The presence of a commercial product which has reviews of varying quality should be a good sign that building something like this from scratch is a non-trivial undertaking. Doing IoT well is hard.
Doing it yourself brings some advantages - you can tune the installation for a specific location, implement your own configuration strategy (most likely cli based), associate with just your home router, etc.
There is no clear answer to which approach is better. Why are you doing this, what skills you you have already, what are you prepared to learn, is security important or can you rely on an ad-hoc implementation providing security through obscurity?
Start by defining your requirements. Lighting, Motion detection, video capture. Look at the commercial product you rejected, and decide which aspects of it you need to improve on (this is subjective).
List out some alternative implementations. This could be:
- MCU on each node, handling each aspect with network connection between them
- Raspberry Pi as a central controller with each part connected through (long?) wires.
- Commercial camera, commercial motion detection
- Commercial camera, MCU for motion detection
Trade off the time/cost/performance of these options, and go with whichever you prefer. You might find that some of the more low-level approaches are harder but are better documented (since the off-the-shelf components are unlikely to be documented, and will rarely be designed with extensions in mind).
Fundamentally, any working commercial product which satisfies your spec (even if it's got some flaws), will be far easier. Far more research is needed.