Garrik's Lunatic Eval Thread

[Garrik]

Ok, in case anyone wants to follow along, here we go.

I realize that around here we take as gospel that EmpireTech cameras are the way to go. And I am not going to try to convince anyone otherwise. But I am an engineer, and I am going to do my own evaluation based on my own criteria. Full color motion at night is cool, but not really what I am solving for as a key eval criteria.

My key goals/criteria (as of right now, ready to input/suggestions/flames ) are:

1) Identify that someone we don't recognize is on the property at a time when that should be happening, and automatically generate an alarm in Home Assistant (my home automation tool)
2) Identify that a vehicle that we don't recognize is at the gate, or parked somewhere near the gate and raise an alarm.
3) Get a good enough shot of either of the above two threats to be able to identify them after the fact if necessary (ie if they stole something)

What am I evaluating? Two big things.

1) "Smart" AI NVR/IVR/VMS boxes. Assuming the weather allows, I will install both a Spot.ai box and a Lumana box on my network tomorrow. These do face recognition and ALPR - so will be comparing what they can do to the edge analytics on the various cameras that have them.

2) Cameras. Right now, I plan to install and evaluate the following cameras (I have these all in hand now):

EmpireTech IPC-Color-4K-X-6mm and IPC-B54IR-Z4E-S3

Hanwha QNO-8080R (5MP, 1/2.8" sensor, motorized varifocal) and XNO-9082R (4K, 1/2.8" sensor, motorized varifocal)

Axis Q1700E (just for ALPR, this is a specialty camera. Will compare Axis edge ALPR analytics to the Spot and Lumana server based ALPR).

Spot.ai bullet and turret cameras. These are probably made by Sunell (some Chinese OEM for sure, not certain which one), and are 5MP cameras with 1/2.8" sensors and 2.8mm fixed focal length.

Lumana bullet cameras, one 5MP (1/2.7" sensor, 2.8mm fixed focal length) and one 8MP (same same). These are also probably made by Sunell.

Several different Reolink models, which I expect to be at the back of the pack, but already have installed from my previous generation system.

A group of these cameras, probably including the Q1700, the Z4E and one of the Hanwha's, will be mounted on a roadside pole for ALPR comparison. The rest will be mounted for perimeter and building security. The setting is a ranch with fairly large amounts of outdoor space and multiple possible entry points (at least until I improve the fencing situation).

More on my evaluation plan, camera locations, results etc. if people are interested.

 

[bigredfish]

So many variables you’ll need a supercomputer to evaluate apples to apples. Hopefully you’re retired

Ambient light
Lens size
Sensor size
Placement and distance to target
IR vs color (daytime is easy for $50 cameras)
Anything 4MP+ with a 1/2.8 sensor will suck compared to 1/1.8
Good luck with ID/Recognition/access control in IR

 

[wittaj]

Ok, in case anyone wants to follow along, here we go.

I realize that around here we take as gospel that EmpireTech cameras are the way to go. And I am not going to try to convince anyone otherwise. But I am an engineer, and I am going to do my own evaluation based on my own criteria. Full color motion at night is cool, but not really what I am solving for as a key eval criteria.

So with your statement "And I am not going to try to convince anyone otherwise", is your mind already made up that EmpireTech (Dahua OEM) isn't gonna cut it and you have bias towards another brand....that doesn't seem very engineer like as one should let the testing determine the answers and you seem to have preconceived answer already...

Many here are engineers and have done their own research and evaluations.

I have boxes and boxes of cameras that I have tried. As have many members here. We share this information to save people from spending and wasting money trying to prove the advice wrong...

And that is why many here conclude that the best bang for the buck are the Empiretech Dahua OEM cameras in terms of build, performance day and night, and cost. Sure there are cheaper cameras, sure there are more expensive cameras, but these tend to win out in evaluations of the important criteria.

Your statement "Full color motion at night is cool, but not really what I am solving for as a key eval criteria" shows how little time you have spent here!

Anybody that has been here long enough knows that good full color motion at night is only obtainable with enough light and that cameras marketed as "full color" are marketing gimmicks as all cameras need light either visible or infrared. If the shutter needs to be slowed down to a crawl to be able to let enough light in to be color, then motion is a complete blur and is useless to most of us here.

Most of get the great captures that the police can actually do something with by using infrared and B/W and use color cameras to supplement the missing information.

But when someone shows that something is better, we go for it as the whole goal here is captures that the police can do something with.

Dahua makes several doorbell cameras, however, they are lacking in certain criteria and most here feel like Reolink currently has the best doorbell camera on the market. Thus proving that we will accept a different camera when it proves it can meet the criteria important to us.

But that is the only camera from Reolink that people recommend.

Here is the unofficial Reolink thread.

You can see all the attempts people have provided to demonstrate the quality of Reolink, and they are all a blurry mess at night or missing body parts or other messes.

We have challenged someone to provide a clean capture of someone moving at night with a Reolink and as you can see with 20 pages, nobody has yet to provide a usable image with motion at night. Maybe you will be that person....

Reolink's algorithm is designed to produce a nice bright static image at night and that comes at a cost of blur and ghost and missing body parts at night.

Reolink: Deconstruction of a dangerous misleading youtube review "Finding the BEST 4K Security Camera NVR Package (Reolink vs Amcrest vs Swann)"

But as @bigredfish says, it is tough to do an apples to apples comparison. To take a fixed 3.6mm camera and compare it to a varifocal at 32mm looking at an object 70 feet away is not a fair competition.

We look forward to you posting you plan and results.

 

[looney2ns]

I sometimes play an engineer on the internet.

 

[TonyR]

I sometimes play an engineer on the internet.

Hell, I can't even spell injunear enginear , see what I mean?

 

[garycrist]

I stayed at a Holiday Inn last night.

 

[Flintstone61]

When did Looney become humorous? I've been gone too long.
z4e's are good cams. Here's a car/ plate at approx 90 feet away. And some other cams to help add a better description of the vehicle. The monster LED street light helps on some angles and hurts on others.
so It's hard at night to find the " perfect settings".

 

[Flintstone61]

Days time is much easier. In the video it seems to be a woman with glasses driving. as the Sun angle changes your view into the car changes. She's facing North. Sun stays in the Southern sky as it Arcs across the daytime skies. hard to get a face shot unless you dedicate a cam to it. and then you'd be hard pressed to get 65%.