Dahua WizColor 5x59-PRO and SmartLight 5x59-IL new series

[alekk]

For completeness, here's the Histograms for those two sets of images above.

There's a noticeable right-shift in the (top) 1509 pics ... whereas the (lower) 1656 is much closer.

I'm not sure this really tells 'ya much (except that the metering was brighter in the earlier 1509 which had more sunlight) ... since the scene is different (bummer I don't have 'em side-by-side) ... so the "Mark I Eyeball" is probably a better judge.

 

[flynreelow]

Yes, ZE is vari-focal ... both set at zoom=550 (on a scale of 0 to ~2,500) ... but as noted before, the focus numbers are not that close ... about 1640 & 1740 ... so make a slight difference in FOV. And yes, it's an overview camera as shown.

Here's a comparison a little more than an hour before sunset. Looking at the histogram in Photoshop, the metering is much closer. I set the White Balance to Auto because it was also closer ... albeit still some difference and maybe some saturation boost in the 1022 Firmware ... which I think is a more pleasing picture. Everything set identical - 5 msec shutter, 100 Iris, ZERO Gain/NR.

View attachment 230824

can u post these 2 pictures separate as well as the animated gif you have.

 

[alekk]

Here 'ya go - PNG's straight from the web interface - had to resave as JPEG's because the 2nd PNG would upload, but not "take" ... for reasons unknown, that seems to happen sometimes - they are almost 10MBytes!

 

[EMPIRETECANDY]

From Alekk's pic compare the new one now not such blue and get sharper on grass and the street trees not too white, pic is sharper, watch on the grass.

 

[alekk]

Now that you mention it @EMPIRETECANDY, there is a blue shift in the histogram - the 1022FW shifts to the left compared to 0827FW.

Again, the scene is different (bummer I don't have 'em side-by-side anymore) ... so it's not directly comparable.

 

[Arjun]

From Alekk's pic compare the new one now not such blue and get sharper on grass and the street trees not too white, pic is sharper, watch on the grass.

Ah yes...the grass is greener on the other side

 

[alekk]

@EMPIRETECANDY : Minor suggested documentation update and/or change in setting behavior.

So during my daytime tests, I have set the NR (Noise Reduction) to 0 ... AND ... toggled 3D NR OFF ... which seemed redundant since the level is to set to zero anyway. Turns out that doesn't appear to be the case!

This is easily seen in a semi-static scene with both the 0827 and 1022 firmware. Set 2D/3D NR to ZERO and toggle it off. Then use some exposure settings (with lots of gain) that generates a fair amount of noise in the image.

Then toggle "3D NR" to ON ... but leave it as zero - the image cleans up quite a bit!

And THEN you can move the slider from zero to 100 there is more "improvement" ... obviously the tradeoff is that you are blurring any motion.


After saying WTF, I went to check the documentation ... the latest I could find was - Web 5.0 Operations Manual 1.2.5 - revision history says released August/2025.

On page 20, it talks about the various exposure settings. While it has separate "3D NR" and "levels" explanation boxes, to me it's not clear that enabling 3D NR (but setting levels to ZERO) actually "does something" .. which I found surprising.

BTW, that is the only place that "3D NR" is talked about ... and "2D NR" is not mentioned anywhere in the document.

I think it would be worthwhile to describe the difference - I assume 2D works on a per-image basis (classic NR such as Photoshop does) whereas 3D NR operates with multiple frames ... which it does say.

And it seems like the "3D NR" Button is redundant ... if you have set it to ZERO, it should simply skip that section of code ... which I think is the desired behavior ... rather than whatever it's doing now.

 

[steve1225]

@EMPIRETECANDY : Minor suggested documentation update and/or change in setting behavior.

So during my daytime tests, I have set the NR (Noise Reduction) to 0 ... AND ... toggled 3D NR OFF ... which seemed redundant since the level is to set to zero anyway. Turns out that doesn't appear to be the case!

This is easily seen in a semi-static scene with both the 0827 and 1022 firmware. Set 2D/3D NR to ZERO and toggle it off. Then use some exposure settings (with lots of gain) that generates a fair amount of noise in the image.

Then toggle "3D NR" to ON ... but leave it as zero - the image cleans up quite a bit!

And THEN you can move the slider from zero to 100 there is more "improvement" ... obviously the tradeoff is that you are blurring any motion.


After saying WTF, I went to check the documentation ... the latest I could find was - Web 5.0 Operations Manual 1.2.5 - revision history says released August/2025.

On page 20, it talks about the various exposure settings. While it has separate "3D NR" and "levels" explanation boxes, to me it's not clear that enabling 3D NR (but setting levels to ZERO) actually "does something" .. which I found surprising.

BTW, that is the only place that "3D NR" is talked about ... and "2D NR" is not mentioned anywhere in the document.

I think it would be worthwhile to describe the difference - I assume 2D works on a per-image basis (classic NR such as Photoshop does) whereas 3D NR operates with multiple frames ... which it does say.

And it seems like the "3D NR" Button is redundant ... if you have set it to ZERO, it should simply skip that section of code ... which I think is the desired behavior ... rather than whatever it's doing now.

Are you saying that all the complaints about "not turning off" 3D NR were due to setting 3D NR to a low level (zero) and not turning off 3D NR using this switch?

if setting 3D NR to zero would disable all 3D, then there will be no separate 3D NR switch..

it has separate functionality - always had - disabling all NR video processing from video pipeline. It has wrong label - it should be simply "Noise Reduction"...

2D/3D NR is most basic functions done in video processing pipeline done in early stages. It has humungous importance for video quality.

In case of WizColor cams - this is done using much more powerful 2D NR & 3D NR AI networks - which use different NR algorithms to different type of "surfaces"..

 

[CamCrazy]

I have one of my 54Pro's on the new 1022 firmware now, it is in a 'normal position' capturing people only. Will report my findings, for now though the 'freeze motion' situation is the same, which I think was not targeted in this update. Hence I am requiring shutter below 2ms to properly freeze a person moving at slow to medium speed.

 

[StevenFromTexas]

Reading these posts makes one wonder if flying a commercial passenger jet is easier than dealing with settings on these new cameras.

 

[CamCrazy]

Question, I switched both my installed cameras to 1022, if I also try one of them with the Dahua 0827 firmware, assume I can still upgrade or downgrade to 1022 and/or 731 ?

What are thoughts on the 827 firmware, I noted a couple of people using it, mentions of Ai being better. I am concentrating on sharpness and stop motion currently.

 

[bigredfish]

Question, I switched both my installed cameras to 1022, if I also try one of them with the Dahua 0827 firmware, assume I can still upgrade or downgrade to 1022 and/or 731 ?

What are thoughts on the 827 firmware, I noted a couple of people using it, mentions of Ai being better. I am concentrating on sharpness and stop motion currently.

I may try the newest version today.

The two areas I’m looking to improve are the softness and IVS detection (missed a rather large possum last night) , but could easily (more likely) have been my rule placement and sensitivity settings.

I’m so far impressed with the enhanced light pickup (3.6 fixed AS) which is really the key selling point of this vs the 5442*, and it seems to have reduced digital noise of the 3449 Tioc Pro. Thus the softness we’re seeing. Tradeoff.
(* I still think the 5442 has better image quality but it’s not as much of a difference as first thought)

Re “stop motion” what we commonly call “freezing” a video to get a snapshot without motion blur. This is the true test of surveillance cameras vs still 35mm photography. Our subjects rarely pose for the camera and we have smaller sensors.
Are you talking night or day? People walking, jogging? Vehicles at 15mph or 50mph? A bird flying?
All very different settings with different requirements, and speed.

During bright daylight it’s relatively easy with this or any camera with manual exposure settings … in fact most any $50+ camera can do so and produce an adequate daytime frozen image. Light can fix most any scene.

Night/Low light, particularly under changing lighting conditions, shadows, is where we separate the wheat from the chaff. Balancing motion blur with noise (or over-soft lack thereof) is the challenge.

The other thing those new to modern video surveillance cameras should understand is distance and speed.

Take a 3.6mm fixed or barely zoomed varifocal, with a moving object say a human or car passing at 5-10ft vs 100+ft
It’s much harder under low light to get a crisp clean, non-noisy frozen still from the moving close in object than the one far away at the same speed. (Light intensity and direction being equal) It’s largely perception as the far away object is much smaller, has far fewer pixels, and lacks detail.

We see camera manufacturers use this trick in marketing videos, often a slow pan of a busy Tokyo skyline from 1000 yards. Lots of an activity, cars, bright lights, colors, we think oh wow what a great clean image. But it lacks fine detail, it’s not an ID shot for the police, it’s a pretty picture.

 

[CamCrazy]

Re “stop motion” what we commonly call “freezing” a video to get a snapshot without motion blur. This is the true test of surveillance cameras vs still 35mm photography. Our subjects rarely pose for the camera and we have smaller sensors.
Are you talking night or day? People walking, jogging? Vehicles at 15mph or 50mph? A bird flying?
All very different settings with different requirements, and speed.

During bright daylight it’s relatively easy with this or any camera with manual exposure settings … in fact most any $50+ camera can do so and produce an adequate daytime frozen image. Light can fix most any scene.

Thanks for responding, well as of yesterday I now have another 54Pro covering people walking at slow speed, distance 15 feet on full zoom (ZE), my second is covering vehicles moving cross scene at 25-30mph. Normally I totally agree that most cameras in the past will guarantee stop motion at sub 4-6ms, sometimes lower depending on conditions. For the 54Pro in my location I need them on sub 1ms, yes, I even have one on 0.7ms. This is unheard of and I have personally used 20+ ip cams over the last 8-10 years (numerous analogue prior to that) and setup many others. I know that normally with photography a 200mm lens would freeze motion assuming all other things equal above 1/200 shutter speed, this theory doesn't really translate to ip cams in the same way, and frankly with photography I would normally run 1/400 on a 200mm lens anyway.

Keeping exposure compensation down to 25 and gamma at 41 is helping with both quality and freeze frame, similarly sharpness down at 15, NR is off altogether at this point. Will give the new firmware some time now and see how the IR/colour looks overnight, willing this thing onwards Yes I am picky with image quality, very, but why not!

The softness as such doesn't worry me as much as the freeze frame struggle, yes of course sharpness is nice, but freeze frame is king. Getting a bit frustrated with things at this point, I guess having the 4K-T and Z4E-S3 (even the older 5442's) as benchmarks is not helping either. I want this 54Pro to match the S3 with the bonus of white LED, odd though because it does have moments of brilliance in low light colour. Still on the plus side, the 4K-T is hugely impressive, followed by the S3, light is often against cameras in this part of the world and those two work very well.

For now I have mixed feelings. Your post is excellent though with some great wisdom

 

[StevenFromTexas]

I guess having the 4K-T and Z4E-S3 (even the older 5442's) as benchmarks is not helping either. I want this 54Pro to match the S3 with the bonus of white LED, odd though because it does have moments of brilliance in low light colour. Still on the plus side, the 4K-T is hugely impressive, followed by the S3, light is often against cameras in this part of the world and those two work very well.

Maybe someone from Dahua needs to read this part of your post?

 

[alekk]

@CamCrazy : I wonder if maybe the difficulty you are having in freeze motion is related to my observation (which I didn't realize) that even if you have the 3D/2D NR levels set to zero, if you have it toggled, you are still getting some NR - as @steve1225's said "In case of WizColor cams - this is done using much more powerful 2D NR & 3D NR AI networks - which use different NR algorithms to different type of "surfaces" ... so even if the camera is capturing a sharp image, that NR is smearing it up by time it gets to you(?)

If you haven't already, maybe try this and see if it makes a difference specifically WRT freezing motion.


Yea, ditto what @bigredfish said - as in the photography world, it's all about the light.
BTW, sometimes our subjects don't pose for the camera - this (small!) bird was taken with a shutter speed of 1/8000s.

 

[bigredfish]

Thanks for responding, well as of yesterday I now have another 54Pro covering people walking at slow speed, distance 15 feet on full zoom (ZE), my second is covering vehicles moving cross scene at 25-30mph. Normally I totally agree that most cameras in the past will guarantee stop motion at sub 4-6ms, sometimes lower depending on conditions. For the 54Pro in my location I need them on sub 1ms, yes, I even have one on 0.7ms. This is unheard of and I have personally used 20+ ip cams over the last 8-10 years (numerous analogue prior to that) and setup many others. I know that normally with photography a 200mm lens would freeze motion assuming all other things equal above 1/200 shutter speed, this theory doesn't really translate to ip cams in the same way, and frankly with photography I would normally run 1/400 on a 200mm lens anyway.

Keeping exposure compensation down to 25 and gamma at 41 is helping with both quality and freeze frame, similarly sharpness down at 15, NR is off altogether at this point. Will give the new firmware some time now and see how the IR/colour looks overnight, willing this thing onwards Yes I am picky with image quality, very, but why not!

The softness as such doesn't worry me as much as the freeze frame struggle, yes of course sharpness is nice, but freeze frame is king. Getting a bit frustrated with things at this point, I guess having the 4K-T and Z4E-S3 (even the older 5442's) as benchmarks is not helping either. I want this 54Pro to match the S3 with the bonus of white LED, odd though because it does have moments of brilliance in low light colour. Still on the plus side, the 4K-T is hugely impressive, followed by the S3, light is often against cameras in this part of the world and those two work very well.

For now I have mixed feelings. Your post is excellent though with some great wisdom

Need pics

 

[StevenFromTexas]

Need pics

Yes! If there are no pics, then it didn't happen.

 

[Chokolinho]

What do you use for white balance?

 

[steve1225]

@CamCrazy : I wonder if maybe the difficulty you are having in freeze motion is related to my observation (which I didn't realize) that even if you have the 3D/2D NR levels set to zero, if you have it toggled, you are still getting some NR - as @steve1225's said "In case of WizColor cams - this is done using much more powerful 2D NR & 3D NR AI networks - which use different NR algorithms to different type of "surfaces" ... so even if the camera is capturing a sharp image, that NR is smearing it up by time it gets to you(?)

On my 54PRO, when I turn off the 3D NR switch, I see an image without 3D noise reduction—with the typical defects of an unprocessed image—noise that forms a characteristic pattern typical of this sensor.

Of course, I'm talking about a very short shutter speed (0.1 ms during the day) and high gain.

So, I don't know what you're talking about @alekk,

 

[Chokolinho]

On my 54PRO, when I turn off the 3D NR switch, I see an image without 3D noise reduction—with the typical defects of an unprocessed image—noise that forms a characteristic pattern typical of this sensor.

Of course, I'm talking about a very short shutter speed (0.1 ms during the day) and high gain.

So, I don't know what you're talking about @alekk,

How would you roughly adjust the 54pro?