Review: Dahua 8MP / 4K 25x Starlight IR PTZ - PTZ5A4K-25X / SD5A825GA-HNR

[Wildcat_1]

All, I’m back in the saddle again with another review, this time for the EmpireTech PTZ5A4K-25X / SD5A825GA-HNR. Thanks to @EMPIRETECANDY for providing the unit in return for an unbiased no holds barred review.

Good to see a new 8K on the block for testing and this time one of the first to feature 2 new pieces of functionality (that will be coming to other units later this year), SMD 4.0 and Quick Pick (this one useful when using Dahua NVR’s). This one also features updated Auto Tracking algorithms, 3.0 so this is going to be interesting too test if this new PTZ lives up to the claims.

This is part of the Dahua Wizsense range so it gets object filtering (Human, Vehicle, Non-Vehicle) from an IVS rule perspective + Face Detection + the new and improved (I’ll be testing this) SMD 4. In the box is the camera, the PFB305W Wall mount bracket and then regular manuals etc. This does NOT come with a PFA120 or other junction boxes so make sure you purchase one if you don’t already have one

As usual I will be covering configuration, AI setup for critical capture, tips and tricks on how to best utilize and obviously a rundown of image quality in day/night capture scenarios. Initial vids and pics in post 2, AI feature breakdown and testing in post 3, tips and tricks on how to make these features work and other settings / menu functions in post 4 and a summary of findings / recommendations in post 5.

If anyone has any questions as I go just leave a comment. Please remember to keep questions/comments on topic with this thread. Thanks

NOTE: This review will take a little while before I can move to capture as I’ve already noticed a HUGE issue with compression. Specifically, the issue I saw on the 4K-T’s that I reported where compression artifacts and macro blocking were highly present. I’ve reported this to Dahua development / engineering already. Been pulling this one apart and made recommendations for adjustments and fixes already. Will update this thread once testing restarts.

----

Onto the specs

Firmware Installed On Delivery

2023-03-25 (not good FW, since been pulled by Dahua)

Specific Revision of Algorithms On This Verions
Thought I would start including this so you had for reference

Algorithm Version - 2.0.2
Security Baseline Version - V2.3


GUI

Uses the new Web 5.0 (V5.01.0.1408349.TA)

Camera Details

Image Sensor - 1/1.8” 8 Megapixel STARVIS™ CMOS
Effective Pixels - 3840(H) x 2160(V), 8 Megapixels
RAM: 1GB
ROM: 4GB
Electronic Shutter Speed - 1s~1/30,000s


Minimum Illumination
Color: 0.005Lux@F1.6;
B/W: 0.0005Lux@F1.6;
0Lux (IR on)
IR Distance - Up to 150m (492.13 ft)
IR On/Off Control - Zoom Prio/Manual/Auto
IR LEDs - 4


Lens Information

Focal Length - 5.4mm~135mm
Max. Aperture - F1.6~ F4.0
Field of View
H: 61.6° ~ 3.6°
V: 35.7° ~ 2°
D: 66.6° ~ 4°
Optical Zoom - 25x
Close Focus Distance - 0.1m - 2m (0.33ft to 4.92ft)

DORI Distance
Detect - 2700m(8858.27ft）
Observe - 1066m(3497.38t)
Recognize - 540m(1771.65ft)
Identify - 270m(885.83ft)

Mounting Options

• PFB305W - INCLUDED Wall Bracket in conjunction with the following (NOT INCLUDED):
• PFA120 - Junction Box
• PFA140 - Waterproof Bracket
• PFA151 - Corner Bracket
• PFB300C - Ceiling Mount Bracket
• PFB303S - Parapet Mount Bracket

 

[Wildcat_1]

Post #2

 

[Wildcat_1]

Post #3

 

[Wildcat_1]

Post #4

 

[Wildcat_1]

Post #5

 

[EMPIRETECANDY]

Thanks @Wildcat_1

 

[JDreaming]

Although we know the size of the camera on paper, but it will be great if you could put some popular cameras like SD4A425DB-HNY, IPC-T5442T and IPC-Color 4K-X next to the SD5A825GA-HNR so people can have a reference to the actual size of the camera when you do the review like you used to with the SD49425 and SD5A425.

 

[Wildcat_1]

Although we know the size of the camera on paper, but it will be great if you could put some popular cameras like SD4A425DB-HNY, IPC-T5442T and IPC-Color 4K-X next to the SD5A825GA-HNR so people can have a reference to the actual size of the camera when you do the review like you used to with the SD49425 and SD5A425.

Will do

 

[RN3KU]

Прошивка устанавливается при доставке

25 марта 2023 г. ( не очень хороший FW, так как его вытащила Dahua )

Конкретная редакция алгоритмов в этих версиях
Думал, я начну включать это, чтобы у вас было для справки


[/ЦИТИРОВАТЬ] В чём разница прошивок не подскажите ? Этой -
Версия системы
V2.820.0000013.5.R, Дата сборки: 01.03.2023
Веб-версия
В5.01.0.1393804.ТА
PTZ v.
V2.401.0000001.42.RHNCT_220811_42531
и той что вышла недавно - DH_SD-Taurus_MultiLang_PN_Stream3-Normal_V2.820.10SA000.0.R.230613 - ?

 

[Tinman]

So what firmware will you be using for your review ?

 

[Wildcat_1]

So what firmware will you be using for your review ?

I'm actually trying a new June Beta as well but unfortunately this also has the same issues.

Quick update on the issues I'm seeing and status. Currently in testing this thoroughly there is definitely a limitation due to SOC being used on this cam. Specifically in its impact on processing into h264. When there is a lot of movement including tracking and zoom into a target you'll see a lot of compression artifacting / macro blocking that only clears up once the camera slows down (has less to process on screen) where you will then see it snap back to good quality. I've reported this back to Dahua who are aware and also concluded same thing, SOC limitations causing impact. Since this is not something that can be fixed easily (SOC is SOC ) I'm doing extensive testing on how and where it's possible to elevate the pressure on the limited system resources. I've already found some ways to assist in this area BUT I'm not happy with the results yet. Will keep on testing and updating here. This unfortunately is the problem when you use an SOC from a fixed or vari-focal cam and place it into a PTZ without fully understanding the additional requirements (from a processing perspective) that are involved. For example in fixed or vari-cams, the FOV is fixed or determined by you, it doesn't change, therefore all you're doing is processing movement within a somewhat static scene. Generally that can be defined as target + any background motion. However in a PTZ you must cater for movement (tracking up down, left, right) + the FOV changing based upon targets within that such as zooming in + tracking, zooming out, revealing more of the background etc. I see this a lot when manufacturers do not adjust or replace SOC's when moving from model line to model line within a portfolio of different cam types. I'll keep working on this and continue to look to motivate Dahua to adjust settings to accommodate for this as I continue this test. I will also post back with ways to mitigate (as much as you can) within the current limitations.

Hope that update helps. Will keep you all posted

 

[Tinman]

I'm actually trying a new June Beta as well but unfortunately this also has the same issues.

Quick update on the issues I'm seeing and status. Currently in testing this thoroughly there is definitely a limitation due to SOC being used on this cam. Specifically in its impact on processing into h264. When there is a lot of movement including tracking and zoom into a target you'll see a lot of compression artifacting / macro blocking that only clears up once the camera slows down (has less to process on screen) where you will then see it snap back to good quality. I've reported this back to Dahua who are aware and also concluded same thing, SOC limitations causing impact. Since this is not something that can be fixed easily (SOC is SOC ) I'm doing extensive testing on how and where it's possible to elevate the pressure on the limited system resources. I've already found some ways to assist in this area BUT I'm not happy with the results yet. Will keep on testing and updating here. This unfortunately is the problem when you use an SOC from a fixed or vari-focal cam and place it into a PTZ without fully understanding the additional requirements (from a processing perspective) that are involved. For example in fixed or vari-cams, the FOV is fixed or determined by you, it doesn't change, therefore all you're doing is processing movement within a somewhat static scene. Generally that can be defined as target + any background motion. However in a PTZ you must cater for movement (tracking up down, left, right) + the FOV changing based upon targets within that such as zooming in + tracking, zooming out, revealing more of the background etc. I see this a lot when manufacturers do not adjust or replace SOC's when moving from model line to model line within a portfolio of different cam types. I'll keep working on this and continue to look to motivate Dahua to adjust settings to accommodate for this as I continue this test. I will also post back with ways to mitigate (as much as you can) within the current limitations.

Hope that update helps. Will keep you all posted

Do you feel this is also present in the SD5A425GA-HNR? I will attach another tracking example here for you to see. I do see some blocking in grass, but I guess I can accept that, the subject seems to remain good.

This is a customer that knows I see everyone before they ring my doorbell unfortunately I was at lunch, so i did miss them. There are also some rain droplets on lens from showers.

 

[Tinman]

I also found a little quirk in the SD5A425GA-HNR system reboot process with Blue Iris. I have the camera configured to do a system reboot once a week but, after the reboot Blue Iris thinks the camera's IP is the default 192.168.1.8 and then of course it never comes back In Blue Iris. The camera itself does retain the set static IP though. By looking at the camera's log I see it appears to load the saved configuration each time. So, during the startup, Blue Iris must get the default IP via onvif before the config file is loaded and thus it can't reconnect. So, my work around is, in Blue Iris is to go to the camera's settings/watchdog and set the timeout period to 150 seconds. This allows the SD5A425GA-HNR to fully restart and the issue goes away. This is the only Dahua camera I have that does this and the only one that runs the new web GUI 5.0.

 

[RN3KU]

Что то убрали с сайта , июньскую прошивку на эту камеру - SD5A425GA-HNR - DH_SD-Taurus_MultiLang_PN_Stream3-Normal_V2.820.10SA000.0.R.230613 ?

 

[juliand]

Thanks Wildcat
Looks like this cam may undergo an SOC refresh which is hardware, and not update-able, am I correct?
Looking for this to be my next buy in the next few days, so I'm thinking, this needs to be put on pause for now.

 

[Wildcat_1]

Still working this issue daily as it definitely needs a fix

 

[Wildcat_1]

Do you feel this is also present in the SD5A425GA-HNR? I will attach another tracking example here for you to see. I do see some blocking in grass, but I guess I can accept that, the subject seems to remain good.

This is a customer that knows I see everyone before they ring my doorbell unfortunately I was at lunch, so i did miss them. There are also some rain droplets on lens from showers.

Yes it is present there as well BUT its worse on the SD5A825 due to the resolution (4K) and the impact on SOC for compression in real-time. The 5A425 I can tune to make the issue you're seeing bearable and in mostly non-existent (dialing in to the max extent but not breaching SOC resources) but the 825 unfortunately can't keep up currently, regardless of setting. This is exacerbated by PFA as well which is a feature you do want for sure. The PFA algorithm tries to pre-determine a path for a target and adjust focus while the camera is in motion (Auto Tracking) and target is on the move. In other words you have a minimum of environmental motion (leaves, trees etc moving), target motion (self explanatory), camera induces motion (Auto Tracking) + PFA all playing a part in a changing scene/FOV that the SOC has to keep up with. While you'll always have some level of artifacting, pixelation and macro blocking caused by compression, if the SOC is not paired well, it can result in extreme issue and is certainly not desired. Looking at the 49225 as an example of how it can be implemented, you have 4MP resolution + an SOC that is not being taxed at all, the result, smooth motion, good PFA, great image while tracking (occasional hunt for focus but that is to be expected on any PTZ). The key here though is the SOC's ability to keep up with all of this video data and compress quick enough that it doesn't impact the overall image. It has to do this while other system resources (Web OS 5.0 as an example) are also competing for the same resource.

 

[Wildcat_1]

Thanks Wildcat
Looks like this cam may undergo an SOC refresh which is hardware, and not update-able, am I correct?
Looking for this to be my next buy in the next few days, so I'm thinking, this needs to be put on pause for now.

I wouldn't recommend buying it yet until I see some movement from Dahua on what is going to be done here. While SOC is the gating factor for sure, that doesn't mean the underlying Web OS 5 cannot be tuned to be less resource hungry OR that other functions can be lowered in priority for the SOC while functions like Auto Tracking + PFA are enabled etc. The key though is in seeing which of the recommendations I've made to code, can, will be implemented and then for me to test again. If nothing changes then that would be a shame as it would then mean a V2 would be needed to truly resolve the issue. I remain hopeful that I'll get some sort of resolution here for all of us but stay tuned and I will keep you updated. Thanks

 

[Alaska Country]

I wouldn't recommend buying it yet until I see some movement from Dahua on what is going to be done here.

Is the above in reference to both the SD5A425 and SD5A825 or only for the 825 which has the higher resolution?

 

[Wildcat_1]

SD5A825 is the one I would HOLD on recommending for now until issue is resolved