Pixel Back mapping from Output to input using Grid like structures

Hi,

Am trying to map the pixel on the output to input using multiple grids. Lets say my output image size is of 100*100. I want to create a grid for my output image, where are certain locations of the grid lets say 25*25 pixels is mapped from image 1 and another 35*25 mapped from image 2,.. and so on. So the output image is a blend of different images merged together in different proportions. Why mesh? Because i want to manually choose which portion of the image (ROI) is sent to the output image from each input specifically. Is there a way to involve mesh in this selection process?