Dennis Baldwin has published a number of videos on this subject on his YouTube channel: https://www.youtube.com/user/dennisbaldwin/videos
He uses OpenDroneMap to stitch the images together and geo-reference them. One of his tutorials shows how to use Amazon EC2 cloud computing to accelerate the rendering process.
Commercial: Structure from Motion (SfM) software Agisoft, Pix 4D are good (demo and 'low cost' research licenses are available).
Opensource - Visual SfM, OpenDroneMap, MS Image Composite Editor (ICE) are all good. ICE is image mosaic software but cannot generate DEM/DTM etc....
I'm getting good results with Agisoft Photoscan and Visual SfM using a 3DR x8+ and Canon GPS enabled SX260 camera. You will need GNSS/GPS ground control in your images to increase the 3D spatial accuracy of both generated surface models and ortho outputs. Onboard camera GPS is good for 'rough' geo-location of image outputs.