Post #2 - Initial Pics & Videos
2.1 - Initial Video Of Auto Tracking
I wanted to get a post up of the auto tracking speed and performance of this camera as that is a) one of the most asked questions of me for this camera so far and b) lets be honest is something that is a phenomenal feature of PTZ’s………when it works.
So before I mount outside, I wanted to perform a lights on / lights off full IR test in a dark room to see how this new PTZ holds up. So this is a little test I like to called, Hide & Seek. Without spoiling the video, this camera is responsive. Even if it reverts to preset (loses a target briefly, tracking time expires, doesn’t match target criteria of what you defined like Human or Vehicle) it picks back up almost instantly and continues to track. Even if the AI blips for a second, Deep IVS kicks in and ensures tracking is fluent. I want to point out again, this test is not for image (more tweaks later for that) but for auto tracking only.
Exposure set to 1/250 (not really needed but was testing with vehicles earlier, will set to more applicable expo and dial in pic later), SmartIR = On, SMD = ON & Human/Vehicle, Smart Plan = Deep IVS (Human/Vehicle). All other settings were set to defaults for initial testing.
Testing was on a small make shift rig for this testing only, excuse the temporary cabling Lastly, thought I saw a slight smudge, sure enough after capturing saw one on the lens, will clean for the next test.
Take a look (especially at 53s, 1:57 & 4:53, 5:55 where I try to hammer the tracking) and let me know with any questions. More to come.
In the first post, the link to Andy's Amazon site takes me to a "Loryta" Amazon page.
Are the Loryta Amazon listings also Andy's?
RodeoGeorge
Loryta is US registered brand, empreitech still under registering. The camera still can be updated via dahua official firmware without problem.In the first post, the link to Andy's Amazon site takes me to a "Loryta" Amazon page.
Are the Loryta Amazon listings also Andy's?
RodeoGeorge
Have you set the Direct To Disc ? I have 3 camera's and my CPU is around 9%
I got mine installed and I'm doing some preliminary experimentation on the auto-tracking and how to direct the PTZ camera to areas the other cams trigger on. Here's a quick video clip where the camera that sees the shed calls the PTZ around to Preset 3 once I walk toward the shed from the dumpster.
Initial notes:
1. the camera's Web interface can consume substantial amounts of CPU power if left open. I was all why is my CPU maxed? and it turned out to be the camera's web interface taking more CPU power than expected, as I experimented with settings.
2. I've had the camera lose its position calibration a couple of times now, where Presets still exist, but the camera's not aimed at the correct location anymore. I'll keep an eye on it.
3. if I have Blue Iris use a different camera to send the PTZ to a Preset, the PTZ does not return to the preset that I specified in the PTZ > Fuction > Idle Motion section. For example, if the shed camera detects motion and sends the PTZ to Preset 3, the PTZ should return to Preset 1 after 1 minute, but it does not go back to Preset 1.
A solution would be to program Blue Iris's Alerts > On Reset to do a 15-second WAIT and then order the PTZ back to Preset 1.
4. The PTZ considers HOME to be looking at the surface it's mounted on, which is a little annoying that I can't set a useful Home position. Is this normal for Dahua PTZs or is it something they might fix with a firmware update?
5. It is not doing as well as I hoped in tracking a person who walks out the edge of the scene. In my sample video, you can see it did OK, and followed me over to the dumpster. I'm still playing with the settings to understand the influence of the different criteria in the Deep IVS panel, like Tracking Target size.
6. It appears to have some trouble tracking a person who is walking directly toward and under the camera.
Update, I might need to back down on my Tracking Target Size, which apparently is how tightly the camera zooms onto a recognized target. Too much zoom and the target may be able to walk out of the scene, is my hypothesis. Trying this now:
View attachment 51838