Re-applying Corrections to a Warp

Working collaboratively can mean that work applied in parallel while a stitch is a work in progress occasionally needs to be re-applied when the warp changes. An inverse ppass_map can be used to re-apply corrections, such as localized GridWarp or RotoPaint, to a new version when the warp changes.

For example, if you're given a warp as shown, add paint, and then the GridWarp changes, your paint doesn't warp correctly.

The original warp.	Paint applied to remove the rig.
The same paint when applied to a new warp is offset.

C_Stitcher generates ppass_map layers for all inputs by default, but you can create the ppass_map layer using C_GenerateMap if your source image is prestitched.

Placing the C_GenerateMap between the source and the warp, passes the warp down the node tree in the ppass_map layer, which can then connect into a C_STMap node. Make sure that the Output of the C_STMap node is set to Warped src (inverse).

Using the ppass_map as the C_STMap's map input creates a faster and more accurate inverse result, which can be reused easily throughout a script.

The paint repositioned with the new warp.	The merged result.