How does a camera autofocus make sure that the subject is in focus without requiring the photographer to move the camera physically? Would this be done by moving the sensor so that the focal point is in the correct location?
BenPust
ellenluo one way autofocus can be implemented is to measure contrast of the image sensor in the camera, so when an image is in focus the intensity difference between adjacent pixels of the sensor naturally increases with correct image focus. Therefore, we compute autofocus by adjusting the lens automatically until the some max contrast is detected.
BenPust
ellenluo there's also a more in depth explanation of autofocus at the end of the slides in the extra section
keirp
So in order for cell phones to calculate how much to blur each object when doing portrait effects they have to calculate the exact depth of each object in the scene?
sirejdua
@keirp Portrait mode uses machine learning to compute a depth map, and uses that to refocus and reblur. iPhones with stereo cameras can also use the stereo information to compute a depth map
michellebrier
Is the image plane a focal plane in this diagram, so z_i is the focal distance?
How does a camera autofocus make sure that the subject is in focus without requiring the photographer to move the camera physically? Would this be done by moving the sensor so that the focal point is in the correct location?
ellenluo one way autofocus can be implemented is to measure contrast of the image sensor in the camera, so when an image is in focus the intensity difference between adjacent pixels of the sensor naturally increases with correct image focus. Therefore, we compute autofocus by adjusting the lens automatically until the some max contrast is detected.
ellenluo there's also a more in depth explanation of autofocus at the end of the slides in the extra section
So in order for cell phones to calculate how much to blur each object when doing portrait effects they have to calculate the exact depth of each object in the scene?
@keirp Portrait mode uses machine learning to compute a depth map, and uses that to refocus and reblur. iPhones with stereo cameras can also use the stereo information to compute a depth map
Is the image plane a focal plane in this diagram, so z_i is the focal distance?