Question: Shooting under sodium vapor lights

Discussion in 'Professional Video Production' started by PTravel, Jul 2, 2006.

  1. PTravel

    PTravel Guest

    Sodium vapor lights are the common night time illumination in most cities in
    the world, and put out a distinctive yellow light. Because they work by
    exciting sodium atoms, the light output is a very narrow spectrum, i.e. it's
    not just a question of being warmer or cooler -- there simply are no other
    bandwidths present. Naturally this results in anything being videotaped as
    appear, essentially, monochromatic (for an example, see here:
    http://www.travelersvideo.com/amsterdam at night.wmv)

    Is there any way around this to get, if not a more accurate color rendition,
    at least a more interesting one? Short of bringing my own lights, which is
    not feasible at all, I can't think of anything that will add spectra that
    simply aren't there.
     
    PTravel, Jul 2, 2006
    #1
    1. Advertisements

  2. You answered your own question already. You would indeed need to bring in
    your own light with a broader spectrum. But I don't see you drag around a
    Musco-truck all the time :)) You could add some filters to the lights, but
    they merely change what you have, like shifting the spectrum a few hundred
    Kelvin, but it stays equally narrow-banded.

    Compare it with playing a CD via the telephone. You cut out everything
    below 300Hz and above 3kHz, and you won't be able to recreate the London
    Symphonic Orchestra in all it's glory again.

    cheers

    -martin-
     
    Martin Heffels, Jul 2, 2006
    #2
    1. Advertisements

  3. PTravel

    Larry in AZ Guest

    Waiving the right to remain silent, "PTravel"
    Experiment with the following if you have some MV lighting in a nearby
    parking lot:

    1. Try manual or auto white balance on a white card right under one of the
    lights. Then, shoot some wide stuff.

    2. If your camera does a constant white balance act, try adding a 1/4 blue
    filter in front of the lens.

    3. Combine the above two.

    These experiments will not help much when your light is mixed, such as
    when you've got those neon signs and other light sources in the shot.

    Nice footage, by the way, Paul. Just one little critique - you seem to
    favor a slight tilt to the right in several scenes.
     
    Larry in AZ, Jul 2, 2006
    #3
  4. To keep the buildings straight?
     
    Martin Heffels, Jul 2, 2006
    #4
  5. PTravel

    PTravel Guest

    Thanks. Neither my tripod or my fluid head has a bubble level, so I have to
    eyeball it. The particular night that I shot was cold and a little rainy
    and I really just wanted to get it over with -- usually, I'll dress for a
    bad weather shoot, but I didn't this time. However, the level problem does
    appear to be chronic problem for me. Does anyone know a source for an
    inexpensive, stick-on level? Or maybe I need to get my eyes adjusted.
     
    PTravel, Jul 2, 2006
    #5
  6. PTravel

    PTravel Guest

    I was thinking more in line with the video equivalent of an audio
    compander -- something that can take the limited range that's available and
    expand it. For example, if Na lights put out light limited to 240-245
    angstroms (I'm just making up numbers here), something that would
    extrapolate on the video so that the lower range fell closer to red and the
    upper range closer to blue would provide a more varied, albeit false,
    coloration.

    Hmmm. Have I just invented a new filter?
     
    PTravel, Jul 2, 2006
    #6
  7. PTravel

    Ed Anson Guest

    You appear to be working against physics.

    According to Wikipedea (http://en.wikipedia.org/wiki/Sodium_vapor)
    "These lamps produce a virtually monochromatic light in the 589 nm
    wavelength. As a result, objects have no color rendition under a LPS
    light and are seen only by their reflection of the 589 nm light."

    Filters work by removing light at unwanted wavelengths. Given a
    monochromatic light source, no filter can do anything but reduce the
    intensity of what is there. Even if the light were simply a narrow
    spectrum, you would need special sensors to do what you suggest. You
    certainly couldn't do it with a normal camcorder.

    I'd be inclined to go with the monochromatic effect. You can replace the
    sodium color with any tone you want, and produce a pleasing image. But
    there's no way you can get a polychromatic image using monochromatic light.
     
    Ed Anson, Jul 2, 2006
    #7
  8. Want to have Ira Tiffen's e-mail address? :)

    You could with some clever mathematics expand the range and fill in the gap
    on your computer, but it won't help you with the environment the
    sodium-vapour lamps are lighting. Your green rental car will still look
    like it has been "pimped" to black while you were eating in the restaurant
    :) And if you didn't remember the license plate......

    cheers

    -martin-
     
    Martin Heffels, Jul 2, 2006
    #8
  9. PTravel

    PTravel Guest

    Sodium lights contain impurities, so the output isn't a perfect 589 nm --
    otherwise you could use it as a laser. If you put the output through a
    spectrum analyzer, you'd see a sharp spike at 589nm, but there'd be fall-off
    on either side of it, i.e. though most of the light output is at 589nm,
    there may be light output as much as 10nm (or whatever) on either side of
    it.

    When I used the term "filter" I didn't mean an optical filter, but a
    post-processing filter that would expand the bandwidth so as to create a
    false-color effect.
    Pure monochromatic light, no. Almost pure, yes.
     
    PTravel, Jul 3, 2006
    #9
  10. PTravel

    PTravel Guest

    That's exactly my thought.
    Because the green car will reflect some of the higher bandwidth "spill" from
    the sodium vapor light, whereas the dark red sweater will reflect some of
    the lower bandwidth spill. It should be possible to mathematically expand
    the departure from the "pure" color of the sodium lamp to produce a false,
    but more varied, color spectrum. The key, here, is that the sodium vapor
    lamp contains impurities and does output light only at the single spectrum
    for Na.
     
    PTravel, Jul 3, 2006
    #10
  11. PTravel

    Rôgêr Guest

    To add to what the others have said, you can correct to some degree in
    post production. I get that same color cast on my GL2 while a cheap JVC
    automatically adjusts the color balance to where the color cast is gone.
    And so is most of all the other colors. There's some color, but just barely.
     
    Rôgêr, Jul 3, 2006
    #11
  12. Nice stuff Paul.
    There really isn't much you can do with limited resources other than
    some correction in post. (bringing a chip chart and shoot that under
    the light for reference can help later in post)
    However............
    Get yourself a rosco cinegel color correction sample set. (1 1/4" x
    3")
    Then you can experiment with white balance through different
    correction filters.
    Or add it as correction filtration
    I'm thinking that something in the Tough Plusgreen family and Tough WF
    green either added at the lens or subtracted through white balance
    might bring you a wee bit closer to what you are looking for.


    Bill F.
     
    Bill Farnsworth, Jul 3, 2006
    #12
  13. PTravel

    PTRAVEL Guest

    Thanks, Bill!
     
    PTRAVEL, Jul 3, 2006
    #13
  14. PTravel

    Toby Guest

    To what end? Who would want such a thing? Besides, you would end up with a
    horrible posterization effect. Imagine trying to recreate 16 bit color from
    4 bit color information...

    Toby
     
    Toby, Jul 3, 2006
    #14
  15. PTravel

    PTravel Guest

    I've never understand a question like this.

    I would want something like this. To my own end. Does this bother you?

    As for posterization, I don't think so.
     
    PTravel, Jul 3, 2006
    #15
  16. PTravel

    Toby Guest

    Not at all, but I'm pointing out that if you want it you'll probably have to
    develop it yourself, unless you are a particulary smooth talker.
    Well, yes indeed. You need to create levels where none exist. I suppose you
    could do some fancy interpolation to blur the edges, but you are still going
    to have to create something out of nothing, so to speak. You are talking
    about taking minute chroma differences and stretching them across the RGB
    plane, so to speak. Where are you going to get all those extra gradations?

    I watched your video (by the way, you can get a small bubble level as a
    keychain. Don't get a round one - you only want to check horizontal, not
    vertical, when you mount it on a camera). You have lots of mixed light
    there. What do you plan to do when you have incandescent at 2700K and
    fluorescent at anywhere between 3200-4700K? What happens to those colors
    when you expand your sodium (non) spectrum?

    I don't really mean to rain on your parade, but it's not so simple as you
    seem to think.

    Toby
     
    Toby, Jul 3, 2006
    #16
  17. PTravel

    Toby Guest

    I've looked at sodium discharge lights through a viewing spectroscope. The
    two dominant D-lines around 589 nm constitute well over 95% of the light
    emission, and that's being generous. They So you have a two-fold problem.
    The first is notch-filtering at 589 nm to get rid of the source that is
    going to blow everything else out of the water. Then you have the residual
    couple of percent to boost up to some level where it does you some good.
    Kind of impossible....

    Your best bet is to make sure there are some other sources of full-spectrum
    light in your shots.

    Toby
     
    Toby, Jul 3, 2006
    #17
  18. Great idea. I knew all those gel sample packs would
    come in handy some day! :)

    Alas no amount of white-balance or painting the R/G/B
    gains/pedestals will produce a "color" picture under a
    source as monochromatic as sodium vapor. Not with
    electronic video, nor with chemical film.

    We have a similar situation in the cleanroom fabs where
    we manufacture microprocessor wafers (perhaps the one
    in the computer you are reading this on!) . Part of the fab
    is illuminated with orange light that looks for all the world
    like those "bug lights" people use for porch lights to
    avoid attracting insects. The photo-sensitive "resist"
    that we use to expose the patterns on the wafers is not
    sensitive to this partiuclar wavelength, so it is very much
    like a photography darkroom "safelight".

    But the light is so unbalanced that it is nearly imposible
    to properly white-balance under. One of our cameras
    will almost do it, but we discovered that people who
    are looking at the video become disoriented when they
    see the "litho area" looking as if it were illuminated in
    white light. So our SOP is to white-balance in the white
    illiminated area, and then to just shoot in the yellow
    litho areas as-is and let the video go yellow.
     
    Richard Crowley, Jul 3, 2006
    #18
  19.  
    Richard Crowley, Jul 3, 2006
    #19
  20. PTravel

    David McCall Guest

    Good points. I once tried to get an HL-79 to balance under
    raw sodium lights and it really looked awful. There just isn't
    anything for the camera to work with. The spikes are in the
    middle between Red and Green, with no Blue at all.

    What you said. If there is any part of the scene (or scenes
    adjacent to this scene) in which the Orange lighting is visible)
    that has full lighting, then white balance to that. Otherwise
    just use your tungsten preset (3200) and let it be orange.
    You can tweak it a bit in post, but it will never look close.
    There just isn't anything but orange. You can reduce the
    saturation and shift the hue a little. But that is about it.

    I think Paul's video looked real good. We are used to that look,
    so it really doesn't look out of place.

    David
     
    David McCall, Jul 3, 2006
    #20
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.