Arjun
IPCT Contributor
2026 New Release from EmpireTech
- Thread starter EMPIRETECANDY
- Start date
Chokolinho
Getting comfortable
Close Focus Distance 2.8 mm: 4.5 m (14.76 ft)
3.6 mm: 5.5 m (18.04 ft)
3.6 mm: 5.5 m (18.04 ft)
bigredfish
Known around here
Holy Shit! And max 8192 kbps bitrate?
Nevermind.
This is for the "but its Bright and I dont care about image quality" crowd...
bigredfish
Known around here
Even the 4MP (24PRO-X) IPC-HDW2449TH-S-PROX
....has long Close focus distances... wonder how/why the 5442 was able to have much better close focus distances?
24PRO-X
....has long Close focus distances... wonder how/why the 5442 was able to have much better close focus distances?
24PRO-X
| Close Focus Distance | 2.8 mm: 1.8 m (5.91 ft) 3.6 mm: 2.6 m (8.53 ft) |
bigredfish
Known around here
Would like to see it... Dahua's own marketing video on their own website shows a fucking basketball with worse trailing motion blur than a Reolink....?
Arjun
IPCT Contributor
Could be sensor related
Who wants a copycat rated cloned Sony/Omnivision sensor?
Chokolinho
Getting comfortable
There really are nothing but awesome cams coming out all the time. Each one is better than the last.
Arjun
IPCT Contributor
Don't say that, I don't think the people at Dahua understand sarcasm, their approach is egocentric and monotonous; they will always think their way is right
Jesus...
Stop this nonsense about sensors already. All these new WizColor 2/3/5 series cameras are built on the same sensors. And they're not bad - they were used in many cams before AI-ISP / WizColor non sense with good effects.
The problem lies in the implementation of video transcoding from the sensor in the SOC. These new chipsets use newer denoising, brightening (always-HDR) and coloring algorithms borrowed from the smartphone / sport cam / drone market, which uses sensors with 48 or even 100 megapixels (so the pixels are much smaller).
The previous implementation handled brightening and denoising very well for static content. Unfortunately, there was no comparable solution for moving elements in AI-ISP 1.0. It's clear that AI-ISP 2.0 (WizColor X) is trying to solve this.
Looking at how the latest action cameras and drones perform at night, it's very easy to predict there will be AI-ISP 3.0 and AI-ISP 4.0. Each subsequent version, using the same image sensors, will provide increasingly better support for people/cars/objects in motion.
today progress in video is less and less from sensors, more and more from newer AI video algorithms in SOC chipsets..
It isn't talked about here because all of the main players here isolate their cameras from the internet, so no Chinese servers are being pinged.
And to date this progress has shown a degradation in low light performance.
And at what point does doing AI on every video end up changing the facial quality of someone to the point that it is a clear face, but not the person in the image due to over AI processing?
I'd rather have a little noisier image yet the police can actually ID the person from it than a clear image but the face was changed just enough to add doubt if that is the same person.
Arjun
IPCT Contributor
Right, but there are so many noobs joining every month asking the same question - its time to either ship out these cameras "clean" from Day 1 or put an auto-configurator / script in the ipcamtalk cliffnotes section and redirect the noobs there
Arjun
IPCT Contributor
Sweet, I'll spend my money on another day then lol, meanwhile save your money in these unpredictable times; if you're someone that needs cameras for 24/7 surveillance then thus far, these new WizColor series are not meant for deployment
@bigredfish agrees
bigredfish
Known around here
Arjun
IPCT Contributor
It's partly due to the reliance on AI-locally based processer
Before AI, SMD was worse
Last edited: