The autumnal glow is still here even with the Google images. I thought it was just a feature of Bing - but not so. You have to look a bit more carefully at the orthos here in Wales. There are some bad patches in both image sources.
The terrain mesh here is 2 m converted to level 13. I did try adding some 1 m level 14 tiles in one place where they were available but they didn't seem to make any difference. Surprisingly the 50 m mesh base layer (level 11) has given me trouble in a couple of places. I was getting deep vertical-sided trenches about a km wide, several hundred metres deep and many km in length. At first I assumed that I'd accidentally geoconverted some "no data" regions in the 1m or 2 m data, but I checked and found that not to be the case. In the end I realised that it was the "OS Terrain 50" data that was causing the problem. When I removed a couple of the level 11 tiles derived from that data the problem disappeared. If the high res data were complete you wouldn't see the trenches at all close up. You only see them where there are gaps in the high res data. Consequently if you remove high res TTH tiles thinking they are the culprit, it actually makes the problem worse. I think it's happening where two 100x100 km OS grid squares meet - but I'm not absolutely sure. This is a bit surprising because I would have assumed that the Ordnance Survey data was perfect.