Extended Data Fig. 3: Graphical workflow for distortion correction using BigWarp.

1) Correction can be performed in BigWarp using either a pre-expansion image or a virtual reference grid generated from the pattern design file. 2) Correction should be performed by selecting the post-expansion image as the ‘moving’ image, correcting to the ‘target’ reference image. 3) Correction is easier when the moving and target views are aligned, and suitable lookup tables are selected to better visualize differences between images. 4) At least 4 corresponding landmarks (GelMap grid intersections) between the moving and target images should be selected to allow for initial alignment. 5) Initial alignment and basic correction are performed based on the initial landmarks. The moving image will be rescaled based on the dimensions of the two images. 6) Further correction can be performed by adjusting the viewer to allow more landmarks to be selected. 7) Toggling between transformation techniques allows the deformation to be easily visualized in an overlay of the two images. 8) Export the deformation-corrected non-linear thin plate spline transformation. The moving image should be exported with the resolution of the target image to avoid downscaling of the moving image due to the expansion factor.