HUD plan backfired

Sabe

Well-Known Member
Joined
Oct 6, 2012
Messages
1,252
Reaction score
565
Country
Finland
Dash Cam
G90, GT300W, Mobius A, C
Since my dvr is not a gps model i try to be clever and use HUD as a speedometer in dvr.
I bought cheapest HUD i could find and this was the result, on video first 30 sec is night and everything is ok, but it was a big surprice to see that at daytime HUD display is brighter and display is out of sync for my dvr, as seen on video past 30 sec mark. To my naked eye display is ok, no flashing or nothing, maybe i could fix it if i adjust video frame rate, 30 fps is evidently not compatible with this HUD.
Maybe it would be better to buy a good quality HUD, sunny day is real problem, it can block everything, also the angle of windscreen can be problem in some car models. All values of HUD can be adjusted in case of you want to show higher/lower speed for some reason...:)
I don't know will i continue to use it, but this was interesting.
Product link is on youtube description, price was 26 euros, now price is higher for some reason.
 
Have you tried special supplied plastic film for windscreen ?
 
Have you tried special supplied plastic film for windscreen ?
Not yet, first i should find the most optimal place for HUD.
 
The shutter speed is slower at night, so it averages out what it sees. E.g. if you are recording at 30fps, the shutter speed could be as slow as 1/30s. This will mimic the persistence of vision that the human eye has.

But in sunlight the camera needs to avoid overexposure. Without an adjustable aperture, the main way to do this is by using a faster shutter speed. Maybe a 1/1000s exposure taken every frame (once every 1/30s.) So if something is flashing rapidly, the camera will only pick it up if it flashes during that 1/1000s. If the flashing happens during the other 999/1000s the camera will miss it. 999 to 1 odds are not good.

Many displays (especially LED based) do not light up continuously. They may light up one segment at a time in sequence. It makes the driver hardware simpler. But human persistence of vision makes it look right. So does a camera with slow shutter speed. But slow speeds will give the strange effect you see.

This is not really about the quality of the HUD, an expensive one might do the same thing. Or a really cheap one might work differently and be seen well by the camera.
 
a faster shutter speed
Wouldn't be more "normal" to integrate the signal for the same period of time (1/30 s) and then, to use some attenuation coefficients (rather than integrate over shorter periods) ? Or maybe there are some saturation issues ?
 
The shutter speed is slower at night, so it averages out what it sees. E.g. if you are recording at 30fps, the shutter speed could be as slow as 1/30s. This will mimic the persistence of vision that the human eye has.

But in sunlight the camera needs to avoid overexposure. Without an adjustable aperture, the main way to do this is by using a faster shutter speed. Maybe a 1/1000s exposure taken every frame (once every 1/30s.) So if something is flashing rapidly, the camera will only pick it up if it flashes during that 1/1000s. If the flashing happens during the other 999/1000s the camera will miss it. 999 to 1 odds are not good.

Many displays (especially LED based) do not light up continuously. They may light up one segment at a time in sequence. It makes the driver hardware simpler. But human persistence of vision makes it look right. So does a camera with slow shutter speed. But slow speeds will give the strange effect you see.

This is not really about the quality of the HUD, an expensive one might do the same thing. Or a really cheap one might work differently and be seen well by the camera.
Thank you very much for that info, somehow i didn't think that this was coming from cameras point of view...i was so locked to that HUD, i think you saved my time a lot:)
That also makes it more challenging to fix i reckon, luckily it was a cheap hud and i can take it like a 26 euro lesson about camera technology.
HUD works just fine, but not the way i was planning to:p
 
Wouldn't be more "normal" to integrate the signal for the same period of time (1/30 s) and then, to use some attenuation coefficients (rather than integrate over shorter periods) ? Or maybe there are some saturation issues ?
That would make for some serious motion blur, don't you think?
 
And here is this effect in real life.
 
And here is this effect in real life.
Perfect; I made a similar test with a LED clock (more light/less light), but I couldn't see any differences.
That would make for some serious motion blur
You're right; ideally is to have a very small time integration, but in low light conditions we have to increase this time, which may cause some blur.
 
Wouldn't be more "normal" to integrate the signal for the same period of time (1/30 s) and then, to use some attenuation coefficients (rather than integrate over shorter periods) ? Or maybe there are some saturation issues ?
It depends what you're aiming for. I always felt that averaging over the entire time period of the frame was better for moving images because it looks smoother and more natural. But if you want to pull out individual frames as stills then you want snapshots in time to get sharp images. Somehow everything went towards fast shutter speeds, and that made camera footage flickery as if being lit by a stroboscope. So that lead to the pursuit of more frames per second to eliminate that.

Some (all?) digital cameras do have an attenuation coefficient in that they can adjust their effective ISO rating, by desensitising the sensor. In digital SLRs this improves image quality up to a point. But I get the impression that in cheaper camera sensors they start at the sweet spot and tinkering with the ISO can make the image worse.
 
...I get the impression that in cheaper camera sensors they start at the sweet spot and tinkering with the ISO can make the image worse.
Not just cheaper cameras. All sensors have an ISO sweet spot, just like all lenses have a sweet spot insofar as aperture is concerned. Most manufacturers don't disclose this because, until you get to extremes, it makes little to no difference in day to day performance for all but the OCD perfectionists.
 
Back
Top