CAM-924
The colour has gone
A pedestrian square at midday, seen from a fourth floor. In the middle there is a round stone fountain, with palm trees and planters around it, a kiosk on the left and a café terrace with parasols at the far end. Along the top runs the road, with cars and a zebra crossing. The whole frame is tinted violet: the paving, people’s clothes, the cars, the parasols. There is no good half and bad half; it goes corner to corner. IRIS still sees the shapes — there are people crossing, there are vehicles, there is a fountain — but colour from this camera has stopped being a reliable fact. What IRIS does not know is why: a setting that drifted, a filter that has not gone back to its place, or the sensor itself.
Turn on IRIS Vision
AI-generated image
What IRIS understands
What it recognises in this scene
- colour cast
The rule
You decide once. IRIS applies it always.
If
a camera’s colour drifts away from what that same camera showed on similar days at the same hour, and the drift pulls the whole frame with it
Then
- alert maintenance with the camera, the time and the video of the moment it changed
- put the reference picture next to the one from now, so the difference does not have to be argued
- flag that camera as unreliable for colour searches until somebody checks it
The instruction that set it up«Tell me if any camera loses its colour and starts seeing everything in a tint that is not there, and tell me which one and since when.»
What changes
What changes for whoever answers for the cameras working
Colour is not decoration: it is something people search with. Somebody asks for “the white van from this morning” and on this camera nothing comes back, because here white is lilac. And the person searching has no way of knowing: they get no results and conclude that nothing happened. That is worse than a camera that is off, because a camera that is off is obvious. With IRIS somebody finds out the same day: the alert arrives with the camera, the time and the earlier picture next to the one from now. And while that camera stays like this, anyone searching it knows that colour is no use there and that the way in is shape, vehicle type and time.
What people usually ask
If I search for “a red car” and that camera has its colour broken, what happens?
That car may not come back at all. Searching by colour leans on colour, and if the camera has it shifted, a red car can arrive as brown or as violet and fall outside the search. The dangerous part is not failing: it is failing silently, because the person searching reads “no results” and takes it to mean nothing happened. That is why IRIS flags the camera and says that there you search by shape, by vehicle type and by time, and review the video by hand.
How does IRIS tell a colour fault from an orange sunset?
By comparing the camera with itself. Every camera builds up a record of how its own scene looks at each hour and in each season, so a sunset surprises nobody: it arrives when it should, it lasts a while and it pulls mostly on the sky. A fault cast behaves differently: it appears at once, it stays, and it drags the asphalt, the walls and people’s clothes along equally. IRIS reports the difference; confirming it is for the person looking at the picture.
Is it a fault, or did somebody change a setting?
IRIS does not say, and it should not. What it states is the fact: this camera is seeing a colour today that it was not seeing yesterday. The cause may be the white balance, a profile left changed after someone configured something, the infrared filter that has not gone back into place, or the sensor. Some of those are fixed from a browser in a few minutes; others mean climbing up. Which one it is is decided by the person who looks at the video and the camera’s history.
See it for yourself
Everything comes in. Only what matters gets through.
This sums up what IRIS does with one alert, from the moment it appears until it is closed. The real screen holds far more; this is the main thread.
- The hour, as a grid
- You look where it's dark
- You jump to that minute
Next step
Who sees the sales that never reach the till?
10 cases: 4 raise an alert when something happens and 6 only measure. You see each image as it is, and then with what IRIS understands on top.
Who is watching when the load swings over somebody?
11 cases: 9 raise an alert when something happens and 2 only measure. You see each image as it is, and then with what IRIS understands on top.
And the best part
At four in the morning,the camera is still watching just as it does at noon.
No heavy eyelids, no distractions, no shift change. What happens at night is understood at night, not the next morning.
Tell us your problem and we will say whether IRIS understands it, or not yet.
We reply the same working day.