updated readme

README.md

@@ -2,6 +2,21 @@
 
 This program composites many images into one big image using a set of images and a reference image
 
+### Installation & Usage
+Check the [releases](https://github.com/ultraflame4/TiledImage/releases) page for the latest version and instructions on how to install it.
+
+### Arguments / Parameters
+#### Required Arguments
+- reference_imagepath: The path to the reference image which is used to determine the look of the final image 
+- out_path: Where to save the results
+- tileset_path: The path to the folder containing the images to be used as tiles to construct the final image
+#### Optional Arguments
+- --resize-factor: The factor by which to resize the reference image. Default is -1 (auto, resizes based on tile size. Final image resolution will stay mostly the same)
+- --process-type: TiledImage uses numba to speed up computation. This argument specifies the method used to do so. Default is guvectorize
+  - guvectorize: Uses numba's guvectorize to speed up computation. This is the default method
+  - njit: Uses numba's njit to speed up computation. This is known to be extremely slow
+  - cuda: Also uses numba's guvectorize but targets CUDA-enabled GPUs. This is known to be slighly faster than guvectorize but requires a CUDA-enabled GPU AND has some overhead costs
+
 ### Limitations
 - All images in the set must be of the same resolution and dimensions
 - The final product may have black bars on the side or have clipped textures because<br/>