1920x800/540 video mode?

[nutsey]

What do you think about it? Is this no-hood-and-no-sky mode in demand by anybody or not? It could add ~25% of bitrate efficiency with no vital image detail loss...

RAW video link

 

[Nigel]

I prefer 1920x540, then you can watch both front and rear in full resolution simultaneously on a standard FHD monitor. I've never found that I'm missing anything at that 32:9 aspect ratio, and you gain even more bits per pixel, in fact double the quality for the same bitrate over conventional video.

 

[nutsey]

I prefer 1920x540, then you can watch both front and rear in full resolution simultaneously on a standard FHD monitor. I've never found that I'm missing anything at that 32:9 aspect ratio, and you gain even more bits per pixel, in fact double the quality for the same bitrate over conventional video.

Copy that!

 

[Nigel]

It also makes it easier for the auto exposure/HDR if you don't have a bright white or dark black hood occupying 1/3rd of the image and a bright sky in another 1/3rd.

Do you have a 1920x800 camera?

 

[nutsey]

Sure. Any Novatek 650-663 cam can shoot in this video mode.

 

[Dashmellow]

What do you think about it? Is this no-hood-and-no-sky mode in demand by anybody or not? It could add ~25% of bitrate efficiency with no vital image detail loss...

The problem with this approach is that there is no indication of the distance between your vehicle and any other car or object out ahead of you. If you are involved in a traffic accident or some other mishap with another vehicle or perhaps with a pedestrian you will be better off, evidentially speaking if you can show the actual impact and nearby circumstances along with the direct consequences. Even capturing a near miss with your vehicle and another could prove vital but would unlikely be captured without your hood/bonnet in the image. Proper souveillance benefits from a point of reference.

 

[kevindd992002]

The problem with this approach is that there is no indication of the distance between your vehicle and any other car or object out ahead of you. If you are involved in a traffic accident or some other mishap with another vehicle or perhaps with a pedestrian you will be better off, evidentially speaking if you can show the actual impact and nearby circumstances along with the direct consequences. Even capturing a near miss with your vehicle and another could prove vital but would unlikely be captured without your hood/bonnet in the image. Proper souveillance benefits from a point of reference.

So which resolutiom do you recommend?

Sent from my SM-G950F using Tapatalk

 

[TonyM]

I agree with @Dashmellow about keeping a point of reference within the frame. Also, when you stopped at the lights around 31s, the red light is cropped off the top of the video. Information like this can be vital to showing who was at fault in an incident. So whilst I see the benefits of eliminating the hood & sky from the exposure calculations, and reserving the bitrate for the most important parts of the view in the middle, there are times when having the top and/or bottom of a conventional 16:9 view are also useful.

EDIT: I have no problem with eliminating the top and bottom strips from the exposure calculation, so that the exposure is based on the road & cars without being affected by the sky and the hood.

 

[Dashmellow]

So which resolutiom do you recommend?

Sent from my SM-G950F using Tapatalk

As far as this discussion is concerned it really doesn't matter I guess. My point is that it's good to have at least a small section of your vehicle in the image as a point of reference between your car and the traffic environment you are driving in or to capture the impact of your vehicle with something else.

To answer your question though, I typically use 1920 x 1080 because it usually does an excellent job for most things and it is the most universal format that anyone you need to hand your video off to such as an insurance company, attorney or law enforcement officer will be able to easily view with any device or screen or software they have access to.

 

[Dashmellow]

@TonyM makes a good point about traffic lights.

BTW, I've tried out these "widescreen" formats and I can certainly see the appeal but I abandoned using them for the reasons outlined above.

 

[Nigel]

Proper souveillance benefits from a point of reference.

Curious - what was the intended spelling of that second word? Or is it a USA spelling for something?

Sousveillance - Wikipedia

en.wikipedia.org

My point is that it's good to have at least a small section of your vehicle in the image as a point of reference between your car and the traffic environment you are driving in or to capture the impact of your vehicle with something else.

Depends on the vehicle, even with 32:9 I still have the very front of my car in the image, and our traffic lights seem to be positioned lower than yours so I also have them in view.

 

[Dashmellow]

Curious - what was the intended spelling of that second word? Or is it a USA spelling for something?

Sousveillance - Wikipedia

en.wikipedia.org

Depends on the vehicle, even with 32:9 I still have the very front of my car in the image, and our traffic lights seem to be positioned lower than yours so I also have them in view.

My bad, I misspelled sousveillance and apparently my spellchecker doesn't recognize the word.

You are quite right, I guess it would indeed depend on the particular vehicle (or how you aim your camera) but every image you've ever posted using such formats, especially the one you've been posting a lot lately does NOT have "the very front of your car" in the image. Is this a cropped, post processed image? Kindly show us what you are talking about.

Also, time and time again you make assertions and broad generalizations about how things are in the USA. Fact is that the height (and style) of traffic signals can vary widely depending upon where you go, even at times within the same city.

 

[Nigel]

You are quite right, I guess it would indeed depend on the particular vehicle (or how you aim your camera) but every image you've ever posted using such formats, especially the one you've been posting a lot lately does NOT have "the very front of your car" in the image. Is this a cropped, post processed image? Kindly show us what you are talking about.

Front of the car is hidden behind the rear camera in that view, here is one I posted a while ago with only front and back. Although front is not 32:9 in this case, there is plenty of height for even the tallest of traffic lights:

Also, time and time again you make assertions and broad generalizations about how things are in the USA. Fact is that the height (and style) of traffic signals can vary widely depending upon where you go, even at times within the same city.

Not living there, maybe it is not surprising that I wouldn't know in great detail about things like "the height (and style) of traffic signals can vary widely depending upon where you go, even at times..."! UK traffic lights have a standard set of rules for positioning.

 

[SawMaster]

I agree on needing a car reference point- in fact I prefer seeing both front corners as that's where most forward-facing crashes will make impact. And in capturing traffic signals. Otherwise there is merit in not wasting data space on unnecessary parts of the view if that can be used to raise performance elsewhere.

Phil

 

[Dashmellow]

Front of the car is hidden behind the rear camera in that view, here is one I posted a while ago with only front and back. Although front is not 32:9 in this case, there is plenty of height for even the tallest of traffic lights:

Well, having the front of your car in the image like that is an improvement in that it establishes a reference for your vehicle, but just barely. With footage like that I would hate to have a vehicle strike me on one of my fenders from either side just outside the field of view, as @SawMaster is talking about. My hunch is that like most members here, you have never been in a situation where you have even been required to provide footage to the authorities, and like many here on DCT your concepts of dash cam capture during an incident are theoretical. Of course, if that is your preference that is what you should go with.

Not living there, maybe it is not surprising that I wouldn't know in great detail about things like "the height (and style) of traffic signals can vary widely depending upon where you go, even at times..."! UK traffic lights have a standard set of rules for positioning.

So, then kindly don't keep making assumptions and blanket assertions about how you believe things to be here in the US. FYI, we have standards here as well. We have different standards and codes depending upon whether you are on a federal highway or a state highway. In some circumstances the standards are set by an individual smaller jurisdiction, such as a town of a certain size or a city of a certain size. Unlike the UK, the US is a huge nation of widely different entities in widely different environments. All in all, traffic engineers are responsible for making these decisions and establishing the standards for each jurisdiction along with state and local governments. In my local small town alone we have everything from enormous traffic light arrays controlled by cameras monitoring the flow of cars at major intersections to smaller traditional traffic signals on ornate cast iron posts at a more "human" scale only blocks away. This is on a state highway that passes through the middle of town where it transforms into "Main Street" until you eventually leave the other end of the town.

 

[crabu2]

I need the hood, at least some of it as a reference point. The sky can go.

 

[Michael_K]

That's my favorite format! (Well, close enough to 64:27.)
I wonder if it's just a slightly low FOV. I don't see any fisheye (or very little).
Yes, I'd also like some of the hood. But, with 16:9, you get dashboard, too!
Maybe the OP just has a car with a very short front bit?

Perhaps, with a slightly wider FOV, this would work.

 

[Nigel]

Well, having the front of your car in the image like that is an improvement in that it establishes a reference for your vehicle, but just barely. With footage like that I would hate to have a vehicle strike me on one of my fenders from either side just outside the field of view, as @SawMaster is talking about. My hunch is that like most members here, you have never been in a situation where you have even been required to provide footage to the authorities, and like many here on DCT your concepts of dash cam capture during an incident are theoretical. Of course, if that is your preference that is what you should go with.

Well then we need a 360 degree view so that the fenders are properly visible, although it is still hard to see any contact unless the camera lens is outside the cabin, or you have a very rectangular car!

One I posted a couple of weeks ago, not much of the hood visible in this one, but enough to be a reference, and since my car has curved edges rather than 90 degree corners it is always going to be impossible to see where the real front of the car is unless I put a camera in the radiator grill - now there is an idea, I have a waterproof remote camera waiting for a use...

 

[SawMaster]

This seems to be about one's approach toward whatever they consider optimal usage. For me, a dashcam is an evidence gathering device so I want it to clearly show where my vehicle is as well as what is happening nearby around it. I do not need to have images from a quarter mile away nor do the images need to be perfect in color or focus; they just need to be present and identifiable and cover as much area as possible while doing this. Those images also need to capture the conditions of traffic control devices where possible. I really don't care if it shows the dashboard or sky or the hairs in my nose- none of that is directly relevant to my purposes.

Where this subject does become relevant to me is if the excess image area or coverage has any adverse affect on the performance of my cams. Which we know that excessive sky in the FOV can do, and also dashboard reflections. And perhaps even in processor performance which is a hard limit we're working within. So if in losing some of the unneeded capture area (such as my nose hairs) lets my cam perform better that's a good thing. But if it detracts one bit from my being able to see and view the things I deem necessary to it's function as an evidence-gatherer then it's a bad thing. I'm not recording for cinematic purposes nor am I concerned about the recording's aspect ratio for viewing pleasure purposes.

You must clearly have visible some reference point(s) to your vehicle in order to have important factors like your lane position, apparent distances. and traffic signal conditions clearly visible in the recording or it's not going to show what needs to be seen to assist you in a traffic incident. Having the front corners of your car visible is by far the best way to clearly show your lateral lane positioning regards road markings. Having only a centerline reference from a center-mounted cam does not do this; but only shows your relation to the lane center which is not important in terms of legality. If for whatever reason you can't capture your front corners they you have an even greater need to get some kind of side references in the view to better show you were within your lane limits. My minivan has a sharply sloped front so I had to aim the cam downward more than I preferred to in order to capture those critical reference points. It was OK as I could still capture all overhead traffic signals, but just barely, and barely is enough here. Luckily dashboard reflections weren't a problem in doing this. In my big workvan I can aim higher and get a better capture of everything so I do.

Long before dashcams I was involved in a crash where a woman had wildly crossed the centerline in a sharp curve then was attempting to get back over when the impact occurred. Momentum of her vehicle carried her back over into her lane post-impact so there was no clear evidence by vehicle positioning to prove her claim or mine regards who had crossed the line- it could have been either of us. What saved me in that case were my skidmarks, which started in my lane then jolted over at the impact point at the edge of my lane. It was raining and my tracks were almost washed out when the Cop arrived; had he been a couple minutes later in arriving I would have had no proof of my story versus her lies. If I had a cam in that crash it would have showed her wildly across the centerline and the point where impact occurred, but without having reference to the corners of my car I would still not been able to prove that at impact I was fully in my lane; the sharpness of the curve would have made a centerline reference only useless for evidence purposes.

It was a crazy crash and would have made an equal to some of the Russian vids as she was not only in my lane, but at one point was almost completely in the lane beyond that with her car well heeled over from cornering forces as she flew along at a reckless speed. As we all sat in the Cop's car while he was writing it up she claimed I was flying and that she never left her lane. The Cop, rather miffed, said "Lady, I saw his skidmarks and there's no way you weren't in his lane. It doesn't matter how fast he was going, you weren't supposed to be in his lane so it's your fault completely" She opened her mouth to say something and the Cop jumped in and said angrily "I don't want to hear another word from you. This was your fault and you're going to be charged with it." It was all I could do not to laugh out loud at that point but I managed to stay composed. This crash totally convinced me of the huge importance of being able to show the front corners of my car when I later got into dashcamming. It doesn't matter at all if you see my dash (or my nose hairs) but it matters greatly that you can clearly see me within the limits of my lane in all circumstances. Here in the Southern US we have some quaint words and phrases and one of them is "wrong-headed". I think it's meaning is obvious so don't be wrong-headed in thinking about what your cam captures and it's importance

Phil

 

[TonyM]

UK traffic lights have a standard set of rules for positioning.

It's not that straightforward. I've just had a quick look through the "Traffic Signs Regulations and General Directions 2016", the DfT "Traffic Signs Manual", and the "Specification for Highway Works - Series 1200". It was no surprise that I did not find a simple diagram showing how high above or to the side of the carriageway each sign or signal should be placed.

I would agree that for most of the time, the central 32:9 shows pretty much everything you might need. But there are occasions when having coverage of birds in the sky, low tree branches, tall-sided vehicles next to you, overhead signs & signals and so on may be beneficial in recording the circumstances surrounding an incident.