Can machine vision inspect this part?
Machine vision is feasible when the defect is visible under some lighting you can control, the part is presented consistently, and the decision fits inside the cycle time. It becomes a study when the surface or the variation fights the camera, and it is not a vision problem when the defect is not optical at all.
Computes in your browser. Nothing is saved or sent unless you ask. The link updates with your inputs, so a result can be shared.
Reviewed
How long the camera has per part, including the decision.
The single biggest factor.
Feasible.
Controlled lighting, consistent presentation, a defect the camera can see, and time to decide. The remaining work is choosing the lens, the lighting geometry and the reject mechanism — engineering, not research.
How this is calculated
The verdict follows the first constraint that decides, in this order: is the defect optical at all; can lighting be controlled; does the surface fight the camera; does presentation give a repeatable view; does variation defeat classical tools; is there time to decide.
“Needs a study” means a short feasibility study — a few hundred images of good and bad parts under candidate lighting — answers what nobody can answer from a description.
Planning guidance, not a quote — a written scope follows a look at the part.
How to use it
- 1Choose the defect type.
- 2Describe how the part is presented and its surface.
- 3Enter the cycle time.
- 4Say whether lighting can be controlled and how much the parts vary.
- 5Read the verdict and the constraint driving it.
These are planning numbers, not a quote — a written scope follows a conversation. We publish no prices.
Goes with
- CapabilityVision & Advanced Sensing
- Engineering notesSeven Ways Vision Projects Fail (and How to Avoid Them)
- Engineering notesMachine Vision Lighting: Why Most Failed Inspections Are Lighting Problems
- Engineering notesAI Vision vs. Traditional Machine Vision: Which One Does Your Application Need?
Other instruments
Asked about this tool
Why does lighting matter this much?
Because the camera only sees contrast. A scratch that is invisible under ambient light can be obvious under a low-angle dark-field ring. Most vision projects are won or lost at the lighting stage, not the algorithm.
What does 'needs a study' mean?
That the answer depends on data nobody has yet: images of good and bad parts under candidate lighting, a few hundred of them. A short feasibility study produces exactly that before anyone commits to a build.
Can AI vision handle what classical vision cannot?
Sometimes — subtle cosmetic defects, high variation. It needs labelled examples and it still needs consistent lighting and presentation. It does not rescue a bad camera position.
Strategy. Software. Systems.
Bring the number to a conversation.
Describe the line, the part, or the process. An engineer replies within one business day with whether and how we would approach it.
Book a callOr describe it here
One line is enough. A person reads it and replies within a business day.