- News
- Reviews
- Bikes
- Accessories
- Accessories - misc
- Computer mounts
- Bags
- Bar ends
- Bike bags & cases
- Bottle cages
- Bottles
- Cameras
- Car racks
- Child seats
- Computers
- Glasses
- GPS units
- Helmets
- Lights - front
- Lights - rear
- Lights - sets
- Locks
- Mirrors
- Mudguards
- Racks
- Pumps & CO2 inflators
- Puncture kits
- Reflectives
- Smart watches
- Stands and racks
- Trailers
- Clothing
- Components
- Bar tape & grips
- Bottom brackets
- Brake & gear cables
- Brake & STI levers
- Brake pads & spares
- Brakes
- Cassettes & freewheels
- Chains
- Chainsets & chainrings
- Derailleurs - front
- Derailleurs - rear
- Forks
- Gear levers & shifters
- Groupsets
- Handlebars & extensions
- Headsets
- Hubs
- Inner tubes
- Pedals
- Quick releases & skewers
- Saddles
- Seatposts
- Stems
- Wheels
- Tyres
- Health, fitness and nutrition
- Tools and workshop
- Miscellaneous
- Tubeless valves
- Buyers Guides
- Features
- Forum
- Recommends
- Podcast
Add new comment
59 comments
Here are my main points:
* Battery life - 5-plus hours (including for light if rear camera included) in all temperatures
* Preferably able to change batteries - ie removable
* Good, secure mounts that don't interfere with riding - some rub on legs
* Small and lightweight - made from decent materials
* Good image quality able to pick up registration plates clearly in (nearly) all situations/conditions
* Reliable - so waterproof and drop proof (probably too much to ask to be crash proof)
* Charges quickly
* Companion app easy to use
* Simple way of uploading footage
* Buttons that are easy to use with winter gloves
* Some sort of way to automatically record - ie crash mode and maybe close pass - so that footage is easy to find
* Doesn't cost the earth - some front and rear systems cost an arm and a leg - so I'm thinking sensibly priced as all these requests don't come cheap
Achieve all (or most being truthful) of those and I'll buy without hesitation
Image quality by frame rate.
While 60fps gives a better quality image, it leads to significantly longer processing times.
This might not be an issue of you are adding telemetry to a 5 minute clip, but when you are adding it to a 2 hour ride it becomes quite serious.
My i7 laptop, using Dashware, will normally take around 3 hours to add telemetry to a 2 hour 30fps video.
As an experiment, I've changed from 30fps to 60fps and it took over 15 hours to add the telemetry on a 2 hour recording.
I would suggest a user selectable variable frame rate (30 / 60 fps) to accommodate different end video usage.
Back again with a thought on image stabilisation: don't bother with it.
Only optical image stabilisation is capable of improving the clarity of the image, but it requires moving parts and is quite complicated to implement in a way that will survive the vibrations when mounted to a bicycle.
Digital stabilisation that uses an accelerometer to compensate for motion, is done to make the footage more pleasant to watch. It does this by using a wide angle lens, but then throwing away imaging data in the extremities in order to smooth out motion. This takes quite a bit of power consumption to do on the device. The reduction in image data and battery life is not worth it to me.
Please let us know what difference in image data and battery life you got, with or without digital stability, on which device?
That would be a great help...
Presumably if the accelerometer data is recorded then digital image stabilisation could be carried out in post processing just as effectively. Which is to say that it still won't be able to fix subject movement, which can only be ameliorated by shorter shutter times.
Also, I don't see any effort to design the camera mount to damp vibrations before they can impact the image.
@siracha Yes, you could embed the accelerometer data in the video file to enable post-processing digital image stabilisation. Capturing this data is very low power and an accelerometer/gyro would be running anyway to do the crash detection function. This approach would also retain the entire captured image data from the sensor.
I think a shock absorbing mount is a good idea. But if your
frame rateexposure is fast enough and you are using a global shutter image sensor, then it should not be needed for clear imaging - just to make viewing less vomit inducing!@lonpfrb I've not done any testing, but image stabilisation does consume quite a lot of power. So much so, that GoPro's won't do it when the battery is cold! The only benefit of digital stabilisation is to make the footage nicer to watch. For a camera that's focused on evidence capture, that is a lower priority.
Maybe we are talking about two different degrees of image stabilisation:
1) "anti-vomit"; stabilising the horizon, smoothing out the general sway and tilt of the camera (ie low frequency, high amplitude), can be done in software but crops the frame edges. Generally this is what is meant by video image stabilisation.
2) "anti-blur"; still-image stabilisation, counteracting movement of the camera during the frame capture (shutter open time), so that the individual frame is sharp. This can't be done in software, the movement has to be physically counteracted by moving either or both the lens and sensor - or minimised by using a shorter shutter time (needs a bigger aperture and/or more sensitive sensor). Shutter speed is also the only way to contend with subject (as opposed to camera) movement blur.
The second is the important one if you want to read a number plate. The individual frame has to be sharp, regardless of any nauseous movement between frames. Active OIS obviously eats battery. It's also a potential weakness in the face of heavy vibration.
Quadlock have made a vibration damping phone mount for motorcycles. There is some discussion on their site about reported vulnerability of phones' OIS mechanisms when subjected to vibration on motorcycles, and their aim is to protect the OIS (rather than to enable clearer video). However the same approach could be used to improve the video - effectively it is passive OIS.
Spot on. Your example 1 is digital stabilisation. Example 2 optical stabilistation.
Optical can be done by moving just the lens/sensor (better for small amplitude and fast vibrations, as the lens has low mass/inertia so can react fast), and/or by moving the entire camera on a gimbal (better for larger amplitude and slower movements).
OIS can remove vibration blur from a longer exposure. The problem is when the object you wish to image is also moving, you need an exposure time that is short enough to capture that moving object without blur too.
Some testing is needed, but it's possible that an exposure time short enough to capture a car numberplate passing at 60mph without blur, is fast enough to deal with any vibration blur from mounting directly to a bike. With active illumination at night, my hunch is OIS will not be needed to get a clear image.
I've been mulling over the passive stabilisation idea. I think you could be onto something here. I'm a bit sceptical about damped phone mounts, as the phone/camera is quite light and you do need a bit of mass in your sprung object. So I got onto thinking about passive gimbals like this...
The downside to these is the extra weight added to a camera system. But what if you designed your camera to be shaped like a passive gimbal, with the heavy batteries low down and the lighter camera up top, the mounting point providing the articulation? It's a bit of a wacky idea and might not work too well on a platform that's moving a lot. It could just end up flapping all over the place, but maybe worth a look.
Hello Steve and forum members.
I have been using your DC-1 camera for over a year now and I am generally very pleased with it. I submitted a sample of my experience of a bad driver to you which is up on your YouTube page (https://www.youtube.com/watch?v=6UJk1sirRww). The view of the car's number plate was quite visible in many of the frames and my own file is less compressed that the YouTube footage so a bit better still. My memory is that the light level of the footage is brighter than the light level as perceived by my eyes!
What I really like about the camera is that I have two cameras but only 1 battery/charging point, one on/off switch and one memory card. This is the kind of convenience I'd like to see in a bike mounted two camera setup as the thing I don't really enjoy about the DC-1 is that I have to wear it on my head. If my camera was split into two parts with the front lens, battery, controls and mem card mounted to my bars, then a single wire going to the little rear camera attached to my rack or seatpost, that would be great.
The other thing I'd like to see is a way to change settings via my PC rather than a phone app. I have a smartphone, but I am very wary of the apps that I install on it due to privacy concerns. I did look at the VF Cam app that is used with the DC-1 but found that it came with a bundle of Facebook trackers that I am uncomfortable with. I know that puts me out of step with most of humanity but I don't have a FB account, I haven't ticked the box to accept their terms and conditions, and I don't see why they should be able to track my location, contacts, media files etc, etc. This is not a problem for the most part as I can download and view my files on my PC and I was able to position my camera with a little trial and error by the same method. It would be nice to be able to view the settings though, and I guess that my time-stamp is only correct for 6 months of the year.
Anyway, I like the camera, appreciate the cutomer service response time (my experience is counted in minutes rather then days/weeks!) and would buy again from you.
Best wishes.
Back here with an imaging sensor suggestion:
The OmniVision OG02B1B
This looks to be perfect for a bicycle camera:
- Global Shutter: no wibbly, wobbly, jellyvision artefacts
- B&W, supports IR at 850 & 940nm, with strobe sync: crisp licence plate capture in the dark with active infrared led illumination.
$10 unit cost (MOQ applies)
Available as a board camera for RaspberryPi, useful for prototyping.
Waterproof; image stabilisation; at least 1080p quality image @ 60 fps; day and night capability; 2 hour battery life with ability to connect external battery via USB; small size (about matchbox size) and no screen; handle large capacity micro SD cards; user determines file size and looping or not; not a GoPro; use standard mounting fittings that can be purchased at reasonable prices from any photographic equipment supplier.
I am currently using this https://m.youtube.com/watch?v=uooaSBe7sxc Its only deficiency for me is that it's not waterproof. Price is right. Images could be slightly improved but still quite good.
I have to carry four batteries around for my current (admittedly cheap) camera because none lasts more than an hour before running empty. It's affixed to the bike and is very bulky. As a result of the bulk (and I am slightly ashamed to admit this), I don't use it on the summer bike or TT bikes. Decent image clarity, good battery life, and a compact/aero design all key for me. I'm sure that's really easy to achieve!
As a lot of people have said already, number plate clarity is key.
Other features that could work for me would be a radar sensor recording actual closeness of vehicles, and multiple fixing options at the rear including a rack mount.
Contrary to what some others have said, integrated lights would appeal, particularly for the front where space to fix things is at a premium. There is a front light that acts as an out front mount (can't remember who makes it) so that sort of idea might be appealing.
Yes, if there was the option to wire it into my dyno hub and still have it weatherproof, I'd be sold!
As we're talking about ideals here - why not one that has some sort of sensor/radar that will automatically sense when a vehicle is passing too closely and save that section of video instead of having to stop and press a button (which I can't properly see without my reading glasses anyway!).
Being able to read the number plate of any vehicle involved in an incident is probably the most important feature. The police can only act on video evidence if the number plate is clearly visible.
Previously I have mounted a garmin elite camera facing rearwards on underside of the right hand drop of the handlebar of my road bike. In that position you get a good view of overtaking traffic, but more importantly, it gives a very clear view of how close they are as they pass because your leg and the rear of the bike are visible in frame for reference. A bullet type front and rear combined cam might work well like this. An expanding pug that fits into your bar end might be a nice solution to mounting this way and it wouldn't take up any room on the handlebar itself.
Maybe some sort of lidar or distance measurement built into the side of the camera so segments of video where you're being close passed are automatically highlighted? The distance could also be overlaid on the video. Also the time and date stamp need to be automatically updated, the cycliq ones are a faff to set.
I haven't seen anyone mention temperature, yet. As an owner now of two generations of Fly6, they are useless in the cold. My first one would silently shut down below about 45 °F / 7 °C. I actually made a cosy for it out of foam beverage bottle/can holders, which helped a little. My current one doesn't do that, but the battery life is useless below freezing, maybe an hour.
I ride in all weather and would like my equipment to work in it. I think a reasonable range would be -30 to 40 °C (-22 to 104 °F.) Yes, I have biked in that entire range.
My ideal camera is one that can cope with an all day ride, so I am looking at 80+ miles at say 15mph moving average, what's that? Quite a lot of battery. I already charge my lights after a ride, so I have a mains connection and USB multi-charger ready to hand. Standard handlebar mounting and consideration on the rear for mounting either on seatpost or seat stay.
I do not want built in lights, I want them compact.
Absolutely waterproof - enough to cope with a car deluging me through a puddle as if I had several buckets of water thrown over me.
Quality, good enough to pick up number plates in all weathers and under street lights.
Ideally rather than storing a ride's worth to scan through, I'd like a alert button which would then trigger a function on front and rear to keep the previous 5 minutes and the following minute, to allow for incidents. A crash trigger would automatically do the same. All the rest of the video can go in the bin as far as I am concerned. The important thing is an easy way to set the time (yep, I've had a Fly6).
I want an easy way to clear and reset the memory and a quick way to access the video without having to unmount the camera and ideally without having to unload video cards. I would quite like it to detect my home wifi, which extends to where I store my bike and upload the video automatically after a few minutes when it is convinced it is back home.
Basically, I want to spend the minumum time reviewing, just be able to check and upload. I then want some easy viewing software that allows a very quick review of any incident, together with very lightweight editing tools, basically trimming the video and very little else. Perhaps something to merge front and rear views automatically. The other thing is a blur enhancer, I've had success with a shareware blur enhancement tool to go all Ethan Hunt and retrieve a difficult number plate. Anything more, there are plenty of proper video editing tools to play with.
How about having a front camera with a smallish video screen on it that's visible whilst riding. Then, have a rear camera that links to it (WIFI? Bluetooth?) and have the rear view displayed on the front camera.
Yes, I have not seen that but I woyld like it.
I stopped using a cam because I hated charging it.
I think there's scope for a front and rear cabled camera which places a large capacity battery on a bottle cage mount. Like a pump mount so you can still use the cage for a bottle. That way you can shrink the cam, place the weight lower down in the frame. You can also potentially use security mountings for the cameras because they won't need to be removed regularly. Cables for camera plug into battery mount, battery is removable for charging or carrying a spare, and contains the SD card. Basically like how first gen Di2 worked
I guess you could even tie the cameras to the battery and mount cryptographically so they're useless on their own, reducing risk of theft
I doubt the perfect camera can be made and so here are some observations.
For commuting my priority is plate recognition in low light and removable battery with 2 hours battery life and a separate charger. Easily removable from mount as bike will be left unattended.
For longer leisure rides battery life is the most important attribute for me. I tend to do longer rides in good weather.
OK to combine rear camera with a light but not the front.
Not interested in loop recording or emergency write protecting files as I always view and download incidents when I get home and large capacity cards are relatively cheap now. Ease of formatting the card in camera I would find useful for the rare occasions when there is nothing worth viewing.
Don't need a screen. A small LCD display that can be used to change settings like the front gopro ones would be useful but I'd be happy to use a PC app to do the settings if it reduced cost and made it easier to waterproof. Or do we need settings at all. 1080 at 60fps would seem to be adequate for what we need so one button to start and stop recording may be enough.
The major issue that no one seems to be able to crack is low light performance so the ability to record a spoken plate reliably may be an alternative. This would probably mean some sort of external mic which could be worn or one directed straight at the rider from where the camera is mounted.
It seems that a 1" sensor is now available with claimed better low light performance but the cameras cost around £500. If the cost comes down it may be worth looking into.
Getting the number plate is the key problem. Subject movement blur seems to be the usual problem, and that can only be fixed by short shutter times. When ambient light is low, that's difficult.
I wonder whether it could be solved by synchronising about 2 fames each second with a burst of flash light, at the highest possible shutter speed. With number plates being retroreflective they will be bright under the flash illumination, and should be correctly exposed at the commensurately short exposure time, with movement blur eliminated.
The rest of the frame may well be totally underexposed, but who cares? The dynamic range necessary to get reflective plates and the rest of the scene both correctly exposed is doubtless lacking in any case. The remaining frames will capture the scene.
The light flashing at 2Hz will double as a visibility aid.
The same scheme could work on a simpler unit without a built in light, with the proviso that the cyclists runs a front light on constant beam, positioned close to the camera.
Watching the footage might be unpleasant with two frames every second underexposed. However I would think that desktop software with some AI smarts could reconstitute the exposure using the adjacent frames.
but in the digital world its not really about capturing frames anymore, its all to do with light hitting a CMOS sensor chip, and the quality of the CMOS sensor chip that captures the photons of light and then the conversion process of that into a digital value is key.
Cheap cameras tend to have cheap sensor chips and a conversion process that leads to poor quality images, especially under low light levels because they struggle to pick up the change across the pixels which leads to a blurring effect.
some more expensive cameras have more expensive sensor chips, but are let down by a poor conversion process which might just be because theyve tailored the end footage to look a certain way, which isnt ideal for capturing numberplates in the dark, but is great for skiing down a mountainside and looking visually impressive.
ultimately it all comes down to the sensor chip & what price point the camera aims at completely fixes the quality of that chip you will get.
A CMOS sensor is not so different from a piece of film, both are just a mosaic of buckets that fill up with photons in proportion to the image brightness falling upon them. To capture a moving image on film each frame requires a whole new piece of film. With a CMOS chip the buckets just all get emptied each time to give a fresh slate, the photons are counted before they are emptied and the numbers are recorded as a digital frame.
Regardless of how you capture the image, to freeze subject motion and hence eliminate blur you must have a very short capture time. Otherwise the same location on the subject is smeared across multiple locations on the frame. Other things being equal, to capture the necessary number of photons over a shorter period of time requires a brighter subject (ie a greater rate of photons admitted over a shorter window of time). And yes, it is still frames, digital or otherwise.
This is the key missing feature from all bike cameras available today.
It's missing because most are being sold as 'action cameras' for capturing vlog worthy video, rather than 'safety cameras' for capturing evidence (even Cycliq are guilty of this).
It's not hard to find the answer. Just look at any ANPR camera (Hikvision, Petards, etc.), do they have the latest high-res, high dynamic range, backlit sensor? Nope, they have a black & white sensor and active illumination (usually with infrared light in the non-visible spectrum).
A B&W sensor uses the whole pixel to capture light. A Colour sensor has to split that pixel three ways and light has to pass through IR and colour filters, reducing intensity prior to even hitting it. Numberplates in the UK are reflective and will throw back any light you aim at them.
Brilliant idea. My initial thought was that you wouldn't have the colour of the car but that be got from DVLA site so no problem.
Yep, it's also easier to remember the colour of a car than the numberplate and only one of those is enough evidence to uniquely identify it!
In low light conditions, especially under the old sodium streetlamps, colour reproduction can be way off anyway. So of little value. A B&W sensor in the daytime would enable a faster shutter speed for less motion blur, which is more important to me than colour information.
An IR LED and a B&W CMOS camera would get the job done for a low cost.
Pages