View Full Version : Would polarizing filter help with oncoming lights?
Tom Gull April 10th, 2010, 08:01 AM I'm off to film the Amtrak Capitol Limited today and had a polarizing filter mounted for taking pictures of cherry blossoms on our street. It occurred to me that this filter might help damp down various kinds of light flare I see when filming a train coming right at me either at night or at certain angles during the day. I know it's not really aimed at reducing flare, but I'm wondering if it might in this limited case.
For reference, here's nighttime headlight glare in a clip I took in California - the first I've ever taken of a train approaching at night. The train was actually on camera for probably 30 seconds prior to this but I edited that out because the lights formed a big blob.
YouTube - Amtrak Surfliner at night at Grover Beach California 26 March 2010 (http://www.youtube.com/watch?v=jhzYaKbtqEM)
Does anyone have an educated opinion on this one, or has anyone tried using a polarizing filter to reduce flare from light sources moving right towards the camcorder?
Thanks / Tom
Jay Bratcher April 11th, 2010, 03:41 PM I don't think it would help. A polarizer will reduce reflected light, but lens flare is reflected light from inside the lens, after the light has passed through the polarizer (reflecting off various lens elements). Unless you can put a polarizer on the back of the lens, it wouldn't help. Even then, it might not help, and would probably force you to compensate for the darkening effect of the polarizer.
For what it's worth, I didn't think that lens flare in your sample was too bad, considering what the footage is. It looks like your lens is coated, so that may be as good as you can get...
Tom Gull April 11th, 2010, 05:27 PM Thanks, it looks like you're right. I tried three clips with trains coming at me yesterday and it didn't seem to have much effect in broad daylight, comparing those against prior clips. The cam actually does well in broad daylight, flaring a bit only if the light itself pulses strongly. Someone suggested that a polarizer might cut down on light reflected off the rails, which seemed to happen, but that light never bothered me in any case.
In daylight, the polarizer lens doesn't look to darken the video much if any. I was worried about the effect at night, though. Filming through a lens that looks black (the polarizer) seems counterintuitive at night at best!
David Sholle April 11th, 2010, 06:08 PM I don't think it would help. A polarizer will reduce reflected light, but lens flare is reflected light from inside the lens, after the light has passed through the polarizer (reflecting off various lens elements). Unless you can put a polarizer on the back of the lens, it wouldn't help. Even then, it might not help, and would probably force you to compensate for the darkening effect of the polarizer.
For what it's worth, I didn't think that lens flare in your sample was too bad, considering what the footage is. It looks like your lens is coated, so that may be as good as you can get...
Your comments got me thinking. Lens flare is partially polarized because it is reflections off of transparent surfaces. When unpolarized light (can think of as a mix of two perpendicular polarization directions) reflects at a glancing angle off of a transparent surface, the reflection becomes partially polarized (actually one of the two perpendicular polarization directions is more strongly reflected). Therefore, a polarizer on the front of the lens might actually help decrease flare, by limiting the light hitting the lens to be in the polarization direction that is less strongly reflected.
In theory, it should work, but in practice, I don't think it will. Many years ago, polarizers were simple linear polarizers. Such polarizers would block one of the two perpendicular polarization directions, and allow the other to pass, sending linearly polarized light to the film. This was switched over to circular polarizers because of the way SLR's began to use reflections off of the reflex mirror to judge exposure, etc., and the linear polarizers could give incorrect results. I believe all modern polarizers sold for camcorder and camera use are now circular polarizers. A circular polarizer blocks one of the two perpendicular linear polarization directions of incident light, but with the addition of a quarter wave plate, it then generates and passes circularly polarized light (as opposed to linearly polarized light) through the lens to the sensor. Unfortunately the circularly polarized light will have no effect on the flare in the lens.
Graeme Hay April 14th, 2010, 08:41 PM I have one of those 2 linear polarizers the scene turns dark(back one is circ but that is so the digital camera will meter), but from what I can see it doesn't reduce lens flare beyond make it less because less light is coming in, but once you up-ed the gain your probably no better off.
Guy McLoughlin April 15th, 2010, 10:45 AM The sample "YouTube - Amtrak Surfliner" looks like sensor bloom ( light that is too bright for the sensor to record properly effecting surrounding pixels ) to me, not lens flare. It's obvious that the train's headlights are at least 6-8 stops too bright for the camera to record them properly. I suggest that you might try under-exposing some trains to see how the lights record.
I suspect that there is a proper exposure for the train lights, which means to get your "night shot" you will have to shoot in early dusk light that might look too bright to your own eyes, but will photograph correctly with your exposure set for the train's headlights. ( i.e. Expose for the train lights, with enough daylight to record the surrounding people and buildings properly )
Tom Gull April 15th, 2010, 01:03 PM I'm fine with "sensor bloom" - I'm just not enough of a videographer to have known the term <g>. If lens flare is reflections bouncing around in-cam, then this was undoubtedly bloom - the light compared to surroundings just overwhelmed the sensor.
I very rarely film at night. The only reason this was an exception was because that's when the train went through and I wasn't in my home town. It was "catch it then" or never. So I haven't had a chance to practice technique for handling this kind of filming.
This clip was done at twilight right after I got the camcorder and I made no exposure adjustments.
YouTube - Amtrak Capitol Limited Harpers Ferry WV Twlight CX500V handheld (http://www.youtube.com/watch?v=JuJzXZTSkTY)
I see the bloom most of the way across the bridge. This is the only other clip like this I've taken, I think.
I wonder if I could keep the exposure different up to a point and then move the wheel fast enough to get back to what most of the night clip showed after the bloom...
Guy McLoughlin April 15th, 2010, 01:36 PM I wonder if I could keep the exposure different up to a point and then move the wheel fast enough to get back to what most of the night clip showed after the bloom...
I don't think this will give you what you are after, as proper exposure for the train lights will make everything else look almost pitch black in comparison. You really need to shoot earlier in the day, just after the sun has gone from the horizon, so that when you expose for the the train lights everything else will appear to be lit as early evening to night-time.
Sensor Blooming
...Just in case you think I made this term up, here's a quick link. Check out figure 3 :
Awaiba - CMOS Image Sensor Technology (http://www.awaiba.com/en/technology/)
Tom Gull April 15th, 2010, 02:59 PM I figured "sensor bloom" was a real term - it sounded exactly right to me.
Oddly enough, I can't set the time of day for these filmings. The trains run when the trains run, and the interesting ones with schedules I can look up run during the day. I suppose I could practice on some Washington DC metro trains, but then I have to explain to my wife why I'm wandering out at night, which may or may not go over well <g>!
Dave Blackhurst April 15th, 2010, 05:11 PM I'd suggest you may have to go "off axis" - keep in mind those lights are probably pretty powerful, but they also are fairly focused/narrow beam. SO, even a couple degrees off axis and you'd reduce the bloom/intensity of the light beam.
The problem is relatively simple - those lights are designed to make the track in front of the train as bright as possible for safety (anyone got a guess as to how many candlepower and the "throw"?), contrast this with a dusk/dark and you'll have issues - think of how our eyes iris changes between a bright day and and in the dark... or my favorite "concussion check" using a flashlight to test iris response - you can see the iris open and close to adapt!
The camera has to do the same thing, and while the EXMOR R has perhaps the best latitude to handle the range from bright to dark that you'll find, there are still limits - thus why you'll see "HDR" or high dynamic range tricks in some of the new sony P&S cams - the only way to increase the dynamic range is to combine two identical shots, one "over" exposed and one "under"... Sony actually has "DRO/Dynamic range optimization" built in as well, but again, there are limits.
Tom Gull April 15th, 2010, 06:38 PM I'm not real surprised at your advice, Dave. I didn't really expect too much filming this way - the opportunity was just there and I figured I'd try it. In fact, I really like the train from the side and I was fortunate to be filming right where the engine stopped. Considering I was in a strange town and had no idea how long the train was, I felt pretty lucky, headlight overload or not. My original polarizer question really was aimed more at daylight filming.
|
|