Is it possible to use embedded script with visual capture API to optimize latency ?

## Why this question ?
I'm a research engineer who works in autonomous driving and i need to acquire images with minimal latency. For the moment axis images gives me latency of ~250ms, which is quite huge for my application.
I want to reduce this latency, but it seems it can't be reduced anymore with the web interface (i'm already using all i can, like low latency mode, no image correction, ..)

## Video capture API
I want to know if the video capture API can help reduce latency. I think it could be useful, since some parameters like bufferisation or pipeline are not customisable in the webui interface. The goal is to have an embedded script that take snapshot and send it via websocket (or other protocol) to a server connected.
Fps do not need to be very high (15fps is already quite good) but a good latency should be under 100ms.

What do you think of this idea ? Do you have already any code template that could fit this purpose ?
Thanks in advance,
M M

Start by reviewing the video capture API and the existing web UI capture options, focusing on bufferisation and pipeline controls. Determine whether an embedded snapshot sender using WebSocket or another protocol can meet the requested 15 fps and under-100-ms latency target; done means documenting feasibility and a suitable code template.

Written by the indexing model from the issue text.