From a different thread, but to include it in our master thread:
So I am still missing the beginning frames.. see attached. Anything else I can do? Also, Disk is at 99% and CPU runs typically 71-78% Using an Asus N501VW laptop with 24 gig RAM, i7-6700HQ CPU 2.6 GHZ, 64 bit Windows 10 Running 12 cameras. Mix of Dahua, 1 Reolink PTZ (I had 2 and replaced 1...
ipcamtalk.com
So, let's take a look at this night video from this post.
They have incredible light and streetlights. Here is a freeze frame capture from that camera of a vehicle traveling maybe 10-15ish MPH under the streetlight, so about as ideal as you can get. Looking at this picture one would think it was middle of the day because you have so much light, but it was middle of the night 1am. Look at all that blur. I can tell it is a white car, but cannot tell make/model or if it is even a 2 or 4 door.
I have nowhere near the light quality this person has at night. Here is a capture from my Dahua OEM camera at middle of the night 2am with no streetlights and just the floodlights off my house at about 15ish MPH:
I can make out color, how many doors, make, model, etc.
The only way this is possible is by having a camera with the proper focal length for the distance one wants to cover, proper MP/sensor ratio, and a camera where you can manually change the parameters and the camera actually adheres to your settings.
Many cameras like Reolinks and other cheapo cams let you "set" the parameters, but the camera will override any user settings the the camera believes are in error because those cameras algorithms are written to provide a nice, bright,
STATIC image over anything else. I have a cheaper camera that lets me "set" the shutter. If you set a shutter for 1/10,000 at night, the image shouldl be pitch black. But nope, the image still looks nice and bright because the cheapo camera internally says "user error on the shutter speed" and makes it what it wants it to be for a nice bright image...