Mobius 2 Action/Dash Cam support thread

I have a feeling night improvements may prove impossible. My understanding of cameras is that usually the night performance is a measure of how sensitive the sensor is at low light levels, the noise usually occurs because the camera has to up the gain (ISO in stills language) to compensate. I'm not aware of anyway to improve that as the sensor needs the light that the sensor needs. I'm afraid it could simply be a bad sensor choice by Mobius.

It's a pity no-one in the dash cam world can get hold of a sensor comparative to that fitted in the top stills / video cameras (many of the the same sensors wouldn't fit in any event due to size), but if someone could get a hold of a cheaper smaller version, then the performance is now amazing.

A cheaper cam with great low light performance is the Sony UMC-S3c (£300 approx):


A more expensive and definately higher quality picture option in my opinion, is the Sony Alpha7S. Here it is in action, video takes longer to get going! Great video picture up to ISO 208,000 (!) in my opinion, ISO 40,000 is amazing quality, and it's probably usable at 500,000 ISO although in a car cam probably too much highlight would be lost. However a similar sensor with an auto range of ISO 50-40,000 (or gain equivalent could be a great sensor (assuming daylight performance is still good):

 
Last edited:
@TonyM and @richcam, can both of you kindly report on whether your tests were done with H.264 or H.265?

I set both the M1 and M2 to record in H.264 due to the problems I had with replaying the video footage when recorded in H.265 using VLC on my iMac.
 
I have a feeling night improvements may prove impossible. My understanding of cameras is that usually the night performance is a measure of how sensitive the sensor is at low light levels, the noise usually occurs because the camera has to up the gain (ISO in stills language) to compensate. I'm not aware of anyway to improve that as the sensor needs the light that the sensor needs. I'm afraid it could simply be a bad sensor choice by Mobius.

It's a pity no-one in the dash cam world can get hold of a sensor comparative to that fitted in the top stills / video cameras (many of the the same sensors wouldn't fit in any event due to size), but if someone could get a hold of a cheaper smaller version, then the performance is now amazing.

A cheaper cam with great low light performance is the Sony UMC-S3c (£300 approx):


A more expensive and definately higher quality picture option in my opinion, is the Sony Alpha7S. Here it is in action, video takes longer to get going! Great video picture up to ISO 208,000 (!) in my opinion, ISO 40,000 is amazing quality, and it's probably usable at 500,000 ISO although in a car cam probably too much highlight would be lost. However a similar sensor with an auto range of ISO 50-40,000 (or gain equivalent could be a great sensor (assuming daylight performance is still good):


There is quite a lot more to night time camera performance than "how sensitive the sensor is at low light levels" and the idea that "someone could get a hold of a cheaper smaller version" of the full frame sensors in any of the three mentioned cameras seems wishful thinking more than anything else.

Firstly, the full frame 35mm size of the sensor is one of the primary reasons for the excellent low light performance. If you were to make a smaller (and cheaper) version of the sensor it could never perform as well.

Secondly, in order to take advantage of the performance of these highly sensitive sensors one needs a much larger, faster, higher quality and far more expensive lens than any that you will ever likely see in a dash cam.

Thirdly, the outstanding low light performance you see in these cameras is in large part due to the capabilities of the powerful and very expensive processors used in conjunction with these advanced larger format sensors. Sony sometimes refers to their DSPs as the "image processing engine".

The Sony SNC VB770 is a camera that costs $7000 (NOT including the lens). The Sony Alpha7S is nearly $2000 NOT including the lens.

The Sony UMC-S3c is probably the closest in size to a camera that might be suitable as a dash cam at 104mm x 84mm x 48mm but that does not include the SEL35F14Z Distagon wide angle E-mount lens Sony offers for this camera which will protrude another 4.5 inches, has a 72mm front filter size and adds another 1500 dollars to the more than $5000 (£3869.97) cost of the camera itself. (not £300 as was stated above)

I'm not quite sure I understand the logic of suggesting that the currently achieved M2 night time performance was perhaps "a bad sensor choice by Mobius" and then offering up such an extravagant and essentially unrelated set of comparisons. After all, we are talking about a tiny $90.00 camera here.

In any event, I do believe we will eventually be seeing improvements to the performance of the M2 across the board, including low light as time goes on, much as the Mobius team managed to squeeze every last drop of performance out of the M1.
 
Last edited:
If would perhaps be more appropriate to compare the M2 (or any other dashcam) to a small point and shoot camera. I do find it interesting that, to my knowledge, none of the big names in cameras make a dashcam (Canon, Nikon, Sony, Panasonic etc.) Surely there's a market out there for them to tap into? The nearest we get is the use of Sony's sensors. Anyway this is a topic for another forum and o/t here.
 
There is quite a lot more to night time camera performance than "how sensitive the sensor is at low light levels" and the idea that "someone could get a hold of a cheaper smaller version" of the full frame sensors in any of the three mentioned cameras seems wishful thinking more than anything else.

Dashmallow, don't get me wrong, I'm not saying that Mobius necessarily got the sensor choice wrong by not choosing one of the above sensors, what I said was that it's a pity no-one in the dash cam world can get a hold of a sensor of such capability.

You can also see from the recommended list of dashcams on here, the most expensive camera doesn't necessarily have the best low light performance. Even at low to medium cost, and at the same size, sensors vary in low light performance and all I'm suggesting is that in the current camera, out of the existing processors available for this type of camera, it might be the case that Mobius chose a sensor that was good in the day but not so good at night.

Firstly, the full frame 35mm size of the sensor is one of the primary reasons for the excellent low light performance. If you were to make a smaller (and cheaper) version of the sensor it could never perform as well.

Sorry but I'd have to disagree as whilst that maybe used to be true, with advances these days, at least one knowledgeable author suggests that is wrong. The biggest sensor doesn't necessarily mean the best low light performance:

http://www.clarkvision.com/articles/does.pixel.size.matter/

You're also quite right that there are a lot of factors that influence night performance.

......in order to take advantage of the performance of these highly sensitive sensors one needs a much larger, faster, higher quality and far more expensive lens than any that you will ever likely see in a dash cam.

Not necessarily sure of good lenses being beyond dash / action cams anymore. The Yi 4K has already moved onto Schott b270i low iron glass and a 7 element coated lens.

Thirdly, the outstanding low light performance you see in these cameras is in large part due to the capabilities of the powerful and very expensive processors used in conjunction with these advanced larger format sensors. Sony sometimes refers to their DSPs as the "image processing engine".

There's no doubt processing plays a big part, but necessarily expensive? I understand a lot of the processing is down to the firmware. Surely the processor is simply a means of processing data as per any cpu style chip. In fact the Canon DIGIC processor is said to resemble an X86 processor according to the Wiki - https://en.wikipedia.org/wiki/DIGIC#DIGIC_DV_5
 
I set both the M1 and M2 to record in H.264 due to the problems I had with replaying the video footage when recorded in H.265 using VLC on my iMac.

I use H.265. I can make original files available if that is useful?

Thanks guys. That's exactly what I expected based on your screen shots.

The developer has explained to me that the H.264 coding was proved not good enough and problematic. It is being worked on. If I understand correctly, much effort originally went into the H.265 coding on the M2 and that is the reason for the currently superior performance in H265. The H.264 coding issues coupled with an issue when using WDR is causing poor image quality and a lack of apparent sharpness, especially in areas of texture during motion such as tree leaves. I do hope this gets fixed soon as I consider it a serious problem because if you shoot in H.265 in order to achieve the better image quality and then find that you need to submit this footage to law enforcement, insurance companies or attorneys, these entities may not have cutting edge H.265 playback capabilities currently available to them and may not for some time to come.
 
Dashmallow, don't get me wrong, I'm not saying that Mobius necessarily got the sensor choice wrong by not choosing one of the above sensors, what I said was that it's a pity no-one in the dash cam world can get a hold of a sensor of such capability.

You can also see from the recommended list of dashcams on here, the most expensive camera doesn't necessarily have the best low light performance. Even at low to medium cost, and at the same size, sensors vary in low light performance and all I'm suggesting is that in the current camera, out of the existing processors available for this type of camera, it might be the case that Mobius chose a sensor that was good in the day but not so good at night.



Sorry but I'd have to disagree as whilst that maybe used to be true, with advances these days, at least one knowledgeable author suggests that is wrong. The biggest sensor doesn't necessarily mean the best low light performance:

http://www.clarkvision.com/articles/does.pixel.size.matter/

You're also quite right that there are a lot of factors that influence night performance.



Not necessarily sure of good lenses being beyond dash / action cams anymore. The Yi 4K has already moved onto Schott b270i low iron glass and a 7 element coated lens.



There's no doubt processing plays a big part, but necessarily expensive? I understand a lot of the processing is down to the firmware. Surely the processor is simply a means of processing data as per any cpu style chip. In fact the Canon DIGIC processor is said to resemble an X86 processor according to the Wiki - https://en.wikipedia.org/wiki/DIGIC#DIGIC_DV_5

Technology marches forward but you can't ignore the laws of physics. I think you are comparing apples to oranges here. For example, while you can certainly improve the quality of the glass in a lens and improve its performance but you'll never get an action camera lens to perform like a lens costing thousands of dollars that has much larger elements. While I would agree with the authors statement that the key to light gathering ability is the size of the aperture, the fact is that it is the sensor, lens and DSP working in concert with one another that determines the performance and it is highly dependent on the particular camera. Relatively inexpensive imaging processors and sensors in dash cams are not likely to ever outperform much more sophisticated ones costing many, many times the price. Accordingly, for processors I think comparing the DIGIC to an X86 vs the sophistication of the camera you cited in your precious post is a vast oversimplification. Either way, we are still talking about an SoC camera with 1/4 or 1/3 inch sensors in the 100 dollar range, including the optic (which in the case of the Mobius are excellent ones from Japan rather than China) and there is a limit of what one can expect in what such cameras can do with the current state of the art. While it is indeed perhaps a "pity" that no-one in the dash cam world can get a hold of a sensor of such capability, it because they don't currently exist and would otherwise make these cameras unaffordable.
 
Last edited:
lets just hope they come up with something that's better at capturing photons than the current materiel in use, it would be nice to have smooth 4K color night footage, and captured with a higher exposure time.

I would pay good money for such a thing ( even just 1080p ) if it was a good stealthy design and it was a actual step up in regard to night performance.
But i know we will have to make do with the slow exposure times for some time to come.
 
There would be a market for a super-duper dashcam but it would be very small which would make the cam very costly. We're currently at the technology point that will sustain the mass-production market even if better is possible, which it always is. And that goes for everything mass-produced, not just the M2 or dashcams in general ;)

My personal limit of what I'd spend for any dashcam is around $200 and it would have to be fantastic before I'd consider spending that much. I'm happy with the ones I have now selling for around half of that. IMHO that's about double of what most people want to spend till they discover that most cams in the $50 range are so crappy as to be useless. Professional quality results always takes professional equipment and that stuff is never cheap.

Phil
 
Technology marches forward but you can't ignore the laws of physics. I think you are comparing apples to oranges here. For example, while you can certainly improve the quality of the glass in a lens and improve its performance but you'll never get an action camera lens to perform like a lens costing thousands of dollars that has much larger elements. While I would agree with the authors statement that the key to light gathering ability is the size of the aperture, the fact is that it is the sensor, lens and DSP working in concert with one another that determines the performance and it is highly dependent on the particular camera. Relatively inexpensive imaging processors and sensors in dash cams are not likely to ever outperform much more sophisticated ones costing many, many times the price. Accordingly, for processors I think comparing the DIGIC to an X86 vs the sophistication of the camera you cited in your precious post is a vast oversimplification. Either way, we are still talking about an SoC camera with 1/4 or 1/3 inch sensors in the 100 dollar range, including the optic (which in the case of the Mobius are excellent ones from Japan rather than China) and there is a limit of what one can expect in what such cameras can do with the current state of the art. While it is indeed perhaps a "pity" that no-one in the dash cam world can get a hold of a sensor of such capability, it because they don't currently exist and would otherwise make these cameras unaffordable.


Some good points although I'm not sure were necessarily at apples and oranges. I think you think my expectations are probably a little higher than they actually are! I'm not expecting to see a dashcam to rival eg A Canon DSLR with an L series lens for video / stills quality. I do think however, that there's a lot of room for low light performance to be improved just as overall picture quality is improving, the latter being something we are starting to see already with the Yi 4k and SJ7000. I do also think we can get much closer to DSLR / higher end video performance, even if the true broadcast picture quality cannot be achieved.

Talking generally and about no brands in particular, just because performance hasn't improved in the past to closer to higher end video performance, doesn't mean it can't be in my opinion. A lot of industry is about turning out a product to a cost, not necessarily turning out the best, which means there's a difference between what can be done and what is often done depending on the target market. Manufacturing is often about a compromise between price and quality. There's also an element of attraction to providing to a low cost volume market, as the lower the cost the bigger the potential margins and with the ability to set a lower price point, potentially higher sales. Why would you produce something to an expensive and maybe technically difficult standard when the market (in the majority) will accept something far cheaper and easier to produce? I'd bet the majority of dashcams sold probably sell for £100 or less globally. So for many manufacturers, volume sales are probably more interesting than producing something more expensive to an ultimate quality.

However, with falling prices of technology, ever advancing technical quality, and the emergence of a more premium dash cam market in the £150-200+ range, I think there's room for improvement. Even a few months ago if I'd said someone would be using a 7 element lens with coatings and low iron glass on a dashcam, you'd have probably fallen off your chair with laughter, yet we have the yi 4K doing just that. (BTW I accept the Yi isn't a dedicated dashcam, however there is a crossover with the action cam market).

As for the Mobius, I hope they can squeeze more out of the firmware. I'm just thinking there's more to a good picture than firmware alone, especially where low light is concerned.

As for the compression, if there's more to come from the H.264 then great. However, this also suggests to me there should be more to come from the H.265 as H.265 should offer better quality because it has the potential the offer the same quality for 1/2 the file size. So back to back with h.264, a eg 20mbs file should be far superior in h.265 as it's equivalent encoding in h.264 would be around 40mbs. There is of course a ceiling of little to no visible returns with compression, but until that's hit, h.265 should offer not only the opportunity to get smaller file sizes, but also the opportunity to get higher quality by running H.265 at compression levels that were simply too high for the card / camera's bandwith on h2.64 before it. For sure once compression artefacts are out of the picture, then quality comes down to sensor, glass, firmware etc.
 
Even a few months ago if I'd said someone would be using a 7 element lens with coatings and low iron glass on a dashcam, you'd have probably fallen off your chair with laughter ...

The lenses of my 2 Mobius B cams are 7 element glass in a metal housing. Made on Japan. Immune to temperature-related focus shift. 3 years old & working perfectly.
 
The lenses of my 2 Mobius B cams are 7 element glass in a metal housing. Made on Japan. Immune to temperature-related focus shift. 3 years old & working perfectly.

And multi-coated as well! The later C and C2 Mobius lenses are unusually effective at minimizing flare.
 
After a few weeks of testing I have come to my own conclusion that the M1 is currently a better all-round camera than the M2 for dashcam use. I know some other members already had that view a while ago. I suppose I 'wanted' the M2 to be better than it is and was prepared to give it a go despite the various warnings I read on these forums.

I have now relegated the M2 to the rear view window and reinstated the M1 up front, since I have more confidence in the M1 reliably recording the better footage of an incident.
 
After a few weeks of testing I have come to my own conclusion that the M1 is currently a better all-round camera than the M2 for dashcam use. I know some other members already had that view a while ago. I suppose I 'wanted' the M2 to be better than it is and was prepared to give it a go despite the various warnings I read on these forums.

I have now relegated the M2 to the rear view window and reinstated the M1 up front, since I have more confidence in the M1 reliably recording the better footage of an incident.

I'm afraid I have to agree. The Mobius 2 has a variety of shortcomings that will need to be addressed for it to be a worthy dash cam.

New firmware is on the way that will enhance WDR, H.264 and low light performance among other concerns.
 
After a few weeks of testing I have come to my own conclusion that the M1 is currently a better all-round camera than the M2 for dashcam use. I know some other members already had that view a while ago. I suppose I 'wanted' the M2 to be better than it is and was prepared to give it a go despite the various warnings I read on these forums.

I have now relegated the M2 to the rear view window and reinstated the M1 up front, since I have more confidence in the M1 reliably recording the better footage of an incident.

What did you find for M2 limitations that were a problem?
 
What did you find for M2 limitations that were a problem?
The M2 performs rather well in very good light. WDR handles high contrast on bright sunny days. It's mostly when the light falls that it starts to struggle.

1080p60 tends towards over exposure, leading to washed out images, and also motion blur due to using a longer exposure time than necessary.
1080p30wdr blurs details when combining two frames to get the HDR effect.

I have had better results from the M2 at times, even at night, but the performance varies a lot. The M1 is capable of more consistent output at the moment. Exposure and colour balance are much better.

It's still early days for the M2. I bought it knowing that the firmware had a lot of development work ahead. The underlying hardware appears to have good potential. I am aware that the first public firmware update is coming soon, and I hope that it will deliver much improved results. Until then, I'm still using the M1 as my primary camera.
 
I tried an M2 and was impressed. Really sharp and clear. I had trouble playing h.265 films so I reverted to H.264. However, after about five weeks, it suddenly stopped working. I eventually managed to reset it to the point where I could reinstall the firmware and get it going again, but a few days later it died again. This time I couldn't get it working again. Luckily I'd bought it through Amazon so it was easy to return for a refund. The question is whether I was unlucky to get an unreliable unit, or is this a common problem? At present I'm making do with a Mobius 1 which is itself giving some problems after three years of reliability.
 
I've not heard of behaviour like that with the M2. In what way did it stop working? Was it not recording, or not turning on?

It seems to be an odd coincidence that the M1 is also becoming unreliable too. What problems are you having with it?
 
I saw something new yesterday on my drive home.

 
Back
Top