Quick Menu Icon Creator

Quick Menu is a graphical user interface for MS-DOS developed by OSCS Software Development. Quick Menu allows users to create their own menu structures for efficient launching of games and applications without having to navigate a drive via MS-DOS commands such as cd and dir.

Quick Menu employs 32x32 icons with the EGA (4 bits per pixel) color scheme. The icons are stored in Quick Menu in an internal archive, but can be exported to or imported from .icc files. This repository contains a set of Python-based scripts to read those .icc files or to convert .png files to .icc files.

The LAB40 result is the fifth column, labeled "Perceptual / 40% diffusion". The imported result is shown below under LAB40 imported into QuickMenu.

ICC File format

The ICC file format stores an icon. These files are always exactly 1032 bytes long and can be separated into two parts, being the icon bitmap and the mask bitmap. Each bitmap start with a header sequence of 4 bytes (0x1F, 0x00, 0x1F, 0x00) followed by the pixel data. Each 16 bytes of pixel data corresponds to a row of the icon. The pixel information is stored on per-bit basis. The first four bytes correspond to the LSB of each pixel for each of the 32 pixels in a row. Each subsequent set of 4 bytes correspond to the next bit for each pixel in that row such that exactly 16 bytes are needed to store all the pixel information for a given row (giving a color resolution of 4 bits per pixel). The mask bitmap uses the same encoding as the icon bitmap, yet each pixel either corresponds to a value of 0x0 (transparent) or to 0xF (opague).

Conversion

The Python script in this repository allows for a conversion from a .png to an .icc. It is assumed that these .png files carry an alpha channel (allowing for transparency) and are exactly 32x32 pixels in size. Since .png files er encoded with 24 bits per pixel (bpp), a downconversion to the EGA color scheme is required. This is achieved by calculating the vector difference between the 24bpp color and each of the 4bpp colors. The color in the EGA pallette whose difference has the least Euclidean distance with respect to the original 24bpp color is then chosen. Although this conversion is automatic, it should be noted that this is not always the most ideal result and the user is recommended to make further adjustment to the final icon in Quick Menu.

Improved EGA conversion

The original converter matches colors by RGB distance using an approximate palette. This can turn the Dune II soldier's olive armor into gray or black. You can opt into the standard 16-color EGA palette, perceptual CIELAB matching, and partial Floyd–Steinberg error diffusion:

from qic import QIC

q = QIC(palette='ega')
q.png2icc('png/dune2.png', 'icc/DUNE2_NEW.ICC', metric='lab', dither=0.4)
q.icc2png('icc/DUNE2_NEW.ICC', 'png/dune2-new.png')

dither=0 disables diffusion; dither=1 diffuses the full error. At 32x32, strong diffusion can obscure small features. The transparency threshold remains alpha > 100, and error does not spread into or out of transparent pixels. QIC() preserves the original palette and conversion defaults.

Run python3 dune2-compare.py to reproduce four ICC/PNG alternatives and examples/dune2/comparison.png, with enlarged previews and actual-size views against gray, blue, and black. The original DUNE2.ICC is preserved. All candidate PNGs are decoded from their ICC files and checked for EGA colors, binary transparency, and the original silhouette. EGA has no dark olive color; pixel editing is still useful for emphasizing the helmet and armor at this size.

LAB40 imported into QuickMenu

This DOSBox capture shows DUNE2_LAB40.ICC imported into QuickMenu's icon editor, including its transparency mask and previews. QuickMenu's display palette differs from the standard EGA RGB values used in the comparison above; the imported icon retains the same EGA color indices.

Recreate the imported file with:

python3 png2icc.py png/dune2.png icc/DUNE2_LAB40.ICC --algorithm lab40

Dependencies

The scripts in this repository depend on the PIL and NumPy modules. You can install these modules easily using

pip install -r requirements.txt

Usage

Conversion

1. Clone this repository.
2. Place source .PNG files in the .png folder
3. Select an algorithm and specify the input and output paths:

python3 png2icc.py png/dune2.png icc/DUNE2_NEW.ICC --algorithm lab40

LAB40 is the command-line default. Algorithm names are case-insensitive.

Algorithm	Palette	Color matching	Error diffusion
legacy	Original approximate palette	RGB	None
rgb	Standard EGA	RGB	None
lab	Standard EGA	Perceptual CIELAB	None
lab40	Standard EGA	Perceptual CIELAB	40%
lab70	Standard EGA	Perceptual CIELAB	70%

Add --preview png/dune2-new.png to save a decoded PNG, or --show to open a comparison in an image viewer. Conversion runs without opening a viewer by default. Omitting both paths uses png/dune2.png and icc/DUNE2.ICC (overwriting that output). The Python QIC() API retains its original defaults.

Visualization

1. Clone this repository.
2. Place .ICC in the icc folder.
3. Use icc-show.py to visualize an .ICC file