Night Vision Comparsion on 8 Popular Dahua IP Cameras

[Jamestysecurity]

Hi everyone! My name is James, the Technical Manager in an authoried Distributor of Dahua in Melbourne Australia. Recently, increasing customers and clients have become interested and have enquired more about the night performance of cameras, I will therefore perform a series of tests to evaluate and compare the quality of several Dahua IP cameras with night vision functions.
The tested cameras are listed as follows:
IPC-HDW2431TMP-AS 4MP Startlight IP camera
IPC-HDW2831TMP-AS 8MP Startlight IP camera
IPC-HDW3441TMP-AS Lite AI 4MP Startlight IP camera
IPC-HDW5442TMP-AS Pro AI 4MP Startlight Plus IP camera
IPC-HDW5442TMP-AS-LED Pro AI 4MP Full Color IP camera
IPC-HDW5541TMP-AS-PV ActiveDeterance 5MP AI Camera
IPC-HDW4631EMP-0280 6MP IR IP camera
IPC-HDW4831EMP-ASE 8MP IR IP camera


In scenario 1, 8 cameras were installed in front of the door and were tested one by one, with some road lamps under this condition. When a camera was tested, the IR or LED lights of the other cameras were covered by black tapes in order to eliminate the influence. Pictures taken at night are shown below.


It is clear that 5442-led, 5442, 5541, 3441 and 2431 provided color pictures in auto mode (Default settings—cameras automatically choose the color image or B/W image), whereas black/white images were shown in 2831, 4631 and 4831. It was strange that although 2431 and 2831 were both Dahua’s new-generation starlight cameras, 2831 is in IR mode in the same environment compared to 2431’s color picture. As such, if we manually change 4631, 4831 and 2831 to a color mode, what will happen?

The image quality of 2831 improved after being switched to the color mode and the image even reached to the level of 2431. Therefore, if the light is not too dim at night in certain situations, it is much better to manually change the day/night settings of 2831 from auto mode to color mode.

In scenario 2, all cameras were installed in a dark room in order to stimulate a very dark environment.


Under this totally dark condition, 7 cameras were switched to the IR mode because the light was so weak that it could not reach the minimum illumination of the camera. However, the fullcolor camera still provided a color image due to its LED supplement lamp.

According to the observations above, the form was created in order to help you compare the night performance of these cameras much more easily.

Model​	Environment has some light​	Complete Dark Environment​
IPC-HDW5442TMP-AS-LED​	5 Stars​	5 Stars​
IPC-HDW5442TMP-AS​	5 Stars​	4 Stars​
IPC-HDW5541TMP-AS-PV​	4 Stars​	3 Stars​
IPC-HDW3441TMP-AS​	3 Stars​	3 Stars​
IPC-HDW2431TMP-AS​	3 Stars​	3 Stars​
IPC-HDW2831TMP-AS​	2.5 Stars​	2 Stars​
IPC-HDW4831EMP-ASE​	2 Stars​	2 Stars​
IPC-HDW4631EMP-0280​	1.5 Stars​	1.5 Stars​

According to the above information, we can get the testing results listed as follow:

1. 5442-LED and 5442 are the top two night vision cameras, while 5442-LED is better in a complete dark environment owing to its extra LED lamp. Therefore, if you want get the color image in a very dark condition, 5442-led fullcolor camera is a only choice.
2. 3441, 5541 and 2431 are in the second echelon. Because of the new generation COMS startlight sensor, all these three cameras have a desirable performance at night. Even in a totally dark condition, the IR images look brighter and make less noise compared to 2831, 4831 and 4631.
3. 2431 is the most cost-effective camera in terms of night vision. If you want your camera has some AI functions, like SMD or Smart IVS, new 3 series lite AI camera is also a good option.
4. As for 2831, we recommend to manually adjust the Day/Night Mode instead of Auto mode. Because this camera is more likely to stay IR mode than color mode in a lowlight environment at night, which, to some extent, diminishes its night performance.

I also recorded some videos of these cameras and will upload to the youtube later.

 

[alastairstevenson]

Thanks for doing this - it's good to have a controlled comparison to help make choices.

Default settings—cameras automatically choose the color image or B/W image

It's fair to say that a key aspect of night-time video is the amount of motion blur that occurs.
It's easy to get great night-time images when there is no movement, especially with long exposures.
It can be incredibly disappointing to have a video of an incident to then discover it's useless due to blurred detail.

If it's still feasible - can I suggest that the comparison is made with an exposure setting of either 1/50 or 1/100 or even better an exposure that's been set short enough to start reducing image quality or increasing image noise?
Maybe also with IR/LEDs off and Day mode forced, such that each is exposed to the same scene illumination.
That will really start to differentiate the low-light capabilities of each model which I'm not sure that your static images bring out.

 

[Jamestysecurity]

Thanks for doing this - it's good to have a controlled comparison to help make choices.


It's fair to say that a key aspect of night-time video is the amount of motion blur that occurs.
It's easy to get great night-time images when there is no movement, especially with long exposures.
It can be incredibly disappointing to have a video of an incident to then discover it's useless due to blurred detail.

If it's still feasible - can I suggest that the comparison is made with an exposure setting of either 1/50 or 1/100 or even better an exposure that's been set short enough to start reducing image quality or increasing image noise?
Maybe also with IR/LEDs off and Day mode forced, such that each is exposed to the same scene illumination.
That will really start to differentiate the low-light capabilities of each model which I'm not sure that your static images bring out.

Definitely the image quality will drop if there are some movements in the environment. I do recorded the video of these 8 cameras, however I don't have time to do the post-production for these videos in order to intergrate all 8 cameras' footages simultaneously. I have already carefully watched these cameras' videos one by one and found that the comparison between these static images is pretty similar to the playback video. That's why I post the static comparison pictures at first.
Your suggestion is a good idea and I will try to make a short exposure settings in order to get a more accurate comparison of low-light capabilities. And the comparison videos will be upload to the YouTube soon.

 

[samplenhold]

It's fair to say that a key aspect of night-time video is the amount of motion blur that occurs.

I agree with this and encourage you to put a person walking around in your test at night. The main reason for a camera is to capture a good enough image of a face to be able to identify the person.

So even if you do not want to post videos, then a still capture like you have done, but with a moving person's face for comparison, is what would be the next best comparison.

 

[Jamestysecurity]

I agree with this and encourage you to put a person walking around in your test at night. The main reason for a camera is to capture a good enough image of a face to be able to identify the person.

So even if you do not want to post videos, then a still capture like you have done, but with a moving person's face for comparison, is what would be the next best comparison.

Like I said, I do recorded some video for these 8 cameras but no time to do the post-production job, but will upload the comparison later. This weekend,I will post the comparison images with people waling around. But if you carefully see some images that I posted, you can find a cat, which to some extend, help you imagine if there was a person standing in front of the camera. hahahha

 

[SouthernYankee]

Post 8 videos with people walking an walk past the cameras at night.
Are the cameras all using there default settings?

 

[dcmkii]

I'm glad I came across this forum again. Currently in the market for a good low light performing camera but at the same time needing a camera which can capture licence plates clearly during the day.
I've been narrowed myself down to the 2831 and already started asking for quotations. But now coming to think about it, i would imagine higher fps at 25/30 would identify cars driving pass more clearly than 4k @ 15fps.
Can someone confirm whether I've made the right decision about choosing the 2831. Would the 5442 be able to capture licence plates good enough?
I can see why this forum raves about the 5442 for low light, but how about clarity and details? In the dark room example above, the items on the table seem washed out on the 5442, whereas the 2831is a little less intense and seemingly retains the details.

I would imagine a 6x IPC-HDW5442TMP-AS cctv package would be insanely expensive? as opposed to 6x IPC-HDW2831TMP-AS. There doesn't seem to be anyone in Melbourne which offers the 5442 as a package at all.

 

[mat200]

I'm glad I came across this forum again. Currently in the market for a good low light performing camera but at the same time needing a camera which can capture licence plates clearly during the day.
I've been narrowed myself down to the 2831 and already started asking for quotations. But now coming to think about it, i would imagine higher fps at 25/30 would identify cars driving pass more clearly than 4k @ 15fps.
Can someone confirm whether I've made the right decision about choosing the 2831. Would the 5442 be able to capture licence plates good enough?
I can see why this forum raves about the 5442 for low light, but how about clarity and details? In the dark room example above, the items on the table seem washed out on the 5442, whereas the 2831is a little less intense and seemingly retains the details.

I would imagine a 6x IPC-HDW5442TMP-AS cctv package would be insanely expensive? as opposed to 6x IPC-HDW2831TMP-AS. There doesn't seem to be anyone in Melbourne which offers the 5442 as a package at all.

Hi @dcmkii

Remember to check out the LPR section.

IN low light conditions the 4MP 1/1.8" sensor cameras out perform the 8MP 1/1.8" cameras ( of the same sensor chip generation / design ).

Then you need to decide how many pixels on target you need and the range from the camera to the plate you are wanting to read. See IPVM calculator to get an idea of this during the day light conditions.

 

[dcmkii]

Hi @dcmkii

Remember to check out the LPR section.

IN low light conditions the 4MP 1/1.8" sensor cameras out perform the 8MP 1/1.8" cameras ( of the same sensor chip generation / design ).

Then you need to decide how many pixels on target you need and the range from the camera to the plate you are wanting to read. See IPVM calculator to get an idea of this during the day light conditions.

Hi @mat200

Unfortunately the IPC-HDW2831TMP-AS wasn't on the IPVM calculator. Instead I used the DH-IPC-HDW4831EM-ASE to test the megapixel vs the IPC-HDW5442TMP-AS.
The longest range to identify plates is 11 metres. or roughly 35 feet if you like, I was looking at the sample pic of the guy holding the car plate, not sure if I'm using the tool correctly but I think the 8mp did slightly better vs 4mp. Wasn't a drastic enough difference though.

How does the price of the 5442 compare to the 2831? I can't find a site in Australia from the same vendor which can give me a price for both. 5442 is rare.

 

[mat200]

Hi @mat200

Unfortunately the IPC-HDW2831TMP-AS wasn't on the IPVM calculator. Instead I used the DH-IPC-HDW4831EM-ASE to test the megapixel.
The longest range to identify plates is 11 metres. or roughly 35 feet if you like, I was looking at the sample pic of the guy holding the car plate, not sure if I'm using the tool correctly but I think the 8mp did slightly better. Wasn't a drastic enough difference though.

How does the price of the 5442 compare to the 2831? I can't find a site in Australia from the same vendor which can give me a price for both. 5442 is rare.

Hi @dcmkii

I actually do not have the newer models in, so much of my knowledge is theoretical and from what others have been sharing on those.

In theory a 8MP "4K UHD" camera has 2x the ID distance of a 2MP "1080P" camera at the same FOV based on basic math. ( see cliff notes for more ).

A 4MP camera at the same aspect ratio will be 1.5x the 2MP camera.

This is for well lighted conditions, assuming effective pixels.

In practice in low light conditions a 4MP or even 2MP can often out perform a 8MP camera - it really depends on the size and quality of the sensor as well as the lens, DSP, and ambient light levels. ( total darkness is a real challenge )

I'd check with James ( Original Poster ) and see what he has available. Looks like he's got a decent selection. ( In the USA we often have to find vendors which can ship us Dahua International models as local vendors often have a very limited selection of models to pick from - one vendor well like for this is in the vendor section here, Andy EmpireTecAndy )

 

[dcmkii]

Do you think IPC-HFW1831E could be a justifiable camera for low light and clarity for the day.
8mp with 1/1.8" sensor. Probably better than the 2831 I was looking at earlier. That one has 8mp with 1/2.7" sensor.

 

[samplenhold]

Unfortunately the IPC-HDW2831TMP-AS wasn't on the IPVM calculator

You can ask them to add any cameras that you are interested in. When you go to select a camera, down at the bottom right is a link. Select it to open a new browser tab and enter the info.

 

[aristobrat]

i would imagine higher fps at 25/30 would identify cars driving pass more clearly than 4k @ 15fps.

IMO, what usually helps the most when trying to identify something is the quality of the image in each frame, not the quantity of frames.

FPS controls the quantity of frames.
Exposure settings influence the image quality in each frame. Specifically, shutter speed plays a big role when it comes to image quality of something that's moving (i.e reduce motion blur).

If the image quality is good, you can often ID something with just one frame. On the other hand, if the image quality isn't good, getting an extra 10-15 frames every second probably won't be helpful.

Where FPS can come into play is when you play back a recording of a faster-moving object... If recorded from a camera set with a lower FPS setting, the object may appear to move jerkily across the scene. If recorded from a camera with a higher FPS setting, the object would move more smoothy. In either case, the image quality in each frame should be the same. If I had to guess, I'd say that most folks here run most of their cameras between 10-15 FPS.

 

[alastairstevenson]

This weekend,I will post the comparison images with people waling around.

Hi there @Jamestysecurity - will you be able to link to some comparison videos with movement?
I think forum members (and your customers ...) would be quite interested to see that.
It would differentiate the models so much better.

 

[c_snyder]

Do you think IPC-HFW1831E could be a justifiable camera for low light and clarity for the day.
8mp with 1/1.8" sensor. Probably better than the 2831 I was looking at earlier. That one has 8mp with 1/2.7" sensor.

wopi82 did some night comparisons in this thread, one camera being the 1831E against others. The video might help you decide. Personally based on my experience when it comes to 4MP vs 8MP, at night the 4MP wins in detail as the 8MP cameras become pretty blurry and dark.

Multiple camera low light test

Corona virus is spreading all over the world, and so we're all forced to stay home. Here in Poland all shopping centers, schools, kindergarten, playgrounds, etc. are closed. So, together with my wife, we have to find way, to do our regular job, and take care of children. It is time consuming and...

ipcamtalk.com