CAM-920
Out of focus
A city crossroads at midday, in full sun. Every piece is there: a white car pulling out of the junction, a blue one in the middle of the road, a red one on the right, the zebra crossings, the awnings on the corner and a group of people waiting on the near pavement. And none of it is sharp. This is not a fast-moving car or a distant background: the blur is the same across the whole picture, from the foreground to the furthest façade. You can see shapes and colours, but no faces, no plates, no sign you could read. IRIS sees that sharpness in this frame has dropped below what this same camera used to give. It does not know why; that is for whoever goes up to look at it.

Turn on IRIS Vision
AI-generated image
What IRIS understands
What it recognises in this scene
- frame out of focus
The rule
You decide once. IRIS applies it always.
If
sharpness across the whole frame drops below what that same camera used to give and stays low on the following checks
Then
- alert maintenance with the camera, the time and the video of the moment it started to drop
- keep the reference picture next to today’s, so the difference can be seen instead of argued about
- flag that camera as unreliable for anything that needs fine detail, until somebody checks it
The instruction that set it up«Tell me when a camera loses sharpness compared with how it used to see, and show me the earlier picture next to today’s.»
What changes
What changes for whoever answers for the cameras working
A camera that is out of focus keeps on recording. The light is green, the disk fills up and the panel says everything is fine. The trouble does not show up today: it shows up the day somebody asks for the video of a knock-down, a break-in at a doorway or a shunt between two cars, and nothing in that video can be read. By then the fault has been there for months. Nobody looks at four hundred pictures every morning, and this one is not even black: it is blurred, which draws far less attention. IRIS compares every camera with the sharpness it used to give itself and raises the alert the day it drops, with the earlier picture alongside. What changes is not the camera: it is that somebody finds out in time to fix it.
What people usually ask
Is this something the installer fixes, or do we have to replace the camera?
Most of the time it is a visit, not a purchase. Focus drifts over time, a mount that vibrates shifts it little by little, and a scratched or dirty dome does the same from the outside. IRIS does not say which of the three it is: it says which camera, since when, and against which picture it is being compared. With that, the technician goes up once knowing what for, instead of checking four hundred. Replacing the camera is the last option, not the first.
Shouldn’t the camera notice by itself? It has autofocus.
Plenty of cameras refocus on their own, and when they manage it there is nothing to report. The problem is the day they do not: the focus motor falls short, the dome is scratched and no focus will fix that, or the unit believes it is already in focus. A camera cannot judge its own picture. IRIS looks at it from the outside and compares it with what that same camera gave before. That is the difference between trying to focus and knowing whether it worked.
What if that camera has always been blurred? There is nothing to compare it with.
It happens, and the honest answer is this: with no good past, there is no drop to detect. So you do it the other way round: somebody goes up once, leaves the camera in focus, and that picture becomes the reference. From then on there is something to compare against. IRIS also flags the ones that were poor from day one, because it compares each camera with what it ought to be able to see at that distance; but the alert nobody sees coming is the camera that used to see well and has stopped.
Every simulation
On the edge of a drop
Tell me if someone steps to the edge of a drop with nothing holding them.
- The sentence
- Where it applies
- What it finds
- The same rule
Next step
Banking
Who is watching the cameras of a whole branch network?
13 cases: 12 raise an alert when something happens and one only measures. You see each image as it is, and then with what IRIS understands on top.
Casinos
How long does it take to find a hand played two hours ago?
12 cases: 9 raise an alert when something happens and 3 only measure. You see each image as it is, and then with what IRIS understands on top.
And the best part
With two screens you watch.With twenty you look without seeing.
It is nobody’s fault: it happens to everyone. IRIS watches all twenty at once and only interrupts when something deserves a person’s attention.
Tell us your problem and we will say whether IRIS understands it, or not yet.
We reply the same working day.