Selecting encoding options

Using an artificial neural network to generate a result that indicates one or more encoding options 408 to use when encoding all or part of a graphics texture 402. The neural network preferably indicates a partitioning of the image or texture data for encoding. Preferably the training of an artificial neural network for this encoding is done using an intermediate error metric 418. The training method reduces the intermediate error based on the difference between an intermediate result 412 generated at an intermediate or middle layer 404c of the network and a target intermediate result 414. In embodiments the intermediate result represents a partition bitmap, and the final output is a partition vector representative of the partition. Preferably the encoding scheme is Adaptive scalable texture compression (ATSC) for graphics texture data.