to CCD color coding here. Canon uses the CYGM system.
Click to expand...
AlanF said:
The dyes used for the color mask are tuned to visible light, but not (as far as I can determine) to infrared. The IR passes through not only the red-sensitive portions of the mask, but also leaks through the other colors as well -- least of all for the green channel. What's white less a little bit of green? Magenta.
Click to expand...
I did some initial tests with the IR remote from my Pentax Zoom90WR,
which I posted in the Sony Forum.

Now I have a New and improved IR test. With a $4.99 multi-frequency
optical generator .. Red, Green and IR Led's .. aka a TV remote,
and an
occluding disk, aka an anglepoise desk lamp, stuffed with [don't ask]
so no light would leak through the center.

camera: 16 secs F2.0 ....... Occluding disk ..... MFOG

Test procedure: set camera on timed delay. Focus with light on. Start
timer. When blinking light goes off, wave MFOG around behind the
occluding disk. Count to 16.



; (240K)
and 5X clips from the Top and bottom:

http://www.well.com/~af/shot_today/PICT4313_TOP_5X.JPG



http://www.well.com/~af/shot_today/PICT4313_BOT_5X.JPG



Clearly it IS an IRCA (Infra Red Chromatic Aberration) effect : the
IR circle of confusion is very large ... that is, when visible
light is focussed, the IR isn't.

So the green and red STOP at the occluding disk, but the IR keeps on
going, even at the center of the frame (though that's not shown in
THIS shot).

And the purple comes from the relative sensitivities of the Bayer
filters. At high intensity they all saturate, so we see white.

But the green is LESS sensitive to the IR, so as the IR intensity
drops off some green disappears, and we see Red+Blue = purple.
Click to expand...
--
http://www.pbase.com/geoffb
 
AlanF said:
AlanF said:
One small problem I see, and be the same problem faced by the
manufacturers in implementing their blocking filters, is that in
your sample pics, it appears that the IR filter is affecting the
visible color spectrum a bit, as well...
Click to expand...
I bought the cheapest IR filter I could find. If you look at

http://www.maxmax.com/aXNiteFilters.htm

you'll see that the CC1 filter has a peak in the Green, and
starts falling off in the Red.

For 'real' photography you should find a good 'bandpass' filter
more like the BP1.
Click to expand...
Hrmm.. the BP1 looks like it would cause problems in the 600-700nm range. In fact, the CC2 looks like (according to the charts and also their notes about it's application) it would be better suited to IR blocking without affecting visible.

TheBarron
 
Alan,

First off, thanks for all your diligent work concerning this phenomena. But I think that what you may have been measuring is the spectral sensitivity differences of the color filter array in the IR region.

The materials and procedures used to make the color filter array mask on the photo sensor will vary from model to model and even greater variations could be expected between manufacturers. The closest I can find to the spectral sensitivity of the 828 is the one for the 717 which I know everyone would state cannot be the same as it is only a three color and not the four color filter present on the 828. As shown below, this chart does not show us the response in the range of interest, i.e. wavelengths longer than 700nm.





Without the information we need to see what the spectral response of the 828 is in the IR region we are reduced to speculation. But just to show that IR could be detected as any color, here is a chart from a Kodak sensor. This at least shows that, without the addition of an IR cut filter in front of the sensor, it is indeed possible that the sensor would respond strongly to IR at about 800nm.



Depending on the individual color's filter response in the IR range, any color, or any combination of colors, could be seen as the response to an IR source. Example: if the green filter had more attenuation at 800nm than the blue or red the result would be displayed as purple.

This may or may not explain what you have seen in your experiments with your particular camera, or what's happening on the 828, but it may be something to consider.

Take care, good luck.
 
MaxEntropy said:
Alan,
First off, thanks for all your diligent work concerning this
phenomena.
Click to expand...
Thank you.
MaxEntropy said:
But I think that what you may have been measuring is
the spectral sensitivity differences of the color filter array in
the IR region.
Click to expand...
Actually, that's exactly what I HAVE been saying.

The IRCA of the LENS plus the IR sensitivity of the sensors produces A FRINGE.

The COLOR of the fringe depends on the SENSITY DIFFERENCES
of the filter array.
 
MaxEntropy wrote:
.
MaxEntropy said:
But just to show that IR could be detected as any color, here is a
chart from a Kodak sensor. This at least shows that, without the
addition of an IR cut filter in front of the sensor, it is indeed
possible that the sensor would respond strongly to IR at about
800nm.

Click to expand...
Great curves! (And I'm not talking about anatomy) !!!

Here's my guess on the Kodak filter.

At first glance it looks like the IR result would be reddish at 750nm,
and then white above 850nm.

But the peak of the blue filter is only about 1/2 that of the red and
green filters. So the weighting of the De-mosaicing algorithm
would be 2*Blue + 1*Red + 1*Green.

So the results at 750nm would be

Red = 1 * 32 = 32
Green = 1 * 20 = 20
Blue = 2 * 20 = 40

I haven't plugged those in, but I'd guess it's purply.

At 850nm we would have

Red = 1 * 20 = 20
Green = 1 * 20 = 20
Blue = 2 * 20 = 40

so the fringe would be bluer.

I don't know what frequency my IR source is.
I'd guess that a diffuse, sunny sky would produce a fairly broad spectrum.

Of course, those calculations I just did assume that the sensors have a fairly
flat response. If they are more sensitive to blue then one would have to
factor that in.
 
The following crops are all taken from Phil's PF/CA test photo for the F828. You can download the entire photo here:

http://www.dpreview.com/ ... ... SonyDSCF828/Samples/Fringing/DSC03559.jpg

These crops are of some of the small holes places around the test pattern and at the center as noted. The approximate level of the three color channels in the center of each hole is also noted. Please verify this on your own, if you feel the need.





The lateral CA can be seen on all the perimeter holes as a color shift of red/yellow on the edge towards the center and blue/purple shift towards the outside away from the center. Thus the CA is mis-focusing the longer wavelengths (red/yellow) toward the center and the shorter wavelengths (blue/purple) away from the center. Notice also the PF present at the left-top hole, the PF is away from the center, where shorter wavelengths would be predicted to be.

I believe, in this test, for this camera, this photo shows that IR would not be the cause for the PF shown, unless there is some rapid shift in the lens response to IR that would send IR toward the outside away from the center, but I don't think that is possible.

It would seem more likely that there is a shift of the UV, shorter wavelengths, due to the CA present in the lens away from the center and that there is a response to UV on the sensor, and that due to differences in the respective color filters, it is being sensed as blue/purple. For this to happen, the green filter would have to attenuate the UV more than the blue and red filters in the UV region. In the absence of a spectral response chart in the UV range for the 828, this is just guess work.

Is this the answer? I don’t know.
Is it possible? Maybe.
In any case, it’s fun to think about!

Good luck.
 
We know a little about 7x7/828 sensitivity in IR from nightshot ... if IR gets through to sensor it will be seen.

As for the color ... not 100% sure. The nightshot green is not the true color seen by the sensor. I have seen 707 shots taken in normal mode with IR blocking filter removed on the web ... my memory is they were purple but this needs verification.

Geoff
MaxEntropy said:
Alan,
First off, thanks for all your diligent work concerning this
phenomena. But I think that what you may have been measuring is
the spectral sensitivity differences of the color filter array in
the IR region.

The materials and procedures used to make the color filter array
mask on the photo sensor will vary from model to model and even
greater variations could be expected between manufacturers. The
closest I can find to the spectral sensitivity of the 828 is the
one for the 717 which I know everyone would state cannot be the
same as it is only a three color and not the four color filter
present on the 828. As shown below, this chart does not show us the
response in the range of interest, i.e. wavelengths longer than
700nm.





Without the information we need to see what the spectral response
of the 828 is in the IR region we are reduced to speculation. But
just to show that IR could be detected as any color, here is a
chart from a Kodak sensor. This at least shows that, without the
addition of an IR cut filter in front of the sensor, it is indeed
possible that the sensor would respond strongly to IR at about
800nm.



Depending on the individual color's filter response in the IR
range, any color, or any combination of colors, could be seen as
the response to an IR source. Example: if the green filter had
more attenuation at 800nm than the blue or red the result would be
displayed as purple.

This may or may not explain what you have seen in your experiments
with your particular camera, or what's happening on the 828, but it
may be something to consider.

Take care, good luck.
Click to expand...
--
http://www.pbase.com/geoffb
 
Well jeez ... that's a piece of cake to test. Go to almost any photo store and buy a UV filter. Cheap ones can be had for less than $10.

I'd test it myself but I'm lacking the required 828 :-)

Hot filters are a little harder to find and more expensive.

Geoff
MaxEntropy said:
The following crops are all taken from Phil's PF/CA test photo for
the F828. You can download the entire photo here:

http://www.dpreview.com/ ...
... SonyDSCF828/Samples/Fringing/DSC03559.jpg

These crops are of some of the small holes places around the test
pattern and at the center as noted. The approximate level of the
three color channels in the center of each hole is also noted.
Please verify this on your own, if you feel the need.





The lateral CA can be seen on all the perimeter holes as a color
shift of red/yellow on the edge towards the center and blue/purple
shift towards the outside away from the center. Thus the CA is
mis-focusing the longer wavelengths (red/yellow) toward the center
and the shorter wavelengths (blue/purple) away from the center.
Notice also the PF present at the left-top hole, the PF is away
from the center, where shorter wavelengths would be predicted to be.

I believe, in this test, for this camera, this photo shows that IR
would not be the cause for the PF shown, unless there is some rapid
shift in the lens response to IR that would send IR toward the
outside away from the center, but I don't think that is possible.

It would seem more likely that there is a shift of the UV, shorter
wavelengths, due to the CA present in the lens away from the center
and that there is a response to UV on the sensor, and that due to
differences in the respective color filters, it is being sensed as
blue/purple. For this to happen, the green filter would have to
attenuate the UV more than the blue and red filters in the UV
region. In the absence of a spectral response chart in the UV
range for the 828, this is just guess work.

Is this the answer? I don’t know.
Is it possible? Maybe.
In any case, it’s fun to think about!

Good luck.
Click to expand...
--
http://www.pbase.com/geoffb
 
AlanF said:
It's possible. But in the Sony Lens test
http://www.dpreview.com/reviews/sonydscf828/page16.asp

The Purple Fringe around a highlight is towards the Outside of the
frame .. which is most likely what Infra red Chromatic Aberration
(IRCA) will produce.

If it were UVCA it would most likely be towards the inside of the
frame.
Click to expand...
So what? With the Sigma SD9 it was also all over the edges and not "inside" and it was still UV contanimation, as Sigma/Faveon admitted on Phils request. We will never know whats wrong with the 828 till someone tries stronger cut off filters. Either UV or IR...

Sony's "leaked" (?) comment on Lens coatings is already an indication in that direction. Lenses usually have anti UV coating but not anti IR coatings...
AlanF said:
(Though I can't say that for sure .. it depends on the makeup of
the individual lens elements.)
Click to expand...
And on the sensor...
AlanF said:
Also note in Phil's Sony 'torture' test that the PF decreases as
the aperture number is increased. I believe that's because the lens
is can focus the IR
better. Thus the IR is focussed in the place as the visible light,
and all the sensors are saturated.
Click to expand...
Again you don't have a point, stopped down more it also focusses UV better...

--
Bad boys bad boys
Watcha gonna do, whatcha gonna do
when they come for you
 
GeoffB said:
Well jeez ... that's a piece of cake to test. Go to almost any
photo store and buy a UV filter. Cheap ones can be had for less
than $10.
Click to expand...
No, they don't cut into the wavelength you need to filter out...

I would almost blame deep blue for that, therefore you need something that cuts the deep blue as well.

--
Bad boys bad boys
Watcha gonna do, whatcha gonna do
when they come for you
 
MaxEntropy said:
It would seem more likely that there is a shift of the UV, shorter
wavelengths, due to the CA present in the lens away from the center
and that there is a response to UV on the sensor, and that due to
differences in the respective color filters, it is being sensed as
blue/purple. For this to happen, the green filter would have to
attenuate the UV more than the blue and red filters in the UV
region. In the absence of a spectral response chart in the UV
range for the 828, this is just guess work.

Is this the answer? I don?t know.
Is it possible? Maybe.
In any case, it?s fun to think about!
Click to expand...
Imho this is posting is spot on, at least it confirms what I am claiming for days.

I asked someone to do some tests in that direction with a proper UV cut off filter and not just a normal UV 1 filter. Lets hope I get a decent answer sooner or later.

--
Bad boys bad boys
Watcha gonna do, whatcha gonna do
when they come for you
 
BigMouth said:
GeoffB said:
Well jeez ... that's a piece of cake to test. Go to almost any
photo store and buy a UV filter. Cheap ones can be had for less
than $10.
Click to expand...
No, they don't cut into the wavelength you need to filter out...

I would almost blame deep blue for that, therefore you need
something that cuts the deep blue as well.

--
Click to expand...
That exists too ... forget what it's called. Skylight filter?

Geoff
BigMouth said:
Bad boys bad boys
Watcha gonna do, whatcha gonna do
when they come for you
Click to expand...
--
http://www.pbase.com/geoffb
 
I'm not sure what point you think I'm belabouring.

I have show that IR DOES produce purple in my camera.
I have said that UV may also produce chromatic effects.

Why don't you run a similar test to the ones I ran, but with a UV source
and filter?

And in another post I DID recommend a BAND-PASS filter.

OK, so maybe the thread should be called NVCAPDSFS
Non-Visible Chromatic Aberation Plus Differential Sensor Filter Sensitivity.
 
MaxEntropy said:
The lateral CA can be seen on all the perimeter holes as a color
shift of red/yellow on the edge towards the center and blue/purple
shift towards the outside away from the center. Thus the CA is
mis-focusing the longer wavelengths (red/yellow) toward the center
and the shorter wavelengths (blue/purple) away from the center.
Notice also the PF present at the left-top hole, the PF is away
from the center, where shorter wavelengths would be predicted to be.

I believe, in this test, for this camera, this photo shows that IR
would not be the cause for the PF shown, unless there is some rapid
shift in the lens response to IR that would send IR toward the
outside away from the center, but I don't think that is possible.

It would seem more likely that there is a shift of the UV, shorter
wavelengths, due to the CA present in the lens away from the center
and that there is a response to UV on the sensor, and that due to
differences in the respective color filters, it is being sensed as
blue/purple. For this to happen, the green filter would have to
attenuate the UV more than the blue and red filters in the UV
region. In the absence of a spectral response chart in the UV
range for the 828, this is just guess work.

Is this the answer? I don’t know.
Is it possible? Maybe.
In any case, it’s fun to think about!
Click to expand...
I haven't yet found a formula for lateral CA -- and in any case it depends so much on the exact lens shapes and compositions. A particular lens combination might over-compensate in the IR or UV bands, and reverse what you would expect from a simple lens.

But here's another article that says that IR is on the outside, UV on the inside:

http://www.cartage.org.lb/en/themes/Sciences/Physics/Optics/Optical/Lens/Lens.htm

Longitudinal CA:



Lattitudinal CA:

 
You must log in or register to reply here.

Keyboard shortcuts

f
Forum
m
My threads
Mobile site
About
Editorial content
Cameras & Lenses
Community
All content, design, and layout are Copyright © 1998–2025 Digital Photography Review All Rights Reserved.
Reproduction in whole or part in any form or medium without specific written permission is prohibited.
When you use DPReview links to buy products, the site may earn a commission.
©GPS Media - Guides, Products, Services.
Back
Top