Image quality for training

I am planning on training images in Edge Impulse. Should I be looking to train with the best quality images? That is should I think of using a higher quality external camera, as opposed the camera on the H7+, and import into OpenMV IDE?

It’s better to match the images between the camera and the training dataset. I.e. take pictures with the same camera.

Your answer is not clear to me:
A) train images with H7+ camera
B) train images with better quality camera

A. You want to want the training data to look like the data you are processing.