The log tells you when and where. It rarely tells you what happened.
A clean and dry room is monitored continuously, against thresholds that trigger an investigation when a reading moves. The alarm gives you a sensor and a timestamp. Working out what actually occurred in that window usually comes down to asking people what they remember.
Continuously measured
- Airborne particle count
- Dew point and humidity
- Temperature and pressure
- Deposited film thickness
Usually not recorded
- Who entered, and when
- Whether gowning was completed in sequence
- Time spent in the airlock
- How long a door stood open
- What tools and materials came in
- How rolls were handled going in and out
The clip.
Vaidio running a custom-trained model on a moving line. Different material, different industry. What transfers is the method: the producer supplies images of what is acceptable and what is not, those become a training set, and the model that comes back runs against the camera feed. The standard is theirs, not a vendor's.
15 seconds. The original runs two minutes and shows the same thing.
Most of that second column is standard.
Gowning, entry, dwell time, doors, and material transfer are analytics that run out of the box on cameras that already exist, with an alert when something crosses a threshold rather than a person reviewing footage afterward.
The custom model in the clip is for the one thing no vendor can define for you, which is what acceptable looks like in your own material.
Where we think it helps, and where it doesn't.
We can't examine your process parameters, and wouldn't try. What we noticed is that handling and room discipline usually go unrecorded, which means they can't be cleared from a list the way a logged parameter can.
A camera does not diagnose a deposition process. Most of what drives variance there is process control, and that is a different discipline entirely. What a camera does is take the unrecorded branch off the list, so the search narrows to the parts that are actually instrumented.
And the same cameras are pointed at a building where lithium metal is handled, which makes the safety and operations questions answerable by the same installation rather than needing their own.
What we don't know.
We have never been inside a lithium film operation. Everything above is a guess from the outside, and you may already have all of it covered. These are the questions that would establish quickly whether it is a useful guess:
- When an excursion is investigated, how often does it close as undetermined?
- If a door were propped for twenty minutes on a night shift, how would anyone know?
- When a customer asks what happened during a specific shift, what can you produce?