Best resolution to read detail at speed

[Mark Boyle]

Hello there

My question to the Forum is, What resolution is best to read details at speed for example: Number plates ?

1080p at 60 fps or 1440 at 30 fps ?

Thank you for your opinions

 

[jokiin]

There's no correct answer, different viewing angle, different lens type, different bitrate etc all influence the outcome

 

[Nigel]

If everything else is equal then:
1440 at 30 fps
However everything else is never equal, so you need to say which cameras you are comparing.

If it is the same camera, then almost certainly the 1440 will be best.

The fps has almost nothing to do with it.

 

[Error7]

To capture anything moving at high speed, the higher the frame rate the sharper the image.

 

[Dashmellow]

To capture anything moving at high speed, the higher the frame rate the sharper the image.

There's a difference between frame rate and shutter speed. Frame rate refers to the number of individual frames that comprise each second of video you record (FPS). Shutter speed refers to the amount of time that each individual frame is exposed for.

So, at 30 FPS you could be shooting with a shutter speed of say, 1/30th of a second or you could be shooting at 30 FPS and a shutter speed of 1/5000th of a second. The fast shutter speed will clearly capture a falling drop of water while the slower shutter speed will record a blur, but both will use the same frame rate. The camera will determine how fast a shutter speed to use according to how much light is available.

 

[Error7]

There's a difference between frame rate and shutter speed. Frame rate refers to the number of individual frames that comprise each second of video you record (FPS). Shutter speed refers to the amount of time that each individual frame is exposed for..

My point is that a dash cam recording at 60fps must be using a faster shutter speed than it would need to capture at 30fps.

 

[Dashmellow]

My point is that a dash cam recording at 60fps must be using a faster shutter speed than it would need to capture at 30fps.

Not necessarily.

Remember, this is a thread about "detail-at-speed". Unless the shutter speed is fast enough to stop motion, it doesn't matter whether the slowest available shutter speed at 60 fps is faster than that available at 30 fps.

 

[Nigel]

My point is that a dash cam recording at 60fps must be using a faster shutter speed than it would need to capture at 30fps.

It will almost always use the same shutter speed for both fps.

 

[DAP]

The frame rate puts a hard limit on the maximum exposure time, at 60 frames/second, exposure can't be slower than 1/60 of a second. If it is bright out though, both frame rates will use the same exposure, something faster than 1/60 of a second.
With a short enough exposure, the still frame from both frame rates will look the same, but the 60 frames/second will look sharper than 30 frames per second while playing for anything moving in the frame. For the same camera, a still from 1440p30 will look sharper than 1080p60 (assuming it is bright enough for 1/60 of a second or faster), but the 1080p60 will look sharper in motion.

 

[Nigel]

but the 60 frames/second will look sharper than 30 frames per second while playing for anything moving in the frame.

Can you explain why?

I think it just means that you get more frames, the frames should look identical sharpness as long as there is sufficient bitrate increase for the 60fps, if they are the same bitrate then the 30fps is going to have the better image quality.

 

[Error7]

That's my line of thinking. I'm not in a position to prove it, perhaps somebody else is.

 

[kamkar]

My experiemnts with 2 similar cameras one recording 1080/30 and the other one 1080/60, and the outcome was i was not able in several days of testing to find a single pass ( highway speeds where 2 cars meet doing at least 80 km/h ) where the 60 FPS camer got a plate capture and the 30 FPS one dident ( or the other way around )
So i revised my at the time opinion that really 60 FPS cameras got to be better, and in theory they also should as you at least have 2 X more chances of a capture.
But the thing was if the 60 FPS camera did it, so did the 30 FPS camera, its not so much about FPS but more if the camera can choose to use optimal settings to get that capture, and you can use a 1/60 seconf exposure for 30 FPS footage, but you can not use a 1/30 exposure time for 60 FPS footage.
Dont get me wrong i still think 60 FPS are nice cuz that let you slow down the footage of that idiot to 50% so people on youtube really can see him shine.
I would even take a 120 FPS dashcamm if one could be made so it was just half way decent in low light conditions.

 

[Dashmellow]

My point is that a dash cam recording at 60fps must be using a faster shutter speed than it would need to capture at 30fps.

That's my line of thinking. I'm not in a position to prove it, perhaps somebody else is.

Again, if the shutter speed is not fast enough to stop motion it doesn't matter what the frame rate is. Even if the slowest shutter speed @ 60 fps was 1/60th second and the slowest available shutter speed @ 30 fps was 1/30th second the difference would be negligible for the purpose of capturing "detail-at-speed" because capturing "detail-at-speed" would require much faster shutter speeds at either frame rate.

 

[kamkar]

I would take a 15 FPS dashcam it it was able to maintain a usefull esposure timing of at least 1/250 second, not least in low light.

 

[DashTube]

The world around the camera is moving during the time camera is exposing single frame of the video and that movement makes video blurry. General rule of thumb in still photography is that you need minimum of 1/200th second shutter speed to stop nature-based movement (animals, people, etc.) and make image non-blurry. The faster you can cycle the shutter the better.

Basically, what video camera does is it takes 30 or 60 of these still photographs in a second and stitches them together using lossy compression algorithm to produce one continuous video. So, if all the other parameters stay the same then it doesn't matter if the frame rate is 1 fps or 100 fps, the only thing that counts is shutter speed used in each of the frames.

If car drives at 90 kph (approx. 56 mph) then it is moving at 25 m/s. Divide that by, let’s say 200 (for 1/200th second shutter speed) and you can see that during the exposure of one frame the camera has moved 12,5 cm (or 5 inches). When the target is moving as well, the figures can be doubled. Try moving conventional still camera 5 inches during exposure and see if the result is still sharp? If you were to use 1/30th shutter speed at 90 kph, the camera would move almost one meter during the exposure. As you can see, the exposure time needs to be much shorter than what video frame rate allows for.

But fast shutter speed (short exposure time) can't be achieved if there is not enough light, which is often case when driving. Only the brightest sunny days provide enough light for fast shutter speeds. On low light conditions camera must lower the shutter speed to brighten the image which unavoidably leads to blurry details.

Then there is other variables like pixel count (how many MP in the camera), size of individual photon detectors in the sensor, size of the sensor (surface area), lens F-number, lens coatings, bitrate target given to the algorithm, the algorithm itself, processor speed, etc., etc.

All and all, both 30 and 60 fps cameras can produce sharp details if manufacturer parametrizes them correctly, but 30 fps camera can afford to use less aggressive compression which can preserve more details without exceeding write speed of the memory card.

 

[Nigel]

If car drives at 90 kph (approx. 56 mph) then it is moving at 25 m/s. Divide that by, let’s say 200 (for 1/200th second shutter speed) and you can see that during the exposure of one frame the camera has moved 12,5 cm (or 5 inches). When the target is moving as well, the figures can be doubled. Try moving conventional still camera 5 inches during exposure and see if the result is still sharp? If you were to use 1/30th shutter speed at 90 kph, the camera would move almost one meter during the exposure. As you can see, the exposure time needs to be much shorter than what video frame rate allows for.

A better example is when you are turning out of a side street, making a 90 degree turn.
How long does it take to make the turn? 3 seconds?
How many pixels are there in 90 degrees? About 3000 on my current dash camera.
So during the turn, things are moving across the image at 1000 pixels per second.
So at 60fps, that is 1000/60 = 17 pixels of motion blur per frame (if the shutter speed matches the frame rate).
17 pixels of motion blur is far too much for a good image, if we want to bring it down to a sharp image with 1 pixel of motion blur then the shutter speed needs to be 17 times faster than the frame interval - 1020th of a second, and if the shutter speed is 1/1020 then it doesn't matter if we are using 30fps, 60fps, 120 fps, maybe we could even use Kamkar1's preference of 1024 frames per second, the frame would still look identical.

 

[comadose]

I find the main problem with reading detail is not motion blur, it's dynamic range. Especially at night where you have the headlights reflecting off the number plate it becomes a big mass of bright light.

 

[DashTube]

I find the main problem with reading detail is not motion blur, it's dynamic range. Especially at night where you have the headlights reflecting off the number plate it becomes a big mass of bright light.

That can be solved programmatically by adjusting exposure. Some manufacturers actually advertise their product by claiming it's tuned to maintain number plate readability.