Libcamera interface to Pi camera

@samuelmcdermott just showed me the announcement from May that the Raspberry Pi camera is now able to be run via libcamera. From their announcement, it looks like this provides a lot of features we’ve wanted for ages. The camera docs are also updated, and feature a really nice guide on tuning the sensor as well as the best explanation I’ve seen of the image processing algorithms. It looks like this really is the best of both worlds: it can still use the GPU to process images (so it’s still fast), but is now much more under the control of the CPU, so we have a better idea of exactly what processing is done to the images.

I am not sure if there are any Python bindings, it would be awesome to try it out if there are.


If I understand correctly this would allow to combine multiple pictures to create a new composite image with an extended depth of focus on the fly?

I’m not sure if libcamera makes that easier or not - I think the limiting factor there would be acquiring a Z stack of images, so reconstruction could be done “on the fly” but each image would take at least a few seconds to acquire (~10s) even if reconstruction was instantaneous…