Creating a custom filed type seems unnecessary. "upload" fields end up just being filenames. Perhaps you could have a compute field that uploads the image to S3. Retrieving should just be a simple URL reference.
- [web2py] upload - S3 storage Nils Olofsson
- Re: [web2py] upload - S3 storage nils
- [web2py] Re: upload - S3 storage Nils Olofsson
- [web2py] Re: upload - S3 storage pbreit