What to Look for in a Good Image?
A good image is the foundation for reliable AI inspection. This page outlines the visual and technical characteristics that define a high-quality image on the OV20i system.
Key Characteristics of a Good Image
| Feature | What to Look For |
|---|---|
| Sharp Focus | Object edges are crisp; no blur or softness |
| Even Lighting | No extreme shadows, glare, or hotspots across the ROI |
| Proper Exposure | Image is not washed out (overexposed) or too dark (underexposed) |
| Object Centered | Part or ROI is fully inside frame, not cut off |
| Correct Distance | Part is framed appropriately for the lens used |
| Low Noise | Minimal graininess; no excessive digital noise due to high gain |
| Aligned View | Camera is positioned orthogonally unless oblique view is intentional |
How to Get a Good Image
- Use Focus Slider in the Recipe → Image Settings
- Adjust Exposure to reduce blur or washout
- Add Lighting if image appears dim or noisy
- Reposition Camera if part is not centered or angled awkwardly
- Check Gain — reduce if image looks noisy
Example: Bad vs Good
| ❌ Bad Image | ✅ Good Image |
|---|---|
| Blurry edges | Sharp, clear part outlines |
| Dark or shadowed ROI | Bright, evenly lit region |
| Part cropped at edge | Part fully visible and centered |
| Washed out details | Clear texture and contrast |
Bonus: Test Image Capture
- Use the “Test Image” button in the HMI to validate camera setup
- Use that image to begin training a Recipe (Classification or Segmentation)