Another great article. Just a suggestion but whenever you test low light cameras with the test subject specifically being low light, I would like to see video of moving objects. I mentioned in another article that I was also recently blown away by the low light image of a particular camera, then I was let down when objects started moving. The camera manufacturer had the default shutter speed set so low that motion blur was a major factor. When speco intensifiers were popular a few years ago, this was the same deal, looked great on a still image but big time blur when objects moved.
I go back and forth on this issue. Generally, video is not worth the time/upload space because only a very small percentage of users actually look at it.
If we upload it to a streaming service we can cut down some of the storage requirements, but it is compressed and transcoded and not the original video.
We will pull a clip from this test and zip it, but this is why we generally don't do so.
As far as motion blur, I'll update the report momentarily with the following info:
Cameras were all set to 1/30s shutter or faster. None were allowed to use slow shutter. Note that Hanwha used to default to 1/5s shutter (which we would manually adjust), but in this test, the XNO-6085R defaulted to 1/30.
XNO-6085R digital noise reduction was adjusted slightly (from level 12 default to about level 8) which did not make much visible difference but removed some slight blur.
Compression is standardized to quantization of 28-30.
By the way, slow shutter is an issue for sure, but much less of one than it was a couple of years ago. Most manufacturers gave up on defaulting to slow shutter speeds because users became savvier on the topic. However, now you'll see high levels of digital noise reduction applied at times, which looks quite similar. Blurry and smeared. We talk about that in the Camera DNR (Digital Noise Reduction) Guide in depth.
Hanwha Wisenet X series have a feature under Noise Reduction called WiseNR, which I found to be pretty amazing. It analyzes the image and performs 2D & 3D noise reduction depending on if there is motion or not. When there is motion, it changes methods, and you get great motion performance with more noise. As soon as the movement clears, is changes back again, and then noise is gone.
In my tests 1/30s is still much too slow for a walking subject. Maybe my expectations of clear and sharp are too high. 1/60s is better but still not there. 1/120s seems to do quite well for normal walking pace.
Chris, thanks. Please share a few pics from your tests. We've tried various shutter speeds and for LPR with a moving car, for example, we regularly recommend much faster shutter speeds but for walking, we have not seen the benefit.
When we are testing cameras in low light, we always test them with moving object.
We walk towards camera and crosswise camera. We are making comparison from snapshots. Usually we test camera on default setting because we use sub installers and sadly they don't chance camera setups (maybe don't know how or just lazy)
So we try to find cameras which are decent/good in low light, without chancing camera settings.
Usually we test camera on default setting because we use sub installers and sadly they don't chance camera setups
I can sympathize there. At least for something as basic and straightforward as shutter speed, I really think it would be useful to require it. It does not take a lot of talent to be able to change that setting and not doing so, especially if a camera defaults to a slow shutter can be a big problem.
Bosch has a newer 1080p starlight line (tested here) which performs pretty well in color mode, but not on the level of the ExtraLux. They also have a larger imager 5MP starlight, which is also good, but overtaken by new models.
Yes, you made it clear before that you also test with moving objects, but it will be good to have some video to register that some camera is really good (or not) with moving objects under low light conditions.
Regarding Rotakin, I didn't realize it was so expensive...
To clarify, the Rotakin is not 'prohibitively' expensive, at least for us. In the last year, we have bought numerous multi-thousand dollar cameras (e.g., PTZs, mulit-imager, etc.). We could easily afford it.
It does seem to be overpriced for what it effectively is, a cardboard cut. It also has negligible recognition in North America.
As Stian suggested, we will look at including an object that demonstrates movement in the test image but it's unlikely to be the Rotakin.
That said, we'd be happy to talk to anyone involved directly in the management or sale of the Rotakin to hear their pitch.
Thank you for showing the dark scene results with the IR off and cams in color mode.
Your dark scene is pretty darn dark, so I bet this would be a great camera to leave in color mode around the clock for minimally illuminated exterior areas of campuses and parking lots.
Looking at Hanwha's website, the dome versions of this camera are remotely adjustable! That is a real labor saver during install and for when someone later on wants a high-up camera turned "just a little more to the right."
In some scenes that was true, especially when using higher F-stop lenses on the Q1659. The main drawback to the Q1659 vs. the Hanwha is lens selection, because there are not a lot of low F-stop lenses available with wider angles of view. So unless you're looking at narrow areas, the Hanwha is likely to be better in low light.
However, details are much betterduring the day in the Q1659 due to its much higher resolution. For example, at 60' during the day, this is how the two compared, a clear difference.
F1.2 is very fast to be sure, but 85mm seems an odd choice for most cam applications, due to the very narrow field of view you noted. The very cheap and wider but not as fast option is the 50mm F1.8 STM. Again compared to the 85mm, the much wider, almost as fast, and just a bit lower priced is the 35mm F1.4L II. Then there is also the 50mm F1.2 L USM, which again is wider and just as fast, and is about $500 less than the 85mm. Widest fast Canon option might be the 24mm F1.4L II. There's also a Sigma 20mm F1.4 to go even wider, and hundreds less than the 85mm. Any of these will give better wider comparisons than the 10 to 22mm F3.5, although not nearly as wide.
I suggest avoiding the 85mm when using the Q1659, except when comparing zoom levels that match and time allows also showing a wider fast lens in the same review. I don't understand why Axis would only recommend the 85mm as a fast option.