DV Info Net

DV Info Net (https://www.dvinfo.net/forum/)
-   Canon GL Series DV Camcorders (https://www.dvinfo.net/forum/canon-gl-series-dv-camcorders/)
-   -   To purple or not to purple (https://www.dvinfo.net/forum/canon-gl-series-dv-camcorders/5386-purple-not-purple.html)

Aaron Koolen December 9th, 2002 01:42 PM

To purple or not to purple
 
Yet again I shot a friends wrestling show at the same venue as last time and again the purples on the wrestlers tights all came out blue. The camera was on auto, no CP, manual white balance at about f2.8. The lighting was flourescent.

I've read some of buddy's posts regarding colour rendition and people have been saying that maybe setup/brightness was something to do with it because some of the colours needed to make blue into purple get clipped off. I understand the concept, but this example seemed pretty extreme. I thought I would have got it looking some sort of purple, and not totally blue. Anyone got any insight into what could be causing this - would the floursescents do that much damage to purple colours, even though the other colours looked pretty good?

If I get the chance to try again at the venue I'll take something purple and see what I can do.

Ken Tanaka December 9th, 2002 03:05 PM

Hard to say, Aaron. Yes, I believe that the lighting caused this. The only thought I could offer is to make -very- sure that you carefully manually white balance to a white card placed as close to your main subject as possible and be sure to stay away from "Green Box" auto mode. Personally, I always shoot in Tv, Av or fully Manual modes. Check your colors in the viewfinder or lcd before you roll. There should be no hint of green in white areas.

Barry Goyette December 9th, 2002 03:42 PM

Aaron

I also believe the flourescent lighting is your culprit. Most flourescent output is in a relatively narrow band that is almost exactly opposite the purple-magenta range, meaning even with correction from the camera, or a filter, there is probably little left to keep those colors from running blue. My experience with the gl2 is that purples are incredibly vivid...much more so than with my xl1s or gl1.

Barry

Nathan Gifford December 9th, 2002 04:15 PM

I'd bet on the lighting too. White balance, color temp, and anything but auto will help...as does your recently acquired experience!!!

The latter always seems like the best educator...

Happy Holidays,

Bill Hardy December 9th, 2002 06:34 PM

Yes...it's puzzling.
I am especially impressed with the GL2 rendition of the color purple. Perhaps your monitor may have something to do with it. How are you watching the footage? I would recommend a simple test using a SVHS chord from the cam straight into a different monitor. Press play on the cam to check the fundamental signal. From there it should be a matter of tracking down the cause, whether it be florescent lighting or some other anomaly. Do let us know of your results.

Check out TSU Paradegl2b.jpg

http://homepage.mac.com/bhardy3/PhotoAlbum8.html

Graham Bernard December 10th, 2002 12:51 AM

Buddy - You may have a point. I took my spanking new XM2 with some daylight footage that had within it some "red" areas. They looked "wrong" - ie the edges of the red signs looked out of focus - yes? Well, my pro friend adjusted his TV set with just a little less saturation - brilliant! And I could then see that I hadn't got the focussed set correctly. Still learning. - Oh yes his set-up is for his Sony DX9000. Any clue there?

After all a TV is a type decoder itself. Yes?

Grazie

Aaron Koolen December 10th, 2002 03:36 AM

Thanks guys for your input. I was in auto mode but manually white balanced so the cam should not be adjusting any colours further. The wrestling mat was white (almost as white as my white card) and I used that to white balance to as I didn't have my card (Lesson learnt ;) ) Looking through the LCD, EVF and on my computer monitor the purple comes out blue. On the monitor it's not as bad but still no where near the purple that I saw with my eyes.

Barry can you explain what you mean by the flourescent band? I thought that white balancing should counter the green of flourescents and make it all look correct (all the other colours I looked at seemed to be pretty true)

Thanks again all
Aaron

Nathan Gifford December 10th, 2002 05:59 AM

Fluorescents are a different animal
 
Fluorescents emit light differently: one of the reasons why they are so effcient. Color temp and CRI has a big impact on DV reproduction. That's why there are special fluorescent lights for DV work.

If you would like to read more try http://www.studio1productions.com/Ar.../FL-Lights.htm

Happy Holidays,

Andre De Clercq December 10th, 2002 06:03 AM

Aaron, fluorescent lights are apart from being "pulsed" sources which generate flikker and some other strange effects (WB shifting if combined with incandescent sources)) are also lightsources with peaked (non smooth) spectrum. This means that some colors in the spectrum are much stronger (or weaker) represented than others. This "uneveness" is expressed in the CRI (color redition index) for lightsources. As you know, and this is for your problem, purple is composed (additive color mixing) out of blue and (deep)red. Most fluorescents have probems with red rendition. So if therei is no deep red in the lightsource you mis the red component for making purple and you end up with blue. White balancing has no problems with this because it can still balance the redish part of the spectrum against the greenish and blueish part. Also the human eye doesn't notice this uneveness and this is being described in the "color metamerism" effects for human viewing.

Barry Goyette December 10th, 2002 09:34 AM

Aaron,

Andre has described this pretty well, but just to reinforce what he said I'll offer this. All light sources emit light in a spectrum of visible wavelengths from near infrared to ultraviolet. Depending on the source, there is a concentration of output that establishes what we experience as "color balance", yet daylight and incandescant sources do emit light accross almost the entire spectrum.

Flourescent and other discharge light source emit almost all of their light in a narrower color range around their "spike" point. And in your case, the particular tube is probably a cool white, which produces very little red light....thus there is relatively little red light to reflect back from the purple clothes (purple, especially deeper shades, is typically a difficult color to render regardless of the light source, as it absorbs most of the light that hits it). Balancing won't help this situation, as filtration will only remove some of the excess yellow-green in the light source, but will do nothing to add back in the red.

You may find that in a different situation, with a more full spectrum flourescent (very common in retail and office environments), that you won't have this problem.....

This narrow band output phenomenon can be seen most acutely in the sodium vapor street lights that are common in suburban america. They are popular with local governments because of their low power usage, and give off an eerie, yellowish light. If you look at this light's spectrum output curve, it looks like the washington monument on capitol mall. One skinny spike in an otherwise flat landscape.

This light produces virtually NO light in any color outside it's very narrow color range, ( thus it can be used in photo darkrooms almost full strength--no filtration, without affecting sensitized materials...I remember the first time I used a darkroom with a sodium vapor safelight, It was so bright in there, that I literally couldn't believe that I wouldn't fog my photo paper ). This type of light is impossible to white balance ANY color back to normal (except yellow-orange), although, for what it's worth, the gl-2 does a better job than my xl1s with sodium vapor sources.

Hope this explains it.

Barry

Andre De Clercq December 10th, 2002 10:03 AM

These days you don't need an expensive optical spectrum analyser if you want to get a (rough) idea of the CRI and the peaked character of some lightsources. Just put a CD(rom) at the right angle so that you see the "rainbow" of that source. After some exercise you will be able to see e.g. the strong green component and weak red component in the cold light fluorescent tubes. You will also be able to see the broad and smooth spectrum of incandescant lightsources. If you point your camcorder towards the "rainbow" reflected from a spectrally smooth lightsouce you will even be able to see "gaps"(if there are) in the CCD color splitting/filtering system.

Aaron Koolen December 10th, 2002 01:27 PM

Thanks guys, that's clears that up for me nicely. Man, our eyes are amazing things...or faulty, whichever way you look at it :)

Cheers
Aaron

Andre, thanks for that tip about the CD, I must give it a whirl.


All times are GMT -6. The time now is 04:31 AM.

DV Info Net -- Real Names, Real People, Real Info!
1998-2024 The Digital Video Information Network