I wouldn't expect a lot difference at that aperture stop, and that image size. When looking at the bigger versions the first one is sharper (it could actually be enlarged more than the second image) and has a cooler tone and typical G20mm bokeh signs. The CA... maybe you used an Oly camera for this?
Maybe if you show us what you get at f/2.8, not to mention a comparison at f/2 or faster...
For me comparing these two lenses mainly has to be about speed. If you are happy with an f/2.8 lens and don't print big just pick the one suiting you best.
Then we have these photographers used to a certain focal length, like Streetshooter, and then that may be the most important factor. I know I prefer a 25 to the 20, for example. (And I prefer the 20 to the 17 for obvious reasons.)
Jonas
Both images are uploaded at the same pixel size, 3459 x 2594, the minor differences is due to my approximate cropping. When you click on the image it goes to the gallery view of an image, if you click on it again in gallery view, it opens it up in "full size", it took me a while to realise you could do that in the gallery.
The question was the sharpness of the lens, not the FOV or DOF so that's why I cropped and scaled to the same size or it would be unreasonable to compare them, i.e. if I took a picture with a 14-45 zoomed into 45, I would get more detail than a 20mm lens, not because it's sharper but because of the FOV. I think when you view the full size images it's quite apparent the 20mm is actually sharper, the point is to let the "buyer" decide if they want to pay $350 for a second lens for that difference or not.
I agree that if I had no lens, I would seriously consider the extra $100 on the 20mm lens. But if I had a lens already, (given hind sight) I wouldn't pay the extra $350 to get a second lens.
As for F2 etc where the 17mm doesn't go, the OP was already aware of the 1.7 v.s. 2.8 so that was already clearly explained as an advantage for the 20mm, as was the FOV advantage of the 17mm lens.
According to dpreviews tests (take it for what it's worth
Lens Test) I believe the sweet spot for the 20mm is roughly F2.8 to 4.5 and the 17mm is F4.0 to F5.6 so, I think 3.5 is not too jaded of a choice for either lenses, possibly a slight disadvantage to the 17mm if anything.
This wasn't meant to be an end all be all test. It was just a quick snap of the same image with both lens. I know I kind of wanted to know that when I bought the lens. I read all the hype about it too, that's why I bought it. After I bought it and saw the image results, I was slightly kicking myself for having spent the money on another lens which was "so close". With hind slight, I would have rather put the money towards a lens that out right does something different like a 100mm or a 9mm lens, not something which is pretty much the same, just a little sharper and 3mm fov difference. Sometimes the lessons I learn can be ... expensive.