Refactor the separate bitmap + transform data types that are run separately into a new Image data type composed of bitmap + transform data. All nodes work on Image data. No need to route the transform data separately. This makes the Blend node able to work on Images even if they're not the same size. It lets us also entirely remove the concept of a transform belonging to the node graph layer. Then the Output node can generically take Vector or Image data, and in the Image case, it runs it through an Image Frame node automatically (thus solving the issue with previewing a bitmap node causing an error for users currently).
Refactor the separate bitmap + transform data types that are run separately into a new Image data type composed of bitmap + transform data. All nodes work on Image data. No need to route the transform data separately. This makes the Blend node able to work on Images even if they're not the same size. It lets us also entirely remove the concept of a transform belonging to the node graph layer. Then the Output node can generically take Vector or Image data, and in the Image case, it runs it through an Image Frame node automatically (thus solving the issue with previewing a bitmap node causing an error for users currently).