Compare each source image with the same stationary target. In blink mode, watch the ventricular walls and cortical folds: less movement between frames means better local correspondence. Use Toggle target/source to compare without animation.
Both panels blink in sync. The target frame is identical on both sides. Views use fixed anatomical coordinates in both directions at 1 and 2 mm; they were not selected by improvement. Left hemisphere is on the left in axial and coronal views; sagittal views run posterior to anterior.
Each volume is divided by its own 99th-percentile intensity inside the target brain mask, following the original campaign. Grayscale is fixed at 0–1 and differences at 0–0.30 across all views. No local contrast enhancement or display smoothing. Mean residuals use only target-mask pixels in the displayed crop. The full crop is shown, including mask boundaries.
Interpretation: color fringes and blinking show both geometric and intensity differences. Residuals are not anatomical error in millimeters; interpolation and template contrast can also change them. Better-looking alignment does not replace independent anatomical review.
Visual supplement to the frozen 7 October 2026 campaign. Original numerical results and adverse findings remain unchanged. This file works offline; no server or external assets are needed.