Very close to the original, especially the blurry part, but if you think this is bad, just take a look at the videos from '92, it'll make you eyes bleed (mine still are).
IL v2: Same dataset but reprocessed it, and adjusted training settings, less contour bleed and blur also I wasn't expecting a more accurate face. Style suffered a bit, but in my opinion its for the better (its not that flat anymore), so I'll take it!
Outfit: leotard, red cape, elbow gloves, thigh boots, high heel boots