Dahua HFW4239T-ASE review

wopi82

IPCT Contributor
Joined
Feb 9, 2017
Messages
117
Reaction score
798
Location
Poland
Hi everyone. I’m pleased to present you a review of new product from Dahua, the HFW4239T-ASE. This camera was announced by Dahua in first week of may 2018. I’ve ordered it from Andy, who luckily had it already in stock for good price. Shipping was very fast, and took around 10 days.

This is the first model from Dahua with exceptionally bright f1.0 lens. Add to this an IMX290 sensor, and you get a nice set to fight with night darkness. Dahua rates this model sensitivity as 0.001 lux, which is even better than Ultra series model 8232 which is equipped with f1.5 lens and IMX185 sensor. The only Dahua model with better rating is 8835. But this is a different league.

foto.jpg

The camera is definitely bigger than turret style cams. Unfortunately Dahua did not release a turret version so far. The build quality is very good. All metal. At the bottom of the casing, there is a removable plate under which there is a microSD card slot. This is much more handy, compared to HDW4231 or 5831, where you need to disassemble whole camera and sometimes even the lens, to get to the card slot. On the other hand, such construction makes it easier for a theoretical intruder to get access to the card and steal/destroy it. All screws are star type. But getting a star screwdriver is not a problem today.

The first thing I wanted to check, was lens brightness. Before we go any further with testing, I will write a few word about lens brightness for those, unfamiliar with the terminology. Lens brightness tells us, how much light is able to go through the lens and hit the sensor. Lenses use an f-scale which is a sequence of the powers of the square root of 2. And so the scale is f1.0, f1.4, f2.0, f2.8, f4.0, f5.6, f8.0 and so on. The lower the f number, the more light can pass through lens. A f1.4 lens is twice as bright as f2.0 lens or in other words it passes twice more light. A f1.0 lens is twice brighter than f1.4 lens. And so, f1.0 lens is four time brighter than f2.0 lens. Brightness difference between every f-stop is also described as 1 exposition value – 1EV.

My older 4231 cam is f2.0 (also based on IMX290 sensor), the 4239 is f1.0. So there should be a 2EV difference between the two. To check it, I set my old 4231 to 1/25s exposure and 4239 to 4 times shorter 1/100 exposure. So now both cameras act as equipped with f2.0 lens, and we compare only gain scale.

4231 vs 4239 GAIN.jpg

As you can see, at gain 0 both cams give equally bright image. Assuming the gain at 0 is identical in both cams, the difference between lenses is indeed 2EV. However, as I increased the gain the 4239 become significantly brighter. So this was a right time to convert gain scale to ISO and check what’s going on there :D

4239T gain vs 6D.jpg

It turns out, Dahua has stretched the gain scale to a whopping ISO200000!!! All other cams I reviewed were reaching a maximum of ISO20000.

gain-iso-chart.jpg

So not only we get a 2 f-stops brighter lens, but also an additional over 3EVs of gain scale. All of this makes this camera over 5 f-stops brighter than 4231. In other words, this beast can produce an image around 40x brighter than 4231EM-AS and 20x brighter than 5231R-Z.

brightness-chart.jpg

At 1/100s exposure 4239 reaches 4231 brightness at around gain 60. Here is direct comparison of both cams:

max 4231 gain vs 4239.jpg

Despite both cams give same brightness (lenses are equalized by different exposure), the 4239 seems to give cleaner and crisper image. So another question was, whether the sensor noise is lower or is it better noise reduction?

To check it, I turned off noise reduction, switched both cams to intraframe MJPEG codec and maxed out bitrate, to better see noise pattern. Then I went through the whole gain scale of 4231 and tried to match 4239 brightness with its gain.

noise level test.jpg

Both cams have similar noise level. I would even say 4239 is a bit more noisy. Its noise pattern seems to be more bulky while 4231’s grain is finer. But the difference is subtle. And so, it’s the noise reduction, that’s doing the job. Dahua must have implemented a new denoising algorithm. After playing for a while with the settings I have a mixed feeling about this noise reduction. On the one hand it is more effective at higher noise levels. But on the other hand is seems less effective at lower gain values. Here is the same comparison as above, but with noise reduction turned on and with h264 codec.

noise level test nr50.jpg

Notice a shadowed area of the image, near a white car. While 4231 is clean until gain 70, and then it gradually becomes softer and more noisy, the 4239 starts to introduce some dark and bright specks already at gain 29 (equivalent of gain 50 for 4231). Gain 40 looks bad compared to 4231. What’s worse, these specks are static. This is not a glittering noise, but rather a texture that slowly morphs into another texture. Above gain 90 the image becomes significantly soft, loosing lot of detail. At that level it is better to keep noise reduction at lover values. Overall the noise pattern I described creates an unpleasant effect, when viewing the whole image. I had to tweak the settings a bit to get rid of it. Gamma is one of such settings to be careful about. It can introduce lot of noise in dark parts of the image. On the other hand if you set it low, you lose these dark parts.

Now, when I dropped the exposure to 1/25s and went through whole gain scale, it looked like this:

4239T 1-25s g0-100 small.jpg

It turned out, my test scene was just too bright :D

So I had to move to some other, darker place to test this gear at top gain. Here are two different scenes, seen at different settings.

dark-place-2.jpg

In first scene, the far part of the meadow is absolutely dark. There is no chance you can see anything with naked eye there. I was really impressed with the results. Might be wrong, but I don’t think 8232 could compete with this camera. Maybe it has a better quality image at lower gain but it will not reach the brightness of 4239. Honestly I don’t think any Hikvision’s DarkFighter can do this. I’ve seen DarkFigheterX demo on security tradeshow, and wasn’t that much impressed. The image was indeed better, but it was far from what I saw here.

Now, you’ve seen some pros, let me introduce some cons J

I was unpleasantly surprised to find out this camera has no ICR. I knew it has no IR LEDs, but on Dahua’s page it is clearly stated IT HAS a removable IR filter.

web screen.jpg

First time I switched to BW mode was during the day. I already become suspicious something is wrong.

4239T IR test 0.jpg

There was no sharpness loss. Also the leaves and plants on my neighbor’s yard did not turn bright – green plants reflect IR light, appearing white in IR mode. To make sure ICR is not working, I put my dashcam with IR LEDs in front of 4239 and 4231.

4239T IR test.jpg

I don’t really know if it is a mistake on Dahua’s web page, or my cam is broken. It would be great to have a working ICR on this camera. It would make it even better in dark scenes and allow usage of external IR lamps. The whole section of Web interface devoted to color/BW modes makes no sense. But it is still there.

The camera seems to have problems with white balance. It is often too much tinted blue. At night it totally gets rid of orange tint of street lamp, which is actually good. But I like to have just a bit of warmness left. With 4231 and 5831 I had no such problems. I was always happy with the color. It is of course a matter of taste, but for me the image is just to cool.

WDR also works slightly different than on 4231. As always it actives from level 45, but quickly becomes a wonderland scene. The good news is, it no longer introduces pixel shifting and edge distortion. I partly described this problem in 4431C-A review.

WDR test v2.jpg

The oval lens flare arround street lamp, you see on gain testing samples is there, same size and intensity, no matter where the light source is in the frame. I cannot get rid of it, unless the lamp is beyond the frame.

Lastly, when testing the camera in absolute darkness, I noticed some green light leaks, coming from the front side, near the lens. It seems a green led is inside and shining through the casing. I wonder why Dahua is putting anything shining into the camera. Was that really necessary? This might sound like no problem, but when you try to camouflage your gear, this might spoil the whole idea. Also at ISO200000 and 1/3s exposure, this faint light leak may act as strong torch :D The below images are done with long exposures, so don’t get too scared. It is not that bad. But still, with naked eye you can see the light, when looking at the front of the camera.

light leak.jpg

Here is a quick timelapse I did with 4239. Unfortunately Youtube compressed it badly. It’s only 720p because I used screen capture on my laptop rather than SD card. Thus I was limited by screen resolution. The camera was set to gain 0-100 and exposure 0-333ms.

 

EMPIRETECANDY

IPCT Vendor
Joined
Nov 8, 2016
Messages
8,259
Reaction score
23,718
Location
HONGKONG
Hi everyone. I’m pleased to present you a review of new product from Dahua, the HFW4239T-ASE. This camera was announced by Dahua in first week of may 2018. I’ve ordered it from Andy, who luckily had it already in stock for good price. Shipping was very fast, and took around 10 days.

This is the first model from Dahua with exceptionally bright f1.0 lens. Add to this an IMX290 sensor, and you get a nice set to fight with night darkness. Dahua rates this model sensitivity as 0.001 lux, which is even better than Ultra series model 8232 which is equipped with f1.5 lens and IMX185 sensor. The only Dahua model with better rating is 8835. But this is a different league.

View attachment 31471

The camera is definitely bigger than turret style cams. Unfortunately Dahua did not release a turret version so far. The build quality is very good. All metal. At the bottom of the casing, there is a removable plate under which there is a microSD card slot. This is much more handy, compared to HDW4231 or 5831, where you need to disassemble whole camera and sometimes even the lens, to get to the card slot. On the other hand, such construction makes it easier for a theoretical intruder to get access to the card and steal/destroy it. All screws are star type. But getting a star screwdriver is not a problem today.

The first thing I wanted to check, was lens brightness. Before we go any further with testing, I will write a few word about lens brightness for those, unfamiliar with the terminology. Lens brightness tells us, how much light is able to go through the lens and hit the sensor. Lenses use an f-scale which is a sequence of the powers of the square root of 2. And so the scale is f1.0, f1.4, f2.0, f2.8, f4.0, f5.6, f8.0 and so on. The lower the f number, the more light can pass through lens. A f1.4 lens is twice as bright as f2.0 lens or in other words it passes twice more light. A f1.0 lens is twice brighter than f1.4 lens. And so, f1.0 lens is four time brighter than f2.0 lens. Brightness difference between every f-stop is also described as 1 exposition value – 1EV.

My older 4231 cam is f2.0 (also based on IMX290 sensor), the 4239 is f1.0. So there should be a 2EV difference between the two. To check it, I set my old 4231 to 1/25s exposure and 4239 to 4 times shorter 1/100 exposure. So now both cameras act as equipped with f2.0 lens, and we compare only gain scale.

View attachment 31464

As you can see, at gain 0 both cams give equally bright image. Assuming the gain at 0 is identical in both cams, the difference between lenses is indeed 2EV. However, as I increased the gain the 4239 become significantly brighter. So this was a right time to convert gain scale to ISO and check what’s going on there :D

View attachment 31466

It turns out, Dahua has stretched the gain scale to a whopping ISO200000!!! All other cams I reviewed were reaching a maximum of ISO20000.

View attachment 31472

So not only we get a 2 f-stops brighter lens, but also an additional over 3EVs of gain scale. All of this makes this camera over 5 f-stops brighter than 4231. In other words, this beast can produce an image around 40x brighter than 4231EM-AS and 20x brighter than 5231R-Z.

View attachment 31469

At 1/100s exposure 4239 reaches 4231 brightness at around gain 60. Here is direct comparison of both cams:

View attachment 31474

Despite both cams give same brightness (lenses are equalized by different exposure), the 4239 seems to give cleaner and crisper image. So another question was, whether the sensor noise is lower or is it better noise reduction?

To check it, I turned off noise reduction, switched both cams to intraframe MJPEG codec and maxed out bitrate, to better see noise pattern. Then I went through the whole gain scale of 4231 and tried to match 4239 brightness with its gain.

View attachment 31476

Both cams have similar noise level. I would even say 4239 is a bit more noisy. Its noise pattern seems to be more bulky while 4231’s grain is finer. But the difference is subtle. And so, it’s the noise reduction, that’s doing the job. Dahua must have implemented a new denoising algorithm. After playing for a while with the settings I have a mixed feeling about this noise reduction. On the one hand it is more effective at higher noise levels. But on the other hand is seems less effective at lower gain values. Here is the same comparison as above, but with noise reduction turned on and with h264 codec.

View attachment 31475

Notice a shadowed area of the image, near a white car. While 4231 is clean until gain 70, and then it gradually becomes softer and more noisy, the 4239 starts to introduce some dark and bright specks already at gain 29 (equivalent of gain 50 for 4231). Gain 40 looks bad compared to 4231. What’s worse, these specks are static. This is not a glittering noise, but rather a texture that slowly morphs into another texture. Above gain 90 the image becomes significantly soft, loosing lot of detail. At that level it is better to keep noise reduction at lover values. Overall the noise pattern I described creates an unpleasant effect, when viewing the whole image. I had to tweak the settings a bit to get rid of it. Gamma is one of such settings to be careful about. It can introduce lot of noise in dark parts of the image. On the other hand if you set it low, you lose these dark parts.

Now, when I dropped the exposure to 1/25s and went through whole gain scale, it looked like this:

View attachment 31465

It turned out, my test scene was just too bright :D

So I had to move to some other, darker place to test this gear at top gain. Here are two different scenes, seen at different settings.

View attachment 31470

In first scene, the far part of the meadow is absolutely dark. There is no chance you can see anything with naked eye there. I was really impressed with the results. Might be wrong, but I don’t think 8232 could compete with this camera. Maybe it has a better quality image at lower gain but it will not reach the brightness of 4239. Honestly I don’t think any Hikvision’s DarkFighter can do this. I’ve seen DarkFigheterX demo on security tradeshow, and wasn’t that much impressed. The image was indeed better, but it was far from what I saw here.

Now, you’ve seen some pros, let me introduce some cons J

I was unpleasantly surprised to find out this camera has no ICR. I knew it has no IR LEDs, but on Dahua’s page it is clearly stated IT HAS a removable IR filter.

View attachment 31477

First time I switched to BW mode was during the day. I already become suspicious something is wrong.

View attachment 31467

There was no sharpness loss. Also the leaves and plants on my neighbor’s yard did not turn bright – green plants reflect IR light, appearing white in IR mode. To make sure ICR is not working, I put my dashcam with IR LEDs in front of 4239 and 4231.

View attachment 31468

I don’t really know if it is a mistake on Dahua’s web page, or my cam is broken. It would be great to have a working ICR on this camera. It would make it even better in dark scenes and allow usage of external IR lamps. The whole section of Web interface devoted to color/BW modes makes no sense. But it is still there.

The camera seems to have problems with white balance. It is often too much tinted blue. At night it totally gets rid of orange tint of street lamp, which is actually good. But I like to have just a bit of warmness left. With 4231 and 5831 I had no such problems. I was always happy with the color. It is of course a matter of taste, but for me the image is just to cool.

WDR also works slightly different than on 4231. As always it actives from level 45, but quickly becomes a wonderland scene. The good news is, it no longer introduces pixel shifting and edge distortion. I partly described this problem in 4431C-A review.

View attachment 31478

The oval lens flare arround street lamp, you see on gain testing samples is there, same size and intensity, no matter where the light source is in the frame. I cannot get rid of it, unless the lamp is beyond the frame.

Lastly, when testing the camera in absolute darkness, I noticed some green light leaks, coming from the front side, near the lens. It seems a green led is inside and shining through the casing. I wonder why Dahua is putting anything shining into the camera. Was that really necessary? This might sound like no problem, but when you try to camouflage your gear, this might spoil the whole idea. Also at ISO200000 and 1/3s exposure, this faint light leak may act as strong torch :D The below images are done with long exposures, so don’t get too scared. It is not that bad. But still, with naked eye you can see the light, when looking at the front of the camera.

View attachment 31473

Here is a quick timelapse I did with 4239. Unfortunately Youtube compressed it badly. It’s only 720p because I used screen capture on my laptop rather than SD card. Thus I was limited by screen resolution. The camera was set to gain 0-100 and exposure 0-333ms.

Thanks for the fantastic testing @wopi82 use good tools to show the cams working, for the light green I test it, not very easy to watch. I take a pic via iPhoneX.

upload_2018-7-17_22-3-31.png

Here is the pic for the SD1A203T-GN i tested months ago.
upload_2018-7-17_22-4-37.png

upload_2018-7-17_22-6-29.png
Here are the IPC-HFW4239T-ASE i tested right now.

upload_2018-7-17_22-5-29.png
upload_2018-7-17_22-6-0.png
 

Korgoth

Pulling my weight
Joined
Aug 28, 2015
Messages
103
Reaction score
160
Hi everyone. I’m pleased to present you a review of new product from Dahua, the HFW4239T-ASE. This camera was announced by Dahua in first week of may 2018. I’ve ordered it from Andy, who luckily had it already in stock for good price. Shipping was very fast, and took around 10 days.
WOW! Thats is some nice review. I prefer snapshots with movement on them, because static picture looks always nice. But tstill, this is helpful :)
 

aristobrat

IPCT Contributor
Joined
Dec 5, 2016
Messages
2,982
Reaction score
3,180
Before we go any further with testing, I will write a few word about lens brightness for those, unfamiliar with the terminology.
Thanks for this! I'm still trying to understand the terminology better and explanations like this help a lot.

In other words, this beast can produce an image around 40x brighter than 4231EM-AS and 20x brighter than 5231R-Z.
For a newbie (to the technical side) like me, it seems crazy that a new lens can make an image 20x-40x brighter... that's pretty awesome!

The part I'm trying to understand is if a color image with 20-40x more brightness likely to be a more useful image (in terms of clarity) than a b/w image from a 4231/5231, especially if there's motion happening that can be subject to blur (like a person walking by).
 

tigerwillow1

Known around here
Joined
Jul 18, 2016
Messages
3,815
Reaction score
8,424
Location
USA, Oregon
Thanks for the great review. I want to expand on one point that's important to me and I suspect others:

- The spec sheet appears to state that the IR cut filter is removed in b/w mode
- This is incorrect as the IR cut filter is permanently in the incoming light path
- This means that using the camera in total darkness with an external IR light is infeasible

From looking at the sample pictures, it looks to me like there's no purpose whatsoever for the b/w mode. I wonder how powerful a visible LED light would have to be to get a good image in otherwise total darkness?
 

wopi82

IPCT Contributor
Joined
Feb 9, 2017
Messages
117
Reaction score
798
Location
Poland
I've just contacted Dahua Poland. They wrote, this is a mistake on a webpage. There is indeed no ICR in this model. They also told me, there will be another model of this cam, equipped with white LEDs.
 

Korgoth

Pulling my weight
Joined
Aug 28, 2015
Messages
103
Reaction score
160
Thanks for the great review. I want to expand on one point that's important to me and I suspect others:

- The spec sheet appears to state that the IR cut filter is removed in b/w mode
- This is incorrect as the IR cut filter is permanently in the incoming light path
- This means that using the camera in total darkness with an external IR light is infeasible

From looking at the sample pictures, it looks to me like there's no purpose whatsoever for the b/w mode. I wonder how powerful a visible LED light would have to be to get a good image in otherwise total darkness?
EDIT:
I WAS WRONG. The thing I wrote below are corrected by @wopi82 later :)



You would see difference in b/w mode with movement on scene. This is how sensor works.
On sensor you have 3 color sensors RGB. Each one is collecting photons of only one color, ignoring others, to simplify let's say 66% of other photons
When camera switches to B/W sensor is collecting all photons, ignoring their wavelength, and only getting their energy.
That's why in B/W it should be more sensitive, and you should have less motion blur and noise.

Thas of course how it works in theory ;)

Wysłane z mojego SM-G935F przy użyciu Tapatalka
 
Last edited:

LoLight

Getting the hang of it
Joined
Jul 9, 2018
Messages
25
Reaction score
32
Location
WA
Thank you wopi82!
The camera is such an impressive low-light color performer, especially considering the price!
Really appreciate your reviews - they're so painstakingly detailed!
btw, do you have any tips for this model settings?
It will be run at night in a fairly well-lit (street light + porch lights) environment.
 

tangent

IPCT Contributor
Joined
May 12, 2016
Messages
4,339
Reaction score
3,521
OT, but has anybody played around with one of the RGB-IR sensors that doesn't need an ir-cut filter?
 

wopi82

IPCT Contributor
Joined
Feb 9, 2017
Messages
117
Reaction score
798
Location
Poland
@Korgoth

You are right, I should present some moving objects. I’ll give it a go when I’ll have some spare time.

However I cannot agree with what you’ve wrote about sensor sensitivity. It is not true, the sensor becomes more sensitive when switching camera to BW mode. A color sensor has a color filter array implemented over photodiodes. This array is there no matter if you switch to BW or not. And so, each photodiode is still collecting only part of the spectrum, even in BW mode. The sensitivity rises only when IR filter is removed, because than more information is available to the sensor. Now, not only visible light is reaching the sensor, but also infrared light. Color filter array is not blocking IR light. The image is still in color. After IR filter removal the colors are incorrect and the whole image is tinted purple. Then the camera changes the image to BW (by software) to get rid of these false colors. A different story is monochrome sensor. Such sensor has no color filter array, so indeed it can collect more light, because a full visible spectrum is hitting each photodiode. In case of 4239, switching to BW doesn't do anything. You just have no colors. The noise level and image brightness remains the same.

And by the way, pozdrawiam kolegę z Polski :)
 
Last edited:

Korgoth

Pulling my weight
Joined
Aug 28, 2015
Messages
103
Reaction score
160
@Korgoth

You are right, I should present some moving objects. I’ll give it a go when I’ll have some spare time.

However I cannot agree with what you’ve wrote about sensor sensitivity. It is not true, the sensor becomes more sensitive when switching camera to BW mode. A color sensor has a color filter array implemented over photodiodes. This array is there no matter if you switch to BW or not. And so, each photodiode is still collecting only part of the spectrum, even in BW mode. The sensitivity rises only when IR filter is removed, because than more information is available to the sensor. Now, not only visible light is reaching the sensor, but also infrared light. Color filter array is not blocking IR light. The image is still in color. After IR filter removal the colors are incorrect and the whole image is tinted purple. Then the camera changes the image to BW (by software) to get rid of these false colors. A different story is monochrome sensor. Such sensor has no color filter array, so indeed it can collect more light, because a full visible spectrum is hitting each photodiode. In case of 4239, switching to BW doesn't do anything. You just have no colors. The noise level and image brightness remains the same.

And by the way, pozdrawiam kolegę z Polski :)
I think both of us can be right, just talked with R&D of our Machine Vision solution provider, and he said not all cameras work like I said. To be precise, only some specified sensor can do it.

So, when we discuss standard cameras, I'm wrong and you are right, ;)
Thanks for correction.I edited my previous post

Also pozdrawiam ;)

Wysłane z mojego SM-G935F przy użyciu Tapatalka
 
Last edited:

LoLight

Getting the hang of it
Joined
Jul 9, 2018
Messages
25
Reaction score
32
Location
WA
Another wow! A truly great presentation.
Could you please confirm that the camera is based on the IMX290 sensor?
 
Top