This is fairly standardized in HDRP but in URP I can’t find clean examples anywhere in the source of how to construct what GetPositionInput does. Screen space shadows are constructing world position that’s the closest I’ve found, but it’s doing something funky with depth and didn’t seem to work plugging in the same flow to my own shader.
Actually once I figured out setting screen size from render target GetPositionInput works nicely. Although it seems URP doesn’t leverage SRP common shading nearly as much as HDRP does. Makes it a bit more work to figure stuff out.