I am all for local however my biggest issues with video doorbells tends to be the field of view, image quality and operating temperature ranges.
I live in a climate where +30C and -20C will happen at least for a few days a year and quite often I have seen doorbells just go offline below when it gets cold outside or physically degrade in the heat.
I want a highly durable device. These ESPCam devices kinda have crap optics, I want to clearly identify people in the dark.
I’m surprised I haven’t heard more of this (makes me think its impossible), but I feel like the ideal world is buying big tech’s hardware and overriding its firmware/software to talk to home assistance locally. I saw a post recently about someone making an google nest speaker work with home assistance and they really just made their own speaker and shoved it in a nest mini body. I like the hardware I have, I just wish I had more control over it, and sadly I don’t know nearly enough to mess with firmware myself.
I’m in the process of trying something like this. The cameras I bought have serial ports and firmware packages that can be reverse engineered. That’s where I’m currently at; just having the knowledge that it’s possible. I haven’t reverse engineered anything before, so if my attempts don’t work my backup plan is to harvest the the raw components and cobble together my own firmware on a different MCU.
The newest reolink doorbell comes close on most counts but it isn’t rated cold enough and the RTSP stream requires specific firmware updates to be stable with home assistant / frigate.
When I was still playing with it and it was sitting on my desk I did notice that it ran a bit warm. Not hot enough to be a concern but now you have me thinking it might be an advantage in cooler climates.
The lack of a battery makes me less concerned about cold as well.
I’ve only had it a few months but it seems far more reliable than the Ring doorbell I’d been using previously. I do have it rebooting nightly but I haven’t had to touch it in months.
Winters here don’t often go below 0F anymore. I guess I’ll see how it goes.
Sounds like a much more complicated problem… maybe resistive heater would be enough? I’m assuming there isn’t any problem with the cold other than the whole thing icing over.
Better optics/camera is probably where it all falls apart… you’re not getting it out of a pre-made $8 dev board any time soon.
It’s the battery. A lot of doorbell cameras have an internal battery to keep the camera powered when the chime is rung. A few of my friends have to replace their unifi doorbell cameras once a year after winter
If they used electrolytic capacitors they can freeze, and in areas with hot cold night cycles moisture can condense inside the unit which cause problems. It really comes down to some component selection and design.
A camera platform small enough to fit a standard door frame isn’t going to be able to hold serious optics. The sensor sizes are going to be too small for good night vision, if nothing else.
The temperature issue is probably an issue of economics. Most people don’t need such a wide range, so off the shelf products don’t support the application, as it would incur additional cost.
Both of these are solvable issues however if you’re willing to use a standard security camera mounted over the door instead. There are many cameras that you could pick with large sensors, heating elements, and more.
A camera platform small enough to fit a standard door frame isn’t going to be able to hold serious optics.
We clearly have even smaller optics in smart phones, the limiting factor on doorbells seems to be depending on cheap optics, poor FOV lenses and then cutting the bit rate by only supporting 2.4Ghz vs offering 5Ghz or POE support. There are a number of cameras with good optics or secondary package cameras now to cover the FOV issue. There are even cameras that meet all the clarity requirements I have but are cloud only or are knowing to not work in cold weather at all.
The MAIN job of a doorbell camera should be for the security role of identification. As it will point directly at the face of someone the cheap ones do a bad job of this.
There are some videos showing just how WIDE the quality range is, but sadly very few tick all the boxes.
I think a mobile phone camera is vastly superior to these, although might not be great at night vision for the reasons you said, but is it entirely crazy to not just use a spare phone? It has built in backup power, can store videos locally if there is an internet outage, and can use its own data connection if wifi is not available.
TBH the only way I’ve found around this is a full CCTV solution monitoring my perimeter, the doorbell can just identifies those who come really close and push the button.
A doorbell camera should be viewed as for convenience on the door. If you want adequate security you really need to get a dedicated security focused solution.
I am all for local however my biggest issues with video doorbells tends to be the field of view, image quality and operating temperature ranges.
I live in a climate where +30C and -20C will happen at least for a few days a year and quite often I have seen doorbells just go offline below when it gets cold outside or physically degrade in the heat.
I want a highly durable device. These ESPCam devices kinda have crap optics, I want to clearly identify people in the dark.
I’m surprised I haven’t heard more of this (makes me think its impossible), but I feel like the ideal world is buying big tech’s hardware and overriding its firmware/software to talk to home assistance locally. I saw a post recently about someone making an google nest speaker work with home assistance and they really just made their own speaker and shoved it in a nest mini body. I like the hardware I have, I just wish I had more control over it, and sadly I don’t know nearly enough to mess with firmware myself.
I’m in the process of trying something like this. The cameras I bought have serial ports and firmware packages that can be reverse engineered. That’s where I’m currently at; just having the knowledge that it’s possible. I haven’t reverse engineered anything before, so if my attempts don’t work my backup plan is to harvest the the raw components and cobble together my own firmware on a different MCU.
The newest reolink doorbell comes close on most counts but it isn’t rated cold enough and the RTSP stream requires specific firmware updates to be stable with home assistant / frigate.
When I was still playing with it and it was sitting on my desk I did notice that it ran a bit warm. Not hot enough to be a concern but now you have me thinking it might be an advantage in cooler climates. The lack of a battery makes me less concerned about cold as well. I’ve only had it a few months but it seems far more reliable than the Ring doorbell I’d been using previously. I do have it rebooting nightly but I haven’t had to touch it in months. Winters here don’t often go below 0F anymore. I guess I’ll see how it goes.
Sounds like a much more complicated problem… maybe resistive heater would be enough? I’m assuming there isn’t any problem with the cold other than the whole thing icing over.
Better optics/camera is probably where it all falls apart… you’re not getting it out of a pre-made $8 dev board any time soon.
It’s the battery. A lot of doorbell cameras have an internal battery to keep the camera powered when the chime is rung. A few of my friends have to replace their unifi doorbell cameras once a year after winter
Wow that’s wasteful, I’d expect a bit more from ubiquiti kit
You’d think that’d be a perfect job for a supercap.
If they used electrolytic capacitors they can freeze, and in areas with hot cold night cycles moisture can condense inside the unit which cause problems. It really comes down to some component selection and design.
A camera platform small enough to fit a standard door frame isn’t going to be able to hold serious optics. The sensor sizes are going to be too small for good night vision, if nothing else.
The temperature issue is probably an issue of economics. Most people don’t need such a wide range, so off the shelf products don’t support the application, as it would incur additional cost.
Both of these are solvable issues however if you’re willing to use a standard security camera mounted over the door instead. There are many cameras that you could pick with large sensors, heating elements, and more.
We clearly have even smaller optics in smart phones, the limiting factor on doorbells seems to be depending on cheap optics, poor FOV lenses and then cutting the bit rate by only supporting 2.4Ghz vs offering 5Ghz or POE support. There are a number of cameras with good optics or secondary package cameras now to cover the FOV issue. There are even cameras that meet all the clarity requirements I have but are cloud only or are knowing to not work in cold weather at all.
The MAIN job of a doorbell camera should be for the security role of identification. As it will point directly at the face of someone the cheap ones do a bad job of this.
There are some videos showing just how WIDE the quality range is, but sadly very few tick all the boxes.
BEST Wired Video Doorbells 2022 Edition
Ultimate Battery Doorbell Showdown 2022
I think a mobile phone camera is vastly superior to these, although might not be great at night vision for the reasons you said, but is it entirely crazy to not just use a spare phone? It has built in backup power, can store videos locally if there is an internet outage, and can use its own data connection if wifi is not available.
TBH the only way I’ve found around this is a full CCTV solution monitoring my perimeter, the doorbell can just identifies those who come really close and push the button.
A doorbell camera should be viewed as for convenience on the door. If you want adequate security you really need to get a dedicated security focused solution.