10

I've noticed that lights (for instance, car headlights, and sometimes other pinpoints of light) in movies occasionally have a streaky flicker emanating from them on alternate frames.

You can see an example of it in this shot. It's easily noticeable below the headlights, especially in the second half of the clip. This is from Tales from the Crypt (1972). I've also seen this effect on shots where the sun is just out of camera range, but lens flare from it is visible in the shot—the flare will sometimes have this flickering effect on it, so it apparently has nothing to do with whether the light source is natural or artificial.

Look below the headlights to see the flicker.

example clip

Note: The embedded GIF above is a low-quality version of the source video clip, as it had to be resized to meet SE Imgur’s 2 MiB file size limit. You can watch the source video with higher quality and FPS at this link: https://rumble.com/v3xiau1-flickering-headlights.html.

Alternating flicker/streak in sequential frames

screencap

screencap

10
  • 2
    My guess is that it either has something to do with a rolling shutter effect of the Mitchell BNCR they used to film this, or a compression artifact resulting from viewing a different format than the video was originally displayed in.
    – Rick
    Nov 24, 2023 at 13:40
  • 1
    "I've also seen this effect on shots where the sun is just out of camera range" do you remember if these were panning shots, i.e. where the the Sun (although out of view) was moving? Nov 24, 2023 at 13:53
  • 3
    BTW, please add to your question that it is about the downward streaks directly underneath the headlights. It took me ages to solve this spot the difference puzzle. Nov 24, 2023 at 14:13
  • 1
    Joe - do you have a timestamp for this. it might be worth seeing if there's another transfer somewhere to try & eliminate 'original shot' vs 'transfer artefact'.
    – Tetsujin
    Nov 24, 2023 at 18:15
  • 1
    @SeriousBri I missed the flicker at first, but look at the reflection of the lights on the road. There's unrealistic flickering light on top of them, reaching nearly to the bottom of the frame. I the upper still, there's a brighter stripe of light down from each headlamp than in the lower still, especially on the passenger side of the car (apparently a British production so right hand drive)
    – Chris H
    Nov 25, 2023 at 13:46

2 Answers 2

7

The question detail added since I wrote this pretty much eliminates my early theories. I've now split the original answer out below what is still mainly guesswork.

If you examine this sequence frame by frame, you'll see these stripes appear & disappear exactly every other frame. That makes me think it is not in the digital transfer or any subsequent processing, but in the original footage. I cannot think of any transfer process that would divide these frames so precisely.
I have looked at the current BluRay release of this as well as the gif & excerpt posted by the OP & it matches exactly. The BluRay is at 23.98 fps, which is a bit of a kludge to map to the old US/Japanese NTSC video system Don't ask;) , but is close enough to the original 24fps that inter-frame aberration should happen infrequently.
I don't think it could be any kind of physical 'shake' of the vehicle or lights - because the chances of that syncing perfectly to frame rate would be astronomically small.

The only conclusion I can draw at the moment is that there may have been something in the shutter/gate mechanism that presented itself slightly differently on alternate frames. I'm not really sure how this may happen. I've discovered that the camera used, a Mitchell BNCR, may have had an optical splitter fitted, which would present light to the film & to an eyepiece for the cameraman.

Quite how this functioned I don't have the expertise [or the Google-fu] to know for sure.
There are a couple ways this may have been achieved.

There may have been a moving pentaprism design somewhat akin to a [D]SLR ([digital]single lens reflex) camera - alternating between sending light to the film & to the camera op. To my admittedly limited knowledge, there was a 'movie' version of this which rotated to send the light in alternating directions - to film, then to the camera op, 24 times a second.

It may have used a beam splitter, the science of which hurts my brain;) The simplest version of this seems to be the pellicle mirror. These were first introduced as early as 1938, but even to this day they are not popular in 'stills' cameras. The pentaprism remained the most common structure until quite recently when mirrorless cameras started to come to the fore. These use a digital screen as a viewfinder, eliminating the need for a separate optical path.

See When recording on film, how does the crew see the footage? more more info on modern video taps.


Original answer below.

The streaking is an aspect of the lens flare you can see through most of the shot - light being reflected inside the lens and camera body itself. It's at its 'worst' when the headlights are pointing more directly at the lens.

The flicker I can only guess is because they're LED headlights. You can get completely flicker-free LEDs, for such as video lighting or anywhere near fast-moving machinery [drills, power saws etc] where the flicker would be annoying or outright dangerous, but usually LED brightness is determined by how long the light remains on compared to off. This is known as PWM [pulse width modulation].
I am no electronics expert, so I've never fully investigated why they do it this way when it's possible not to, but my first thought would be 'it's cheaper'… as is usually the case.

This gives an explanation - https://www.analog.com/en/design-notes/lowpower-pwm-output-controls-led-brightness.html Don't ask me about anything you don't understand ;)

18
  • FWIW: "an aspect of the lens flare you can see through most of the shot": and because they have very similar characteristics, an aspect of the behaviour of the human eye when confronted to such lights. If you age, if your eyes suffer from night blindness, glaucoma, cataract, diabetic retinitis or some others, you'll see exactly the same halos around car lights, traffic lights when directly seeing them.
    – OldPadawan
    Nov 24, 2023 at 9:00
  • 11
    It's not likely that they're LED headlights. This is from a movie made in 1972. I've updated my question to reflect that. My thanks to whoever made it into a gif.
    – JoeB
    Nov 24, 2023 at 11:43
  • In which case, I'd have to guess self-levelling headlights being a bit twitchy, or even just a slight shake from the road surface. Looks like a Roller, but idk when self-levellers came in. I have them on my Merc, but that's decades newer.
    – Tetsujin
    Nov 24, 2023 at 11:54
  • 2
    For LEDs, its because its a lot simpler to do PWM. The brightness of an LED is proportional to the current, and in the operational range, the current through an LED changes rapidly with voltage, (a mV of voltage can change the brightness by 10% or more). Which means that to control current directly you need expensive precision analogue electronics, whist with PWM you need cheap imprecise current limiters, and a cheap digitally controllable switch. Nov 24, 2023 at 18:48
  • 1
    Thank you all for your comments. My apologies for not including more info in my original post. I've long suspected that the flickering is some kind of shutter issue, as suggested by Tetsujin, Rick, Jaap Scherphuis, and Steve Pemberton. Perhaps there's dust on the shutter, or something's miscalibrated--but I was hoping for a definitive answer.
    – JoeB
    Nov 24, 2023 at 22:25
3

There is a lot going on in this clip, and while I cannot claim to offer a definitive answer, I would like to add some observations.

(Edited) I originally noted that the flickering in the GIF was not every other frame, and only later was advised that the GIF is not a frame-by-frame transfer but is rather running at 20fps. Looking at the video clip, I see that it is indeed showing streaks every other frame.

There is nothing I am aware of in the Mitchell or any other 35mm movie camera that operates on every other frame: the entire mechanism is built around one cycle per frame. Likewise, there is nothing I am aware of in the film printing process that alternates frames. However, this is not the case with the conversion of film to video.

There are several different ways to transfer film to video, and some of them use dual cameras or rotating mirrors or prisms in a way where adjacent frames go through different optical components. My best guess is that this was transferred using a rotating mirror telecine and that one side of the mirror was dirty.

Side note on telecine: a big engineering problem the first telecine machines had to solve was converting 24fps film to 60 fields (30 interlaced frames) per second NTSC television, which made them complex. However, because there was also a need to convert to 50 field (25 interlaced frames) per second PAL television, machines were improved to be variable rate, and then eventually cameras were changed to capture higher resolution progressive scan images. My best guess is that one of these kinds of machines, rather than a more expensive frame-by-frame laser scanner like the kind used now to create digital intermediates for visual effects, was used to transfer this film to video.

One other observation: There is very little motion blur in the image. Looking at frames 110-112 of the GIF (and corresponding frames in the video), you see huge jumps in position, but hardly any blur. This suggests it was shot with a very narrow shutter angle.

Frames 110-112 in miniature

The narrow shutter angle initially interested me as a possible explanation for why the streaks were sharp in some frames and not in others, but seems less relevant when the video is viewed and the flashing is seen in every alternate frame.

2
  • 1
    I was the one who converted the video from the link the OP provided and embedded it as an animated GIF. Due to the SE Imgur file size restriction, I had to convert the video to a 10 or 12 FPS GIF (I forgot which exactly), so it might not be accurate to reference it frame by frame. It would be more accurate to reference the original video provided by the OP. Dec 2, 2023 at 2:38
  • @galacticninja Thank you for the info. I did not realize the "GIF Source" link linked to the full video. The GIF animation runs 20 FPS and I don't know how the rate conversion worked (maybe by dropping frames), so maybe that explains the odd cadence of flickering. My comment about shutter angle and blur still stands, as you can see in the video's final frames, but it may now be irrelevant.
    – Sara
    Dec 3, 2023 at 2:53

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .