Identify which parts of the image contain high-contrast details, like edges and textures, which usually indicates that those areas are in focus.
Activate the module by clicking on the
Focus peaking works by filtering out most of the image noise, measuring the intensity gradients in the image and calculating average and standard deviation statistics. When the gradient of an edge differs significantly from the mean, the associated pixels are marked with a “heat map” indicating how sharp the edge is.
- yellow represents a large (6σ) jump in gradient, indicating a very sharp edge.
- green represents a medium (4σ) jump in gradient, indicating a reasonably sharp edge.
- blue represents a small (2σ) jump in gradient, indicating a slightly sharp edge.
L’image ci-dessus a été prise avec une grande ouverture pour obtenir une faible profondeur de champ, et vous pouvez voir comment l’appareil photo a fait la mise au point sur le caractère chinois écrit sur la deuxième lanterne rouge en partant de l’avant. Il y a également des tiges de fleurs roses qui se trouvent dans la zone de netteté acceptable autour du plan focal, et celles-ci sont également marquées de jaune et de vert.
Note
The focus peaking detects local contrast and uses it as an estimation of sharpness. This is the only known method to do so as of 2023. While local contrast and sharpness are fairly correlated most of the time, edges between two objects having very different luminances (bright sky and tree leaves) will have a very high local contrast even when blurry, which will make them detected as sharp by focus peaking. A mechanism is in place to mitigate this issue in most cases, but very high-contrast edges will still be detected as sharp no matter their actual sharpness.Warning
The focus peaking does not detect if the images are sharp, but where the sharpest region of the image is, which is likely to be on the focal plane. This can help to spot front/back-focus, if any. But there is no guaranty that the most sharp region of the picture is actually sharp by some metric of definition of sharpness.Warning
Downscaled images have their gradients affected by interpolation methods, where noise is made less visible and local contrast is lost around small details, making focus peaking on small thumbnails inaccurate.Warning
The focus peaking is runs the detection on the final image as it appears in the graphic interface, after all processing steps if any. It will therefore take into account all the blurring and sharpening, as well as contrast enhancements, that may have happened in the processing.