Bed Camera Calibration
Support ▸ Bed Camera Calibration · machines with an overhead bed camera
What it does
It teaches ALPS where things really are on your bed, so that what you see in the camera overlay is where the laser will actually mark.
You do it once per machine. It runs in two passes, and the second one is the one people skip.
⚠ Why there are two passes — read this bit
Pass 1 (board) engraves a pattern of markers, photographs it, and works out your camera's lens distortion and exactly where the camera is looking. After pass 1 the overlay lines up with the bed.
Pass 2 (verify marks) marks a small set of test targets and measures where they actually landed.
Those sound like the same job. They are not, and the difference is the whole reason pass 2 exists:
Pass 1 can only ever tell you how the camera relates to the board. It can never tell you where the laser went.
Pass 1 photographs marks the laser made — so any error the machine has when marking is baked into the picture and into the calibration, and it cancels itself out. The board will look perfect. Pass 2 is the only step that puts the laser and the camera on opposite sides of the question, which is what catches a machine that is marking slightly off from where it is pointing.
A half-finished calibration looks finished. That is exactly why this is spelled out — the panel will happily show you a beautifully aligned overlay from pass 1 alone.
Before you start
- A flat sheet that covers the bed, in a material that engraves with good contrast. Anodised aluminium is ideal.
- The sheet butted against the rulers and lying flat — no bow, no rocking.
- Time: the board is a genuine engraving job, not a quick pass.
Step by step
- Open Support ▸ Bed Camera Calibration.
- Pass 1 — board. ALPS builds a plate sized to your bed. Send to Laser and run it.
- When it has finished, Capture & solve. ALPS photographs the board, finds the markers, and solves the camera.
- Pass 2 — verify marks. Run that job, then Capture & measure.
- ALPS reports the accuracy it achieved and saves the calibration to that machine.
The panel answers "Do I have to re-burn?" and "How repeatable is this calibration?" inline — if you are unsure whether an existing calibration is still good, start there rather than burning another sheet.
What "calibrated" gets you, honestly
It gets you a trustworthy overlay. Put a part on the bed, look at the camera view, and place your artwork on it — what you see is where it goes. That is what this is for, and it works.
⚠ It is not a certificate of absolute accuracy. Measuring an engraved feature against a genuinely independent reference — a calipered object the machine did not make — is a harder problem than lining up a camera, and it is one we are still working on. If your job depends on a dimension being right to a few thousandths, measure the part, do not measure the picture of it.
That distinction is worth knowing rather than discovering.
Things that quietly ruin it
| what | why it matters |
|---|---|
| the sheet is bowed or rocking | the camera is measuring a curved surface as if it were flat |
| the sheet is not against the rulers | everything is offset, consistently, and nothing looks wrong |
| running pass 1 only | you get an overlay that agrees with itself — see above |
| re-using a calibration after moving the camera | it is a calibration of where the camera was |
| a different material with poor contrast | markers fail to detect; the panel will tell you how many it found |
Exposure is handled for you
The bed camera's automatic exposure used to settle at one of two brightness levels between shots, and a measured part came out 0.86 % larger in the brighter one — enough to swamp the accuracy being chased. Since 0.1.197 calibration captures lock the exposure and restore it afterwards, so two calibrations of the same untouched bed now agree far more closely.
You do not need to do anything about this. It is here so that if you ever compared two calibrations on an older version and found them disagreeing, you know why.
If it will not solve
| what you see | what it usually is |
|---|---|
| very few markers found | contrast — anodised aluminium works, bare metal often does not |
| solves, but the overlay is out at the bed edges | the board did not cover enough of the bed |
| accuracy worse than you expected | check the sheet is genuinely flat before re-burning anything |
| "No live overhead frame" | another computer holds the camera, or the machine's camera service needs a power cycle |
Related
- Bed Flatness Map (Support ▸ Bed Flatness Map) — autofocuses a grid across the bed and shows how flat it really is, and how much of the unevenness the corner levelling screws can actually take out. No laser fires.
- 3D Engraving (
3d-engraving.md).