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

[Darrin Morrow]

I think the close focus on the vari-focal is supposed to be even better

 

[bigredfish]

Yes much
I may just hang it in the same spot to compare. By spec it won’t be as good at light pickup but given the scene it may not matter.

 

[alekk]

Out of interest, do you plan to run these in colour overnight?

So here's the NIGHT comparison of Firmware 0827 & 1022 on the T54PRO-ZE at zoom level 550 - that scale goes from 0-2426.

First, this is NOT A "FAIR" SITUATION for the camera ... since the settings are 80msec shutter, 100 iris, 65 gain, 50 3D NR, and 50 2DNR. So clearly the light level is waayyy to low to use color (without adding some lights) to capture any motion. BTW, be very careful with the gain - this scene falls apart real fast when you go about 65 gain.

I don't see much difference - again, the the histogram shows the blue channel is reduced slightly (I compare using a 1920x1080 cutout from the middle, so the overhanging roof on the top camera doesn't show) but I don't have a clear preference ... LOL they are both BAD!

The 1022 seems to be doing a bit more "smearing" (see the road and driveway) ... but this is such a mushy image, I'd be careful judging it as such.

BTW, I was kinda surprised (in THIS low light situation looking at STATIC images) that changing the level of NR (both 2D and 3D) didn't make that much difference, although it did reduce the "dancing pixels" in the video. BTW, if you want to see what the NR TOGGLE does by itself, try a low light scene like this, set the LEVELS of NR to ZERO for both ... and then toggle the NR which causes a BIG difference.

To echo how dark this is, it looks like my neighbor across the street turned on their patio lights ... which come out really bright!


In summary, I don't see a big difference (in this situation!) between the 0827 & 1022 firmware - but I would not expect the Dahua engineers to test such an extreme case, but it perhaps gives some idea of the very outer limits of the current sensor/lens/processing.

BTW, in comparison to the 5442-S3 (took that down to put up the 2nd T54), that would have to run at least TWICE as much shutter exposure (160msec!) and it's a pretty dull image without the colors here. So I consider that good progress ... albeit this is static image testing. And I look forward to testing the next generation sensor (hopefully 1/1.2" ... or even 4/3") paired with an F/1.0 lens! ;-)


And before someone says it, sure, I could put up a junky Reolink with a 1/3" sensor & crappy lens and if it allowed me to do a long enough exposure, I would get a similar STATIC picture. But it would need to be much, much, much slower than what the T54 is doing here.

 

[CamCrazy]

I'm done for now tuning on the 5459 AS.

My final NIGHT settings for now. Pretty close to my initial dial in.
I just can't live with the digital fuzzy textures AT NIGHT on moving people that come with this and the 3449 Pro once I get to 03ms
So for now 4ms is good and less distracting. There's not enough motion blur to make a difference, you have to look pretty hard at close in targets.

Remember I'm using this and the 3449 Pro at entry doors for face ID at the doors a priority, in enclosed spaces.

If you're using it as a yard cam in open area, and won't routinely have human targets within 6-8 feet of the camera, you may not notice or see the noise. It likely won't be an issue to you.
I don't see running this at 1ms or 2ms unless you're just using it as a wide overview up high, at distance, without the onboard lights, or in IR mode where you wont notice the digital noise. It's there. Especially at about 10ft on in.

The 5442 S3 is still a better quality image to the discerning eye IMHO, but we know it takes much more light to run it in color. Its not a deal-breaker.

View attachment 230891 View attachment 230892 View attachment 230893

Great write up, I would say very similar findings for mine. Once the darkness falls I have been running one which is also sharing space and LED onboard with a 4K-T, in colour I want it at 0-2ms for freeze frame but it is starting to get a bit dark. The 54Pro running solo with no ambient can still ID vehicles even at 0-2ms shutter. I have taken to turning sharpening to 0 and NR toggled off at night now on these, I will live with some noise for better motion and more natural look.

Where I have a 54Pro and 4K-T covering the same area from two different angles, it is tough to call at night, even with a 4000lumen LED flood for motion backup. The big benefit with the Pro series is having IR as fallback. I feel like overnight I have seen small gains with the new firmware, early days yet.

 

[CamCrazy]

So here's the NIGHT comparison of Firmware 0827 & 1022 on the T54PRO-ZE at zoom level 550 - that scale goes from 0-2426.

First, this is NOT A "FAIR" SITUATION for the camera ... since the settings are 80msec shutter, 100 iris, 65 gain, 50 3D NR, and 50 2DNR. So clearly the light level is waayyy to low to use color (without adding some lights) to capture any motion. BTW, be very careful with the gain - this scene falls apart real fast when you go about 65 gain.

I don't see much difference - again, the the histogram shows the blue channel is reduced slightly (I compare using a 1920x1080 cutout from the middle, so the overhanging roof on the top camera doesn't show) but I don't have a clear preference ... LOL they are both BAD!

The 1022 seems to be doing a bit more "smearing" (see the road and driveway) ... but this is such a mushy image, I'd be careful judging it as such.

BTW, I was kinda surprised (in THIS low light situation looking at STATIC images) that changing the level of NR (both 2D and 3D) didn't make that much difference, although it did reduce the "dancing pixels" in the video. BTW, if you want to see what the NR TOGGLE does by itself, try a low light scene like this, set the LEVELS of NR to ZERO for both ... and then toggle the NR which causes a BIG difference.

To echo how dark this is, it looks like my neighbor across the street turned on their patio lights ... which come out really bright!


In summary, I don't see a big difference (in this situation!) between the 0827 & 1022 firmware - but I would not expect the Dahua engineers to test such an extreme case, but it perhaps gives some idea of the very outer limits of the current sensor/lens/processing.

BTW, in comparison to the 5442-S3 (took that down to put up the 2nd T54), that would have to run at least TWICE as much shutter exposure (160msec!) and it's a pretty dull image without the colors here. So I consider that good progress ... albeit this is static image testing. And I look forward to testing the next generation sensor (hopefully 1/1.2" ... or even 4/3") paired with an F/1.0 lens! ;-)


And before someone says it, sure, I could put up a junky Reolink with a 1/3" sensor & crappy lens and if it allowed me to do a long enough exposure, I would get a similar STATIC picture. But it would need to be much, much, much slower than what the T54 is doing here.

Very close call between the two images at night, maybe the 0827 looks a bit sharper on some of the trees and grass, so close though.

Would like to see those images with no NR and no sharpening, shutter at 0-8ms, just out of interest. Have you tried a 4K-T out of curiosity, sorry, that is for the other thread really.

 

[EMPIRETECANDY]

I dont notice the close focus so much until about 3ft, I dont think most would even think about it.
I have found many/most of their close focus specs are 2x the distance what real world sees
Yes less noisy than the 3449. On both its a fine digital "fuzz" not the normal pixelated noise we think of on the 5442.

I'm still running the older "soft focus" fw, so while a little soft again its not a deal-breaker for me so far at 4ms


View attachment 230896

Yes, the soft pics need some more work on HDR. This is the very important thing for the camera.

 

[alekk]

Here's some morning pictures comparing the 0827 and 1022 firmware.
Settings are identical at 5msc shutter, 100 iris, 0 gain, ZERO NR, Auto WB.

Note there is a little more "speckling/dimpling" in the street/driveway with the 1022 firmware (bit more smearing in the 0827) ... which is "good" because that exists ... and I recall the 5442-S3 picks this up even better ... but I'll put that back up in a couple of weeks.

In terms of sharpness, the 0827 seems a bit sharper on some of the grass/trees ... but this could be due to the slightly different slant angle, which seems to make more difference that you'd think - I actually thought my first T54PRO when it was mounted below the 5442 ... so I think I'm seeing the same thing with the lower-mounted 1022 firmware camera. It does appear that the 1022 has less jaggies on things such as the white mailbox and then sign to the left of it.

In the later shot (08:46), I like how the 1022 handles the highlights better - see leaves in trees to the left and top center. Sunrise was at 7:22AM ... so the more "golden" light from the 1022 firmware is more accurate. The driveway does have a minor tint, so the 1022 handled that better ... but we'll see if a very slight red cast emerges with full sunlight.

As noted earlier, I like the 1022 colors better and the histogram confirms that it is reducing the blue channel.

Today is trash day ... so we've got some "extra targets" to look at! ;-)

Overall, I'm liking the 1022 upgrade - thanks @EMPIRETECANDY for getting this to us.

 

[alekk]

Here's a high Noon comparison of 0827 versus 1022 ... similar observations as before.

At this point, I would be comfortable (and WANT!) to push FW1022 ... but I'll hold off for a bit longer in case anyone has any specific comparisons they'd like me to do.