I want to use blender to render some stuff for use with a 2D orthographic view with a scale that 1x1 blender unit is 64x48 pixels.
I set the camera up by hand (as I would use for normal 3D renders). This is fine for rendering my buildings etc., but causes slight issues around the edges of tileable objects, getting semi transparent pixels where I need solid ones (I used a simple cylinder/"pipe" for testing with the end-faces removed), as well as some unexpected shading effects. While rendering each of the individual pipe parts looks OK, if I then combine them later onto a tile grid, it becomes apparent the edges are not quite right.
I assume because my manual positioning was not perfect. I could fix those manually in an image, but would really rather not have to, especially for my more complex models.
It also seems id have to play manually with the resolution, orthographic scale, and camera position to handle different tile sizes (e.g. this pipe is 1x1, but some objects with be 2x1, 2x2, 5x5, etc.). Is there a better and more automatic way to position the camera for such rendering?
Some of the individual 64x96 renders, the edge pixels are not quite right.
When they are put together on a tile-grid.
EDIT: Can anyone explain @quiliup answer, because if I render that I get the exact same thing, what does it solve?