Can I recognize the content with 4.0x and get the bounding-box of the characters at the same time?

[Zhe Yuan]

I got suffered from the dilemma that I want to use the incremental training feature in version 4.0x, meanwhile get the exact location of a character with "makebox" in version pre-4.0.0.

I am wondering how can I achieve these two aspect at the same time?

Many thanks to your help!