Why 64 Megapixels Beats Stitching on a Large Flat Part
Large parts arrive at a fixed station and move on once inspected. The core conflict is a division: the field of view has to cover the whole part, the defect has to be visible below the millimetre level, and dividing one by the other gives the pixel count needed. If there are not enough pixels, only two routes exist, shrink the field and shoot in sections to stitch afterwards, or accept that some defects cannot be seen. The first stretches the cycle, the second gives away quality.
Field of view and resolution are a division sum
The scheme works backwards from the smallest defect and the physical size each pixel covers: once the field is fixed, the defect has to occupy enough pixels, and the total pixel count follows. The 64 megapixels provide margin in that calculation, keeping enough defect resolution across a wide field so fine defects are found in one shot. Doing this division at selection is more useful than comparing camera specifications side by side.
No stitching saves more than time
The cost of multi-shot stitching is routinely underestimated. Besides capture and stitching time there are two hidden costs: missed defects at the seam, and registration error between frames. A defect landing near a seam can be split into two halves that neither frame shows as a complete defect, and registration error shifts both position and size. Single-shot imaging removes the stage entirely, so judgment works on one complete image and the result is more consistent.
What fills the 8 seconds
The cycle is 8 seconds, subject to site acceptance results, and it holds because the process collapses into three steps: locate, capture, judge. A fixed-station locating mechanism keeps placement identical each cycle, the camera covers the inspection field in one shot, and the algorithm delivers the verdict. What is saved is repeated capture, stitching and judgment time.
Shorter cycles raise the throughput requirement on the processing unit, and the two are tied together. A cycle cannot be promised from camera specifications alone.
Data throughput is a hidden selection item
A 64-megapixel frame carries a lot of data and strains transfer and processing. The scheme processes by region: coarse localisation, then fine judgment on the target area, avoiding a full pixel-by-pixel pass. That is the algorithmic answer and the hardware has to match it, so camera, compute and algorithm are evaluated as one system rather than fixing the camera first.
One exposure leaves no second chance on lighting
Single-shot imaging means one exposure decides the result. Across a wide field the light has to be even, or defects in dark zones lack contrast and detail in bright zones clips, with no second image to fall back on. A stitching scheme can adjust each small image; one large frame has to get the light right first time. That is the real cost of going stitching-free.
When ultra-high resolution is the wrong answer
If the part moves and cannot be located stably, a long single exposure smears and the advantage disappears. Where depth of field must be large, aperture and depth trade against each other at wide field and high resolution. Where the field is small and the defects are not fine, the extra pixels only add data.
It fits a fixed station, a flat inspection target, a wide field, fine defects and a tight cycle. The conclusion generalises across industrial large-part inspection: settle the pixel budget first, then decide between single-shot and stitching. The simplicity of the scheme is part of the value, with fewer stages, a shorter chain and lower engineering and maintenance complexity.
Beyond the pixel budget
Is more megapixels always better?
No. Pixels match field and smallest defect; beyond that it is only data. The lens must resolve it, the light must be even and the processor must keep up.
Does stitching-free reduce coverage flexibility?
Yes. A single-shot field is fixed, and a part outside it needs redesign. Stitching is more flexible for very large parts, at the cost of more stages.
Does the 8 seconds include loading?
No. It covers locate, capture and judge. Loading and transfer follow the station cycle separately, subject to site acceptance results.