In G'MIC, each image is modeled as a 1D, 2D, 3D or 4D array of scalar values, uniformly discretized on a rectangular/parallelepipedic domain.

The four dimensions of this array are respectively denoted by:

There are no hard limitations on the size of the image along each dimension. For instance, the number of image slices or channels can be of arbitrary size within the limits of the available memory.

The width, height and depth of an image are considered as spatial dimensions, while the spectrum has a multi-spectral meaning. Thus, a 4D image in G'MIC should be most often regarded as a 3D dataset of multi-spectral voxels. Most of the G'MIC commands will stick with this idea (e.g. command blur blurs images only along the spatial xyz-axes).

G'MIC stores all the image data as buffers of float values (32 bits, value range [-3.4E38,+3.4E38]. It performs all its image processing operations with floating point numbers. Each image pixel takes then 32bits/channel (except if double-precision buffers have been enabled during the compilation of the software, in which case 64bits/channel can be the default).

Considering float-valued pixels ensure to keep numerical precision when executing image processing pipelines. For image input/output operations, you may want to prescribe the image datatype to be different than float (like bool, char, int, etc.). This is possible by specifying it as a file option when using I/O commands (see section Output Properties to learn more about file options).