VIOFO A119 V3

[vvs49]

What firmware does the camera have installed?

My
St_MOD_9G_FWA_2.3_20210125

 

[relu]

I would like to try it too, can you give it to me please?

 

[Dashmellow]

Over 100$ dashcam has the same issue. A129+.

I've been using dash cams for 10 years now. In the "old days" dash cams could provide a decent 1080p image during the day but they were literally useless at night. You might see a couple of yards ahead of your vehicle with bright headlights on a dark road. There would be a LOT of noise and little else.

The first sensor that really made reasonably priced dash cams viable in low light was the Aptina AR0330 combined with the Novatek NT96650 processor. It could be noisy but you could actually see the nighttime scene clearly. Little by little we've witnessed remarkable improvements in sensor technology and now we have cameras like the Viofo A119 V3 that can perform amazingly well in low light. There still may be noise and motion blur but you can actually SEE the scene in very low light.

It is important to understand that dash cams use remarkably small sensors coupled with fixed aperture M12 lenses. These small sensors have limitations. There is a finite limit to how much light these small sensors can absorb, especially as pixel size becomes smaller and smaller for higher resolution. One day, the average dash cam may use much larger sensors, perhaps with larger pixel sizes as well and when coupled with appropriate lenses we will have much better low light performance with far less noise and motion blur.

Even at 200 dollars or more, dash cams are really just consumer gadgets built to certain price points more than "real" cameras built to higher standards and precision. Many, many dash cams have various problems and flaws and I hope to see that improve in the industry as time goes on. Still, today's dash cams offer performance we could only have dreamed of 10 years ago and I'm happy to have them.

Some here have probably seen the video I posted a few months ago when an owl smashed into my windshield on a pitch dark road at night but for those who haven't seen it here is a screen shot.

If not for the A119 V3 I would not have known what hit me. In person, all I really saw was a few checkered feathers for just a moment at the very top of my windscreen but it was way too dark to really see what happened.

I came away amazed at how well the camera performed in the dark! Not only did it capture what actually happened but when I reviewed the video I could clearly identify the species as a barn owl. So, I have no complaints about low light sensor noise or motion blur with the V3. This is probably the cleanest and most noise free, dark nighttime footage I've yet seen from a dash cam.

I do keep the camera set for 30 fps at all times and this seems to provide me with the overall best performance, least noise and it mostly eliminated the banding I was experiencing in low light. I live in a rural area with no streetlights and to have this kind of performance on a completely unlit dark road is a remarkable leap from what I've experienced with dash cams even only a few years ago.

 

[vvs49]

OK!!

St_MOD_9G_FWA_2.3_20210125.zip

drive.google.com

 

[relu]

OK!!

St_MOD_9G_FWA_2.3_20210125.zip

drive.google.com

Thanks!

 

[jokiin]

Maybe both cameras could be improved in terms of noise, but doing so would probably have a negative effect on motion blur so the change may not be desirable.

I'm sure they could, as you would also know if they do that there's going to be a tradeoff, much less detail overall in low light situations for starters, gathering information about the scene you are in is far more important than trying to make cinema quality movies that some people seem to expect

 

[Nigel]

I see that you speak with figures, percents... in your great wisdom can you tell us the tolerance limit of banding, maybe you know better than us some secret specifications of Viofo's dashcams.

I would prefer there to be no banding, just random noise, even better no noise at all, but it is a compromise, if you reduce noise then you make other things worse.

The image in question had a noise level of about 1 digit, the minimum possible, would have been impossible to see if it hadn't been in bands, so I wouldn't say that there is any fault there.

 

[Cristianson]

Over 100$ dashcam has the same issue. A129+.

Exactly the same with mine on roads without light.

The thing here is that only some have these problems and some dont and why this occurs with newer firmwares but with older no?

 

[ursamajor]

I would prefer there to be no banding, just random noise, even better no noise at all, but it is a compromise, if you reduce noise then you make other things worse.

I see, random noise in bands. Lol. Dude, do you really think that we are blind, or we don't know what we see?
Maybe you can see the difference between random noise and banding@noise in the attached picture. It's a collage with two pics taken at seconds between them (see index time).
I'm glad that you have the support of a manufacturer and a watchdog. They prove once again that they don't care what customers say.
But I hope that those smart guys from Viofo listen the complaints about this phenomenon that I already sent to them. So far they have been receptive to all the suggestions I have made.

 

[Kremmen]

As I posted ages ago, I'd love to get one that has this problem, pop in my firmware version, my SDcard, my settings, and pop it into my mount and see what happens.

That would hopefully pin it down.

 

[Nigel]

Maybe you can see the difference between random noise and banding@noise in the attached picture. It's a collage with two pics taken at seconds between them (see index time).

What I see in that is that the camera has adjusted the exposure between the two images, maybe because the second is on a corner so the headlights are hitting the side of the road thus brighter than when on the straight where they are illuminating the distance.

In the top image, the exposure has pushed the sensor to its sensitivity limit, maybe a bit beyond if you don't like that level of noise; on the second image the sensitivity (ISO) has been reduced so there is very little visible noise.

The two images show the compromise quite nicely, in the top there is little motion blur but a lot of noise, in the bottom there is little noise but more motion blur. Which is best? From the point of view of recording evidence, I think I would choose the top, if I wanted to make a movie then I would choose the bottom, but neither is the correct choice, just different.

I do think that there is a bit too much noise in the top one, and the banding does not look good, but it is still recording evidence, doing its job; and the settings it is using may be really good settings in another camera. The sensors are analogue components, they do all have noise, and the amount of noise will vary from sensor to sensor and from camera to camera, even car to car since the noise may not be coming from the sensor itself, could be RF interference from some other device in the car such as a mobile phone transmitting data.

I am wondering why those images look so dark, almost looks like you have tinted your headlights?

Exactly the same with mine on roads without light.

Can you post a "roads without light" image for us to look at?

Feedback is useful, and Viofo do listen...

The thing here is that only some have these problems and some dont

As above, image sensors are analogue and there will always be some variation between samples when you push them to their limits, there are also a few other components that can vary such as power supply components that may be generating different noise levels.

We are pushing the sensors to their limits "on roads without light", they are not designed to see in darkness.

and why this occurs with newer firmwares but with older no?

I assume there was an adjustment to the exposure settings to reduce motion blur a bit at the expense of a little more noise. Maybe for people with bright headlamps this works really well, for people using paraffin lamps on their steam cars it is probably a bad change! You can't always please everyone.

 

[Cristianson]

Excuse my french! You dont have to keep pointing "roads without light", english is not my native language.
I just wanted to let others know that the this problem some of us experienced in 2.0 is not fixed in 2.3.

Maybe for people with bright headlamps this works really well, for people using paraffin lamps on their steam cars it is probably a bad change!

I assure you audi matrix headlights dont work on paraffin yet the problem still exists so this "maybe" is not the answer.

 

[SawMaster]

Maybe for people with bright headlamps this works really well, for people using paraffin lamps on their steam cars it is probably a bad change! You can't always please everyone.

Hey- quit making fun of my old van At least it has a blinker fluid reservoir which some new BMW's and Audi's seem to lack

But seriously there is quite a difference in car lighting and sensors. With my old-tech halogen sealed beam headlights the best low-speed night-time performance I ever got was with the old OV4689 sensor, yet that same sensor seems to get overwhelmed with brighter headlights and also when approached by oncoming cars.

No "perfect" for everyone and for every situation, only what does well (or doesn't) with specific situations, or when spoken of in only the most general way.

Phil

 

[Cristianson]

gathering information about the scene you are in is far more important than trying to make cinema quality movies that some people seem to expect

I dont expect nothing more than what was already there before. In v1.3 firmaware the picture is clear day and night without any kind of banding in the same car, same power adaptor, same everything, only firmware has changed("upgrade" ha ha).
Then why cant i have 21:9 with the same clear image?

 

[Nigel]

I assure you audi matrix headlights dont work on paraffin

Had to look those up:

Its current state-of-the-art lighting, HD Matrix LED, is an adaptive system that works by splitting the traditional single headlight beam into dozens of smaller rays that can be turned on and off individually to un-light small areas in front. It uses 32 diodes per side, controlled by a forward-facing camera, and it's largely effective. The newer Digital Matrix LED units use a reflective chip about the size of fingernail that contains a million microscopic mirrors measuring a few hundredths of a millimetre each, meaning a much more accurate and pixelated beam can be generated compared with the old system's 32 big blocks of light. Visually it's the difference between playing Minecraft and the latest Call of Duty title.

And it can light up just your lane:

So how does that interact with the camera? Could the clever mirrored LEDs be causing the banding?

Excuse my french! You dont have to keep pointing "roads without light", english is not my native language.

Sorry, I did not see anything wrong with your English, I was just referencing your words.

 

[ursamajor]

What I see in that is that the camera has adjusted the exposure between the two images, maybe because the second is on a corner so the headlights are hitting the side of the road thus brighter than when on the straight where they are illuminating the distance.

In the top image, the exposure has pushed the sensor to its sensitivity limit, maybe a bit beyond if you don't like that level of noise; on the second image the sensitivity (ISO) has been reduced so there is very little visible noise.

The two images show the compromise quite nicely, in the top there is little motion blur but a lot of noise, in the bottom there is little noise but more motion blur. Which is best? From the point of view of recording evidence, I think I would choose the top, if I wanted to make a movie then I would choose the bottom, but neither is the correct choice, just different.

Oh, boy! Here we go again. I don't know if you really don't or you don't want to understand or deliberately hijack the discussions: we talk about banding not about noise. The pattern in that (chroma) noise (in my pics) is banding, not random. So, repeat yourself: banding!
Try to not denaturate the discussions, please. I know very well how a camera works. No need your explainations.

 

[Cristianson]

Could the clever mirrored LEDs be causing the banding?

Dont think so, i doubt everyone with this problem has led headlights and i feel the need to repeat myself, same car, same headlights, new firmware = banding/stripes or what they are called.
I appreciate your intent to help but clearly the problem is in the camera software since V2.0.

 

[jokiin]

So how does that interact with the camera? Could the clever mirrored LEDs be causing the banding?

LED headlights in general can cause flickering in the picture due to the refresh rate

 

[Dashmellow]

LED headlights in general can cause flickering in the picture due to the refresh rate

In theory, that problem could be eliminated if they adjust the PWM (pulse width modulation) properly when they design the headlights (or tailights) but often that seems to get overlooked because how these lights appear in digital video is not a consideration.

 

[jokiin]

In theory, that problem could be eliminated if they adjust the PWM (pulse width modulation) properly when they design the headlights (or tailights) but often that seems to get overlooked because how these lights appear in digital video is not a consideration.

probably could be resolved, as you say though it's likely something that they would never give any consideration, just something we need to live with, for now at least, unless there's some other change in technology