The options window is available from the Edit Menu. There are five important settings that may be configured from this window. Changes to these settings are not retained when the program is exited.
Pen width:
This option controls the thickness of the line used when drawing a character. The width is in pixels. Note that erasing (right mouse button) uses a line twice as wide as drawing. One may wish to reduce the pen width if one tends to draw characters that don't fill the entire image. The picture shows two letters A. The first is drawn with a pen width of 6; it will not digitize as well as the second one which is drawn with a pen width of 2. If one doesn't fill the screen and digitizations tend to be bloated, try reducing the pen width.
Learn depth:
The Learn depth setting determines whether Recog is stubborn or naïve. It determines the maximum weights that links in the neural networks can have. Each time Recog learns a character, the weights are changed by 1 to match the current input. If the maximum weights are large, it will take a lot of training to change Recog's behaviour. If the maximum weights are small, Recog learns quickly but has a short memory meaning that a couple of (possibly incorrect) learnings can completely change its perception of a character. The default value of 3 is on the naïve side. Recog is meant to be an educational program so it is preferable to play with fast-moving volatile nets.
Show digitizing box:
The first step in the digitization process is to establish the boundaries of the image. If this option is selected, a red box is drawn around the image indicating where the boundaries are and whether the aspect ratio limits affected them. Since the digitization process only scans for black, the red lines are invisible and do not affect subsequent digitizations.