How the archive is made
Methodology
Each reconnaissance photograph is produced from an automatically modified IL-2 mission adapted for automated aerial photography. A Python processing pipeline captures and combines four image tiles into a high-resolution composite. Mission data, workbook reference data covering priority targets and the supply-point model, and custom code are then used to orientate and prioritise target groupings before annotation. The annotation intelligence is still being refined. In its current form, the system is designed to present up to four target areas on any individual image. A current limitation is that the annotations are based on Day 1 data, representing targets before they have been attacked.