Reading License Plates at Night, Light Reflection / "overexposure". HID / Xenon, standard bulbs

Have you read it properly or understood because the posts were about exposure and frame rate? I referenced that the camera I used has an APS-C sensor to explain why it could use such settings because a dash cam may not be able to due to the sensor size difference. It was not about putting an APS-C sensor in a dash cam, there was a tiny fantasifull reference to this so I don't know why you plucked only this bit out - well actually I do know. I can't be botherered replying to you properly because you are very clearly looking to antagonise people, or more specifically me and I don't have the same amount of time that you seem to have for big online arguments.

I have no particular interest in provoking any big online arguments as you put it, but merely questioned the point of indulging in the impractical and useless folly of comparing a 1/3 inch sensor, firmware exposure controlled, fixed aperture dash cam with a sophisticated camera that has a 24.3 megapixel APS-C sensor, a variable mechanical aperture, actual adjustable shutter speeds and other advanced features. It tells us nothing useful that applies to our actual experience attempting to read license plates with dash cams but does come across as a bit pretentious. Other than that I chose to ignore all the photographic misinformation and incorrect facts in your posts. I do have to say though that when I hear someone like you say, "I find a lot of similarities cross over between photography and dash cams.", it's hard not to laugh out loud. Both are cameras after all and work on the same basic photographic principles, so I am indeed happy to know you've discovered a relationship.

I'll say one thing, my prediction that you tend to respond with petulant remarks was spot on and no doubt we can now expect another round of peevishness from you in reply to this post but I have no interest in having any further engagement with you.
 
Please keep it like that then for everyone's sake. Thanks.
 
It's a little strange that dashcams max out their exposure time at night resulting in motion blur on everything and unreadable plates.
Mtz's post here https://dashcamtalk.com/forum/threa...r-wifi-dvr-is-coming.15351/page-8#post-236131 shows the difference between (I assume) 1/30th and 1/60th second shutter speeds. You can pause and see the light trails being twice as long on the 30fps cams as the Yi at 60fps at night.

If there was a way to limit shutter speed then I think you should have a better chance at seeing plates at night. The longer shutter speeds currently do nothing for close objects as they move too much during the exposure so they aren't giving a useful image.

My newish phone with one of these sony sensors can do HDR stills which takes multiple exposures in a short burst. I wonder if they will do HDR/multiple exposure video soon.

I'd love to see a combined video stream of short and longer exposure HDR.
 
I watched the video and the Yi is definitely sharper at night, check out the big yellow sign at the start! It does look like the exposure is a little darker than the Street Guardian so it's quite possible that the shutter speed is faster. The fps though won't make any difference to sharpness only the smoothness of playback. You'd probably need the shutter speed to be a fair bit quicker for registration plates, I was working out at least the equiv of 1/125th to even start having a chance of reading stationary plates at night at 30mph. Maybe increasing the shutter speed to the point of being able to read plates makes the exposure so bad that the engineers have decided against it because it will make the night quality of the cameras look really bad? That would make sense to me anyway.

Here's a couple of screen grabs at night, the first I'm doing around 50mph and the other car is passing me slowly, perhaps doing 55mph and the ambient light isn't great but not too bad. His plate isn't that far off from being legible and I reckon it probably wouldn't take much of an increase in shutter speed to make it readable.

In the second photo I'm only doing about 15mph and the ambient light is much better. The car in front doing the roughly the same speed is very clear so it is probably using a higher shutter speed but the bus is completely burred, although it is stationary so that's going to have a big impact. I reckon it would take a fair increase in shutter speed to make that bus plate legible.

I think being able to read stationary plates at night at the moment is probably too big an ask, but it would be cool if you could have more manual control over things to suit your own desires or requirements or even for experimenting!


Night.jpg

Night 2.jpg
 
Just a thought, but motion blur - at least in such short time periods - is very linear. Is there sharpening software available that deals with such blurring. I'd be amazed if there wasn't.
It will never be as effective as fast shutter speeds, but I'm sure the results would be impressive. Especially as you know there is dark lettering on a pretty uniform bright background.
I bet there is a plug in for PhotoShop to do it, or maybe the latest version even has it built in.
 
Adobe Premiere has sharpening tools and there's various sharpening tools in Photoshop also but to be honest I doubt either will help much at this level of blur. I could give it a shot though, no harm in it!
 
The fps though won't make any difference to sharpness only the smoothness of playback

The shutter speed is limited by the fps - it can't be slower than 1/30th for 30fps and 1/60th for 60fps. The light trail distance being half the length on the Yi and the same length on the others makes me think they are all using the slowest shutter speed possible at night time.
 
Last edited:
Good point, I'm always looking at going faster! It would seem most likely that they are going as slow as possible to achieve the best exposure at night. Maybe the Yi is increasing the gain slightly to achieve a decent exposure at the faster frame rate?
 
Thats what you often get from the factory to demo night performance, driving slow in a que of cars in a brightly lit town, so everything the camera capture will be nice and crisp.
But IRL thats not what you get, actually you dont have to be much off a sunny cloudless day before you start to run into problems at least regarding capturing plates.

I have noticed sommthing "funny" in lower light conditions what my front cameras capture right near the edge of the frame have a high degree of blur to it, but the same car captured by my mobius in the left window is allmost perfect, and when i and the oncomming car both do 80 km/h then the mobius allso only capture plates at the very edges of its frame, and i have like 3 frames to choose from, 1 where the front of oncomming car enter frame - 1 where front or rear of car is visible in the middle of the frame - and then i allso get the rear of the oncomming car as it exit the FOV of the mobius.

Okay the mobius B lens is at a 90 degree angle compared to the front cameras, but it deal with the same and that is cars rushing by, but since its a side camera it get 2 chances so to say but in a much lesser timeframe, one where the oncomming car is rushing into FOV - and one where it rush out of FOV.

Front cameras have much more time as in the start of the event the oncomming car just rush towards the camera and then slow move to the side of the frame before it rush out of sight / frame.
I would say a front camera ( 30 FPS ) have like 15 - 20 frames in wich to get a good plate capture on a oncomming car, a side camera at 30 FPS have like 3-4 frames at highway speeds.
I will have to put a X cam at 60 FPS setting next to my moobius one day to see what that get.
 
Last edited:
Back
Top