low light and high frame rates. A waste of time and money?

Jules

Distinguished Member
If you understand the 180 degree rule in videography, you’ll know that to get what we’ve come to expect as natural blur in motion, you should set your shutter speed to twice the frame rate.

So, if shooting at 25fps, your shutter speed (the amount of time each frame is exposed to light) should be 1/50th of a second.


So lets suppose you’ve purchased that fancy camera that shoots at 120fps, 180fps, or even 240fps.


At 240fps, your shutter speed needs to be 1/480th of a second. That means you need about 9 times more light to properly expose your shot than you would’ve done shooting in 25fps at 1/50th of a second.

Think about that for a second. In low light situations, most of us will be struggling to get proper exposure at 1/50th second using lenses with a very wide maximum aperture of f2 or wider.


At 1/480th of a second, you would need the equivalent of an f0.x lens, which doesn’t exist, and even if it did it would yield an unusably shallow depth of field even at ultra wide angles.


Aha… I hear you cry. Why not just light your scene better?

Good question, but remember that artificial light flickers… and in domestic situations in the UK, that flicker will be very visible with any fast frame rate above 50fps. That includes street lighting.


Sure, you can spend a lot of money on video lights designed for flicker free images at high frame rates, but now you’re talking silly money… and can you guarantee no other artificial light will enter your scene?


And here’s the other rub… even if you forked out on such fancy lighting and fully controlled your scene, most cameras won’t autofocus in their high frame rate modes. So you’ll need to be proficient at manual focusing…. are you sufficiently skilled to warrant the expense?

My advice is save super slow mo shots for bright sunny days, then use your iphone to shoot at 240fps.... the tiny iPhone sensor ensures almost everything is in focus anyway, so you don't need to worry about manual focusing.

Don’t let the marketeers lure you into purchasing their top of the line cameras that shoot in 240fps…. Unless you understand the limitations of the technology or you have the budget to work around it.
 
Last edited:

Terfyn

Well-known Member
Good question, but remember that artificial light flickers… and in domestic situations in the UK, that flicker will be very visible with any frame rate faster than 50fps. That includes street lighting
Battery powered LED lights??
 

rogs

Well-known Member
The other alternative - without any worries about lighting - is use an application to apply 'optical flow' to standard frame rate footage.
Apps like Twixtor (or it's freeware equivalent MVTools) allow you to control not only the degree of slo-mo applied, but exactly where and when it needs to be in the clip -- as well as 'easing' in and out if necessary.

Downside?..... optical flow doesn't always work well with all clips, and can give strange artefacts in some situations....It can be very effective though.

Of course if you're really serious about 'slo-mo' you probably need to hire a Phantom Flex :)
 

dosdan

Active Member
They can still flicker.. caused by the dimming control youll find on most of them.

I tried this on my V770 at both 50fps & 100fps. Used a manual shutter speed of 1/8000s. Light source was a 98-LED battery-powered Glanz/LS LED98A light. (Nothing fancy.)

As I understand it, changing the intensity is done though changing the duty cycle i.e. PWM. Pulled out my old oscilloscope to see if I could work out the rate the PWM is operating at through picking up stray EM emanations. But the CRT section in the oscilloscope has failed.

So I took shots of the dimmed light with my DSLR at a shutter speed of 1/8000s. No banding.

I then videoed it at 50fps and 100fps with a manual shutter speed of 1/8000s. Also tried varying the intensity. No banding or flickering/strobing visible.

Dan.
 
Last edited:

Jules

Distinguished Member
I tried this on my V770 at both 50fps & 100fps. Used a manual shutter speed of 1/8000s. Light source was a 98-LED battery-powered Glanz/LS LED98A light. (Nothing fancy.)

As I understand it, changing the intensity is done though changing the duty cycle i.e. PWM. Pulled out my old oscilloscope to see if I could work out the rate the PWM is operating at through picking up stray EM emanations. But the CRT section in the oscilloscope has failed.

So I took shots of the dimmed light with my DSLR at a shutter speed of 1/8000s. No banding.

I then videoed it at 50fps and 100fps with a manual shutter speed of 1/8000s. Also tried varying the intensity. No banding or flickering/strobing visible.

Dan.
What about at 120fps or higher?
 

dosdan

Active Member
What about at 120fps or higher?

I switched the HC-V750 from "PAL" to "NTSC". In slo-mo mode it then shoots at 120fps, but outputs it as a 60fps file, so the clip's duration is 2x the original event. Although 1080p60, the quality is not is good as 1x. I'm sure there some upscaling involved, and the image is softer and I think noisier too. Also the camera drops the recording bitrate to 28Mbps. Obviously, sensor read-out speed is a limitation here.

I started at 1/8000s and the lowest lamp intensity I could set. So the PWM duty cycle is at a minimum. So any change in intensity, at this fast capture speed i.e. high electronic "shutter" speed, will be most obvious. You'll need to watch the sequence at 100% zoom. Any smaller zoom level and the on-the-fly- downsizing in your viewing program will obscure the fine bands. I suggest you play the full sequence, and then check by single-stepping though a few frames (Ctrl-Right Arrow in MPC-BE). Here's a sequence of resulting 1/2x images (it's a good idea to d/l this 25MB Full HD MP4 clip first before playing):

https://dl.dropboxusercontent.com/s/3cbeyqfz7w363e1/LED Lighting Test.mp4


1. 1/8000s, 15dB camera gain, low intensity. Fine horiz. lines are visible on the tops of the lamp's LEDs. These move from frame to frame. But shooting a 120fps recording at 1/8000s is not how you'd normally operate.

2. 1/500s, 3dB gain, low intensity. No lines visible. Some camera noise evident.

3. 1/2000s, 15dB gain, low intensity. No definite lines visible, but the minor patterns in the noise appears to move upwards. More camera noise evident.

4. 1/2000s, 3dB gain, medium intensity (lamp brightness control half-way). Some camera noise visible.

5. 1/8000s, 15dB, medium intensity. No lines visible. More camera noise evident.

Based on this I'd say shooting 120fps under this light at 1/500s or slower would not be an issue. Also, if you had two dimmed LED lamps illuminating the same spot, any minor banding from each lamp, when used by itself, would probably not be perceptible when their illumination was overlapped.

Dan.
 
Last edited:

Terfyn

Well-known Member
My own experience with LED lamps is similar to Dan. No flicker problem when using them. The Neewer LED308 use electronic dimming methods rather than a resistance dimmer and this may make the difference.

Your concern may be resulting from the lamps you use.
 

12harry

Distinguished Member
I read a report from an LED-company that is promoting their own LEDs ( for others to build into LED-Lights ).
They suggested that their phosphors are "better" ( wouldn't they?), and they recommended using voltage to control the brightness, claiming it was worth the loss of efficiency ( Compared with PWM as others have noted ).
So I expect when you see "professional LEDs" which can cost £500 for a 12" Sq panel, these will use voltage dimming..... probably achieved by using a power-regulator chip, on a substantial heatsink.

However, if you read the chip's datasheet you will find that the loop-frequency is typically several 10's of MHz - so the chip is changing the output rapidly - although I must say the changes are very small and most unlikely to be noticed. Where it may be noticed is in the heat - esp. if the LEDs are set to a low-brightness and the battery supply is towards the upper end.

[ Street-lights used to be powered directly from the ac supply, so would strobe at 100Hz (in the UK) - however, modern lights use internal electronics to drive the components and this is at a far higher frequency - modern domestic cf lamps are typically operating at 10's of kHz - Filament lights ( like Halogen floods) have a thermal-lag effect which reduces flicker ]. In the past some folks were troubled by office-lights which did flash at 100Hz, but these types are almost no-longer used.

PWM is the norm for many light-sources . . . and some displays use a similar "chopping" to address the individual parts ( Since "multiplexing" saves wiring and components ), however, these techniques are invisible to our eyes ( unless you look sideways quickly!), as the frequencies are typically 32kHz.

It's an interesting point, re LEDs and SloMo - However, since more-light is Good, the chances are the LEDs will be running at almost full brightness and the flicker will be almost covered by the phosphor, which has a small after-glow.

My own Camcorder Sony CX410 (now discontinued), has a x5 SloMo feature - and it can't Focus at this setting, so you have to ( I prefer Manual, to eliminate any risk of "Hunting" ), set the Focus prior . . . as well as Zoom / frame-up.
Only then select Slo-Mo and wait to press the REC button. Ideally using the wired remote, although it is difficult to introduce "shake" at 5x 25 fps since it looks like a gentle pan.
With Slo-Mo there is no recorded Audio; so you'll have to make a Normal recording for "Ambience / etc."

That many camcorders are offering SloMo, rather suggests that they've cracked the problems re light . . . . probably using all the pixels to increase the active area and introducing extra processing. My camcorder takes its time after you press REC while it creates the new file. After about 12 seconds, or so, you can REC again in SloMo. If you close the LCD (=Power Off) then it's in Normal-mode when you power-up.

In the EDIT, I can stretch the time up to x4, so pre-render time looks like 125 x 4 = 500 fps - but of course the "Edit" doesn't affect light-levels.
I've used it for "wildlife" - like Bees but it is sometimes difficult to appreciate the effect, as we are exposed to TV-Wildlife material that often includes Slo-Mo ( so the audience can see the motion ) . . . bird-flight is regularly slowed, I'm told.

An Edit effect I particularly dislike is the change in speed - it just looks so false - In effect it shows the Editor's hand.
In the past the event ( eg a goal ) would be shown again, but in Slo-Mo giving the audience the chance to see how the score was achieved . . . Sports "Action-Replay" has been used for many years, even in distant analogue days, I understand.
 
Last edited:

200p

Well-known Member
If you understand the 180 degree rule in videography, you’ll know that to get what we’ve come to expect as natural blur in motion, you should set your shutter speed to twice the frame rate.
On TV, isn't stuff with 50Hz motion (like live stuff) usually shot with an "open" (360 degree) shutter (1/50th of a sec)?

If you are shooting at very high frame rates, I don't see why you couldn't use an open (360 degree) shutter/same shutter speed as frame rate. Technically it's the most accurate representation of what the camera sees during the "frame duration" time (talking progressive only here), and it will need less light than shorter shutters (or produce less noixse). Though the EBU subjective tests results said people preferred 120 fps and 180 degree shutter to 120 fps and an open shutter (slightly higher quality results according to that, for some reason - perhaps because it showed less blur?), but they thought 240 fps and open shutter was even better (bigger improvement).

And shooting with an open shutter at high frame rates would let you A) choose a new frame rate in post production (like with Ang Lee's film - shot at 120 fps) and B) it wouldn't (or is much less likely to?) give problems like wheels going backwards or other artefacts that occur due to the shutter only being open a fraction of the frame duration.
 
Last edited:

Johnmcl7

Distinguished Member
I'm not entirely sure what the point is that's being made here, I have a a couple of Sony cameras that shoot at 1000fps (RX100m4 and RX10m2) which I've used indoors with better results than I expected. The 1in sensor is pretty good at high iso so you can bump up the iso to help reach the 1/1000 shutter (I've seen no issues necessitating doubling it to 1/2000), the only issue I've had is with using LED ambient lighting which gives a horrendous flicker at that framerate. However it's not that common a problem and I completely disagree with the advice to only use a slow motion camera on bright days and use the iphone the rest, the phone isn't remotely comparable in speed or high iso.

Based on actual experience, I'll continue to enjoy shooting 1000fps in a variety of circumstances that don't involve bright sunlight.

John
 

dosdan

Active Member
If you had two dimmed LED lamps illuminating the same spot, any minor banding from each lamp, when used by itself, would probably not be perceptible when their illumination was overlapped.

I should add that these lamps should be battery-powered video lamps. Multiple mains-powered domestic LED lights might be synchronised in their flickering. (I'm not sure - I've got none in this house.)

Dan.
 

12harry

Distinguished Member
Using two separately-powered PWM ( ie dimmed ) might "overlap" as dosdan suggests, but suppose they are at 10% say, the black time will be 90% or thereabouts. With two sources the black-time will be something like 80% - I just can't see that PWM-driven LEDs would be a solution to flicker - when using fast frame-rates.
(( In addition PWM is controlled by a timing element, which is probably not a crystal and is likely to drift with supply voltage and ambient temperature . . . not much during any short-burst . . . . but the likelihood of the two exactly switching to minimise any "flicker" is , IMHO most unlikely, as "things" tend to work against one - just when you need it! ))

I don't understand why, with light being an issue in HS filming, any LEDs aren't powered from a steady DC source - ie "ON" 100% ( or whatever the design was ).... with no hint of PWM. Since PWM will ( or a least is Likely-to), cause some flicker/strobe effect on the final media.
For short-duration filming, the LED lights are not likely to be draining the battery significantly, so switch OFF when not needed; would seem to solve any "power" problems....

Incidentally, where mains is used to provide a DC source, while dependent on the complexity of the circuitry, if it's a modern switch-mode PSU ( without the laminated iron transformer, common 15 yrs ago ), then the mains is chopped at a high frequency, any remaining 50Hz ripple is effectively close to zero - the output capacitors are there to provide instant power should the load change be faster than the feedback loop can deliver . . . but IMHO it is likely to be difficult to determine, being rather too quick for available measuring device most folk have. Therefore any "flicker" is likely to be due to circuits in the LED dimmer which, being PWM, is by far the greatest culprit.
For lighting applications there is usually an advantage in having a battery to power the LEDs with the mains PSU acting as "charger" - thus the lights can be cord-free for, maybe one hour, which makes them far more "useful" . . . although IMHO a design-time from fully-charged (overnight) should be minimum 4hrs, to allow for deterioration in the battery-capacity over time.

Direct-on DC ( battery) LEDs won't suffer this, which is what earlier posters suggested should be used....and I agree - Lots of light is usually beneficial in photography matters.
 

The latest video from AVForums

AVForums Movies Podcast: Which is the best decade for horror movies?
Subscribe to our YouTube channel

Full fat HDMI teeshirts

Support AVForums with Patreon

Top Bottom