Best action cam for low light?

TequilaKez

New Member
Joined
Apr 17, 2017
Messages
18
Reaction score
2
Country
Australia
Just wondering if anyone knows which models suck the least in low light. I'd mainly be using it for filming live music situations so there'd be more light than a dark street at night, but certainly no direct sunlight.

Important for my applications are:
Good optics
Decent low light performance
Gyro/accel stabilisation
High bitrate
Very minimal NR, and sharpening, or better yet, manual control of these. I realise these sensors can be very noisy, but I'd rather deal with it in post.
Value for money (so i can get more than one for different angles).

I couldn't care about:
Resolutions above 1080p (accept if it allows for simulating different focal lengths by cropping)
Sound quality.
Water proofing.
Photos / time-lapse.
'GoPro' branding etc. Weird chinese brands are fine with me.

I'm also very interested which have largest hacking/modding interest. I can't live without Magic Lantern on my Canon, and noticed some scripts for the Xiamo Yi which is exciting. There was a Magic Lantern project started for GoPro but it died soon after :(
 
Git1 has good low light performance, doesn't have the hacking community that the Ambarella based models have though

Don't think stabilisation and low light performance go hand in hand
 
Why dont use DSLR ? even a cheap one still better than action cam.
 
Why dont use DSLR ? even a cheap one still better than action cam.
I have 2 DSLRs which I love shooting RAW on magic lantern. I'm looking for action cam for pocket size/convenience of being able to quickly clamp them to things on stage or peoples instruments. I had a GoPro Hero 3 black, and Protune mode was a usufel for me, but it sucked in low light, and I don't have it anymore.
Things have come along way since then with image quality, stabilisation etc and soooo many more choices than GoPro and better pricing. I can't really test anything though so need to rely on reports on forums like these, but there isn't a lot on low-light performance.
 
Then you may try download sample of sj4000+ or git1 and see is it ok or not. These are only 2 action may suit your request.
 
Then you may try download sample of sj4000+ or git1 and see is it ok or not. These are only 2 action may suit your request.

Thanks for the tip. Would you care to elaborate on why you think this? Any ideas where I might find such samples?

On a side note, I'm tempted to wait for the Yi 4K+ as the specs would suggest top of the line quality and I can only assume that goes hand in hand with decent low light performance. (I should note by 'value for money' I don't necessarily mean cheap.)
 
Thanks for the tip. Would you care to elaborate on why you think this? Any ideas where I might find such samples?

On a side note, I'm tempted to wait for the Yi 4K+ as the specs would suggest top of the line quality and I can only assume that goes hand in hand with decent low light performance. (I should note by 'value for money' I don't necessarily mean cheap.)
dont bother to high resolution action cam. They are good at daynight and such at nigh.
If they're both good at day and night, so why do we have to buy DSLR ?
 
I'm aware these cameras are geared towards daylight, but their convenience and size makes them useful elsewhere, they are also incredibly cheap for the quality of video you get compared to DSLR. The fact is that [in low light] some are ok, some are bad, and some are downright terrible. I'm just hoping to avoid the last 2 categories. I'm quite familiar with de-noising low light footage, but it works best if there is minimal in camera NR applied, hence my interest in hacking communities where I can see some have had success disabling it in some models.
 
Last edited:
the two models mentioned have lower megapixel sensors so will offer better low light performance due to larger individual pixel size than most action cameras
 
yes, because of hardware, so, nothing you can do to fix or improre that. :) sorry
 
the two models mentioned have lower megapixel sensors so will offer better low light performance due to larger individual pixel size than most action cameras
Ok that makes sense, thanks for the info.

yes, because of hardware, so, nothing you can do to fix or improre that. :) sorry
Although I appreciate the answers, I think maybe you don't understand my question. I am aware of the shortcomings of action/dash cams, and I don't wish for magical hardware changes, theres nothing to be sorry about :)

But each has a different combination of the following which affects low light:
- Lens quality/speed
- Sensor pixel size
- How ISO/shutter speed is set, and whether it can be set manually like DSLR (e.g action cams love to set high shutter speeds to catch movement which is bad for low light. With manual ISO/Shutter at least I have control over this.)
- how NR is applied, and whether it can be changed.
- how sharpening is applied, and whether it can be changed. (sharpening looks terrible on noisy video, or video with too much noise reduction)

I was just hoping to get a bit of an idea of which models have a combination of the above that favours low light situations more than other models.
 
Ok that makes sense, thanks for the info.


Although I appreciate the answers, I think maybe you don't understand my question. I am aware of the shortcomings of action/dash cams, and I don't wish for magical hardware changes, theres nothing to be sorry about :)

But each has a different combination of the following which affects low light:
- Lens quality/speed
- Sensor pixel size
- How ISO/shutter speed is set, and whether it can be set manually like DSLR (e.g action cams love to set high shutter speeds to catch movement which is bad for low light. With manual ISO/Shutter at least I have control over this.)
- how NR is applied, and whether it can be changed.
- how sharpening is applied, and whether it can be changed. (sharpening looks terrible on noisy video, or video with too much noise reduction)

I was just hoping to get a bit of an idea of which models have a combination of the above that favours low light situations more than other models.
because you said you are interesed with mod/hack firmware. I made some mod myself and i have a conclude: you can made image be better in daylight with mod firmware but you will almost cant made image be better in nightlight. At least with the method ppl doing now: increasing resolution.
 
The vids are scarce and hard to find, but I almost bought a Git1 for it's low-light performance ;) Had someone hacked a super-cap conversion I would have one now, but the world got focused on the Git2 and the older cam is about forgotten now which is a real shame :(

Phil
 
because you said you are interesed with mod/hack firmware. I made some mod myself and i have a conclude: you can made image be better in daylight with mod firmware but you will almost cant made image be better in nightlight. At least with the method ppl doing now: increasing resolution.

Ok right, no just interested in hacks for disabling noise reduction and sharpening so I can do it in post. Also, with Yi 4K+ having 120mbp+ data rate, a hack could potentially grab 1080p RAW, something like magic lantern. Its really quite amazing how good an image can be pulled out of super noisy low light footage, provided no NR, sharpening or compression has been applied to it.

The vids are scarce and hard to find, but I almost bought a Git1 for it's low-light performance ;) Had someone hacked a super-cap conversion I would have one now, but the world got focused on the Git2 and the older cam is about forgotten now which is a real shame :(

Phil

Here's a thought, chip and cam manufacturers pay attention! :)

We all know that 4K sensors look ****e in low light because instead of making a bigger sensor, the photsites are made smaller to squeeze more in, so much less chance of photons hitting them in low light.
So why not have a 'mode' that sums (as apposed to average) 4 adjacent pixels into 1, effectively halving x and y resolution, but resulting in much brighter values and less noise (due to the averaging affect of summing).
So then you could have 4K sensors, with a special low light mode that simulates a 1080p sensor. Half the res, but a lot more light!
This would need to happen in camera though, before processing and compression.

Maybe this is already possible with Ambarella and we just need cam devs to give us a firmware update, or hack.

[EDIT] Ok, seems what I'm talking about is known as pixel binning, and it's been in use for a while. What i don't understand is, if the technique is so well known, why don't we see it on 4K action cams?
 
Last edited:
What i don't understand is, if the technique is so well known, why don't we see it on 4K action cams?
it is in mobius, but not ready work. Only slighly effect.
 
the Git1 does a lot better than most for low light, @Nigel or @Mtz might have some samples you can download to see the results
The Git1 is the only action camera I know of currently available that has a low megapixel sensor with big pixels to do well in low light and decent control of exposure, nearly all others are 12MP 4K sensors even if they don't do 4K.

It does do better in low light, in fact it also does better in brighter light if you are after 1080 video at 30fps and not anything higher resolution or faster frame rate.

None of the other action camera manufacturers seem to be taking any interest in low light, it is all about 4K.

One other that will be available soon(ish) is the Gitup Git3, which will have better control over settings and extra features over the Git1, like gyro stabilisation. If it is available by the time you are buying then it will be worth looking at.

Important for my applications are:
Good optics
Decent low light performance
Gyro/accel stabilisation
High bitrate
Very minimal NR, and sharpening, or better yet, manual control of these. I realise these sensors can be very noisy, but I'd rather deal with it in post.
Value for money (so i can get more than one for different angles).
Don't expect DSLR quality optics, there will be some fisheye, but the Git1 is decent for an action camera, and Gitup does sell a compatible no-fisheye lens.
Git1 Doesn't have Gyro stabilisation, but you don't want it anyway if you are mounting it on stage or on instruments (eg mounted on a violin would look terrible with stabilisation), you would if mounting on a person.
Git1 isn't particularly high bitrate, but it does a decent job for 1080/30.
Gitup does allow you to turn the sharpening down to minimal, although the Git1 normally does a good job of sharpening anyway.
If you are getting more than 1 then I suggest that 1 is a Git2, or preferably a Git3 which has a better sensor for low light. The Git2&3 are a lot more configurable on exposure and apart from low light capabilities are just better (next generation) cameras.

This was taken with my Git2, note the gyro stabilisation in the dark. The audio on that is also all from the Git2.

This was from the Git1:

And the equivalent from the Git2 (note the background and smoke is much darker than the Git1 because of the lower sensitivity):
 
Last edited:
Great thanks for the videos, it shows a lot. Git1 has good light, petty about the bitrate, it looks atrocious. How much of those compression artifacts are youtube would you say? does it look that bad straight off camera?
 
YouTube takes 15mbit files and compresses them to 4mbit, quality suffers as a result
Well no mystery there. I was kinda hoping @Nigel might comment as to whether those particularly bad artifacts are present in his original video (Sidmouth Fireworks Git1).
 

Attachments

  • Screen Shot 2017-04-19 at 6.45.52 pm.png
    Screen Shot 2017-04-19 at 6.45.52 pm.png
    231.6 KB · Views: 10
Back
Top