Why can't dash cams have the same video quality as drones?

vbx

Member
Joined
Jun 22, 2013
Messages
240
Reaction score
22
Country
United States
Drones are able to record super crisp and sharp videos from 400ft in the air. Yet our dash cam can't even record a decent video from 10ft.
 
In that situation you can just set the focus to infinity (or hyperfocal distance) and everything will be sharp.
Also the only thing inducing motion blur is angular rotation. Up/down/left/right/back/forward movement or vibration will have zero visible effect at high distances.
And the angular rotation also has a greatly reduced effect when using an ultra wide angle lens.
So basically it's easy mode. I will probably get flamed for saying so, but in these respects, it's true. I'm sure there are other technical issues to overcome but I bet they are mostly external to the camera, e.g. stopping resonance.
 
Drones are able to record super crisp and sharp videos from 400ft in the air. Yet our dash cam can't even record a decent video from 10ft.
I would think.. it is due to the relative speed. A drone flying at 60 MPH 3 feet above a road would likely have a pretty fuzzy picture as well.
 
Plus drones use things like GoPro's
 
In that situation you can just set the focus to infinity (or hyperfocal distance) and everything will be sharp.
Also the only thing inducing motion blur is angular rotation. Up/down/left/right/back/forward movement or vibration will have zero visible effect at high distances.
And the angular rotation also has a greatly reduced effect when using an ultra wide angle lens.
So basically it's easy mode. I will probably get flamed for saying so, but in these respects, it's true. I'm sure there are other technical issues to overcome but I bet they are mostly external to the camera, e.g. stopping resonance.

Really? Cause my cell phone also does a better job at recording video from the same distance as a dashcam. Try it yourself. Use a window mount and do some quick recording. Compare the quality.
 
It's all about Video data rate - the DR650GW with new firmware has a Video data rate of 10Mbs
Compare that to an iPhone 6 at 17Mbs at 30 frames/sec or 27Mbps at 60 frames/sec
Or a GoPro will start at 45Mbps and go up to 60Mbps depending on settings

On the flip side, a Gopro fills a 64Gb card in 2hours 13 mins at 60Mbps and just under 3 hours at 45 Mbps while the Blackvue will fill a 64Gb card in around 12-16 hours. My daily commute would easily cause a 64Gb cad to be overwritten if the datarate was 45Mbps so that's not much use.

The Blackview is capable of faster datarates (checkout the Russian firmware that runs on older versions) which would give much better video resolution/quality however Blackview have decided that 10Mpbs is the optimal compromise for reliability and use ..
 
Drone shots from action cams aren't as sharp as you think.

Look at this picture I found on google, it looks razor sharp right?

vBzHgCm.jpg


Now get close tot he screen and look close up at the catamaran in the centre of the picture. It's actually quite blurry close up. The people on deck are just splodges of colour, you can barely make out the ropes along the edge of the deck. Take a look also at the RIB in front of it. Standing back from the picture also sharp. Close up, there's no visible detail. A white splodge that's the driver and passenger, a red splodge that's a kid and you can barely make out the grab ropes strung along the side of the tube. The boat itself is a pretty uniform white shape. The centre of the picture looks the sharpest by the beach. But again look closely at the palm trees, you can't see a single individual leaf. The sailing dingies on the shore and the boats close up are just splodges of colour. I'd suggest this is a good example of how Depth of Field gives a picture an overall sharp appearance when in actual fact a lot of detail is lost beyond the focus point.

Read the discussion about DOF we had in this thread where I posted a picture from Google showing a dash cam type shot using what's specified as a GoPro : https://dashcamtalk.com/forum/threads/hd-cameras.21864/#post-288146

Anyway factors for un-readable number plates (general factors and no reference to any cam in particular):

1. Fixed focus lens & DOF - many car cams and action cams used fixed focus lenses, usually focused to a very short distance from the camera, often a few inches, because by making the foreground ultra sharp and using Depth of Field to give apparent sharpness to the background, you can trick the mind into believing the whole picture is really sharp. In reality, the object at the focus point is sharp, and everything else is rendered sharp by the DOF effect by the camera using a small aperture. Whilst DOF gives an overall sharp appearance to the picture, fine detail is usually actually unsharp. Hence why in the picture in the other thread, the picture overall looks quite sharp but the number plate of the 2nd car on the right is unreadable even from about quite a reasonable distance. In fact so much detail is lost that you can't even really make out any lettering. The only way I know to overcome this is - 1. Change the fixed focus to a point further out - you'll still have exactly the same issue, but by changing the focus point to say 10 feet, you can ensure that number plates at 10 feet are clear and then drop off in sharpness from there as they get nearer or further away. 2. Use a video camcorder with autofocus and hope it will lock onto the car coming towards you.

2. Data rate as Ian says. The more information (detail) in the picture, the harder the camera has to compress it to fit it into the size of video file that the camera is locked into producing. eg. Video a sentence on an otherwise blank static piece of paper, and the compression rate will be very low (unless the paper has visible texture) because there's very little information on the screen and very little changing from frame to frame. By contrast, drive down a tree lined street, and each tree has millions of leaves, plus the motion means large amounts of data are changing from frame to frame so the amount of data the camera can discard as carried over from the previous frame is very low. The result is in order to try to record the picture as seen, the camera has to compress the data very hard to make it fit the file size, and compression involves discarding detail. That's why an in car video with a tree lined road typically shows a lot of blur in the trees and / or macro blocking (the appearance of squares). As Ian says, the answer here is to either use a more efficient compression codec such as H.265 (which compresses without as much loss of detail), or to up the data rate. A higher date rate = the camera compresses less and so discards less detail with the result, the picture is sharper, contains more details and less artefacts. There is a point of course where there is little to no return. However, this should be quite high as many professional video cameras use data rates around 100mbs. I think it's fairly safe to assume the broadcast video industry have done their research.

Personally, I disagree with Ian a little here about the significance of overwriting the data with a high data rate cam. Most car cams automatically or manually by pressing a button, allow you to instantly protect the footage at / after an impact is detected. Secondly, even those that don't such as action cams, will still allow you to protect individual files manually. So after an accident, you can simply stop the camera when you're finished at the scene and protect the last couple of video files. If the camera won't do that, simply carry a 2nd memory card and swap them over after the accident flicking the overwrite protection switch on the side of the one you take out (stops you accidentally putting the wrong one back). I use a GoPro 2 currently in my car since dropping my Mobius onto it's lens, and on my commute which is about 1/2 hr each way, my 32GB SD card lasts about a week before I have to reset it (the GoPro 2 doesn't have loop recording), and that also includes non commuting journey's as well.

3. Lens / sensor / chip quality - no matter how good the data rate, focus, etc. ultimately a camera can only record the quality that's fed to it by the lens. The better the lens, the better the picture. Hence why a consumer lens on a digital SLR camera cannot come close in picture quality to the same picture on the same SLR taken through a professional lens. Beyond that, the sensor and processor also play a part in overall quality.

Personally, from what I'm seeing atm, the Yi 4K is the best quality action / dash camera out there in my opinion, if you're prepared to use an action cam in a car, although the Eken 8 pro seems to run it very close for a fraction of the price. I've only viewed online videos though so the actual reality could be different. All I can say, is that if I could afford it, I'd buy a Yi 4K atm myself, as I value quality over dash cam convenience and don't believe in GPS recording (why hang yourself if you accidentally go over the limit by having your speed recorded!).

That said, no fixed focus camera is going to give readable plates at distance if it's focus is fixed to a few inches. If you want plates readable at any distance, then you're going to need a camcorder / or a good video SLR such as Canon 70D, put up with the size / hassle of removal of a large camera and hope that the autofocus locks onto approaching vehicles and not simply down the road.
 
Thanks for putting that together. Very informative.
 
The number one quality killer is the windscreen. Its the poorest piece of glass you could add into an optical train.

What makes it worse is that windscreens are at an angle and curved, meaning poor refractive index and aberrations and distortions.
 
and don't believe in GPS recording (why hang yourself if you accidentally go over the limit by having your speed recorded!).

As an aside, law enforcement agencies/ insurance companies etc could easily determine your speed from how fast you move between markings on/beside the road, so not displaying GPS data doesn't help.
 
As an aside, law enforcement agencies/ insurance companies etc could easily determine your speed from how fast you move between markings on/beside the road, so not displaying GPS data doesn't help.

Unlikely to be accepted as evidence due to parallex / lens errors, especially with a wide angle lens. However, there's a difference between a video where you're 35 or 40 in a 30 and 70 in a 30. In the former it may be impossible to tell with pure video. With the latter, it will be apparent you're way over the speed limit.

The issue I have with GPS is you could have an accident, you may have been at eg 34 in a 30, and so can't use the video evidence, or if you do or the other party can obtain it, can have the evidence used to prove you at fault in a civil claim, whereas the same video without GPS would never show the slight speed error. The flip side to GPS of course is if you're unfairly accused of speeding, it can help prove your innocence so it's a matter of balance. Personally I'd rather take my chance with pure video.
 
Back
Top