You have TexCd so you have to use TexCd instead of up or do:
Float4 uv = TexCd;
Then when you multiply with tTex you have to use uv instead.
But when you do that it will shift your left and right border off whole mesh, so what you have to do is to separate your grid on subset for each wall and move textures with texture transform accordingly to what you have.
Basically when I do such stuff I use bezier grid with 2x2 control points for one bezier on each wall segment, then you can use this shader to move edge of each texture so it aligns.
Assuming that the walls are nice rectangles, I am wondering why you are not using Homography instead of grideditor.
then it would be pretty easy to map the images to the right walls and then use texture transforms to place the parts of the video on the right walls.
As image for adjusting, I normally use the grid itself with no texture and set to wireframe. if you select the resolution of the grid, it is super useful for edgeblended setup. you can align the grids from both projectors perfectly this way and also have an easy to calculate overlap. I know this isn’t a flat edgeblended projection… but the grid is quite helpful to see if the image gets distorted
For this setup, I would use 3 quads with homography and texture transforms.
here @karistouf i made you small example, just connect that to renderer temptarget, and than that texture temp target will give you to your grid texture input, hope that helps mapping-pano.zip (2.5 KB)