The radius and positioning of my viewsheds seem to be doing something peculiar. I have set my CRS to British National Grid for both the terrain raster (downloaded from USGS) and my observation points. On the fly projection is unchecked. From the image below you see I have 2 observation points, one by the coast to the south, another to the north further inland.
My settings in the viewshed analysis plugin:
The results show both viewsheds out of place. The plugin thinks the observation points are some distance to the south of where they actually are. Not only that, but why is the radius of the calculated area egg shaped? Both of these questions I thought related to a CRS issue. I tried saving the layers to World Geodetic System 1984 and WGS 84. Both attempts led to a python error (the same CRS applied to each layer at the time). Looking at the uppermost viewshed, why isn't the entire valley shaded in? If I was standing at that point, I'm pretty sure I could see the ground around me...the plugin seems to be underestimating the results, or perhaps I'm using the plugin incorrectly.