Lotus Text

Originally a thought experiment inspired by Valve’s steam controller UI, LotusText is a demo ROM that experiments with alternative text input for the NES. Typically, text entry on vintage consoles uses slow, boring interfaces that require scrolling back and forth across the alphabet to type one letter at a time. LotusText instead maps each symbol to a button combination, making everything quickly accessible. Possible uses for homebrew include things like quiz games, text adventures, dynamic personalization, or password systems.

Download ROM

Though a runnable demo ROM is available, LotusText is intended to be usable as an embeddable library. This is a little tricky to do, as NES homebrew doesn’t have many tools for creating reusable code, but the source code of the ROM demonstrates typical usage.

Usage

lotus-hello

On screen are 12 leaves, facing away from the center. On each leaf and in the center are three symbols, left, right and center. By pressing a direction on the d-pad the user highlights a leaf, using diagonal presses for diagonal leaves, and double taps for leaves far from the center. Once highlighted, the user can press B or A or both to pick a symbol to type. When B and/or A are released, the symbol is inserted. Start is used to confirm the inputted text.

Technical description

LotusText uses 3 of the 4 background palettes, and 1 of the 4 sprite palettes. The remaining palettes may be utilized by host programs to provide custom decoration. It only uses 2 sprites, and 14 bytes of zeropage RAM. CPU usage is low, requiring about ~160 cycles per frame and ~70 cycles in vblank.

Required assembler is ca65. Ports may be made in the future if demand exists. Graphics are compiled using makechr. Building is done using Make.

The main operations performed by LotusText are input smoothing and attribute changes. Leaves are highlighted based upon which direction is being pressed by the user, and the attributes used to highlight each leaf are decided from a pre-compiled metadata image. Changing this metadata requires python and PIL / Pillow.