Eclipse depth constraint for all simulations in the IPS scenario, which each had a single pixel contaminated with time-correlated systematics. The left plot shows the results from using aperture extraction with a GP. The right plot shows pixel-fitting accounting for both independent pixel systematics (with a shared height scale for cleaner pixels) and common systematics. Pixel-fitting outperforms aperture extraction here in both precision (smaller RMSE) and reliability (${\chi }_r^2$ closer to unity) as it can weight away from the single pixel contaminated by systematics.