lwilliams3 said Bayonetta's base model was 23k without weapons. Nothing too out of the ordinary. I've seen the screenshot showing the poly count, but there's no evidence that that's the in-game model. You've outright stated that you believe it is, with no evidence whatsoever.
I believe you are aware of the difference between tessellation and normal mapping, which is why it was so odd that you claimed that the 190k poly model wouldn't be being used for baking normal maps because the Wii-U could just tessellate the extra detail.
Also, tessellation is not a simple technique to set-up. For effective use, you have to make the models with tessellation in mind. Alot of devs will not take full advantage of tessellation/displacement maps for awhile.
Your are making every bit as much assumption as you are claiming I am making and I stated nothing about my belief aside from that we have been shown that polygon count.
As I said, that is what we have and that is what I shall go by until information that states otherwise is stated. Given the listed specs of the Wii U GPU over the last gen consoles, it is not unbelievable at all. I point again to the comparison between Left 4 Dead and ZombiU.
I don't believe we know ZombiU's polygon count. Anyway, what the others are saying should not be discredited. Normal maps are used to give a lower-polygon model the same detail as what a higher-polygon model has. Quite a few developers may start with higher polygon models, then switch to a lower one with normal maps. In the case of the Wii U, the increase of RAM and GPU features over the current-gen systems will help it be able to use higher-resolution normal maps and improved lightning techniques on character models. Good use of tessellation/displacement maps may even be possible, depending on how efficient it is designed on the Wii U.
However, Wii U's tri-setup is likely 550M tri/sec (1 tri/cycle @ 550MHz), which is little above 360. I wouldn't expect polygon counts well beyond what we are seeing now. In fact, depending on far the other effects are pushed, the Wii U could see some slight decreases in raw geometry over other things. There are alot of factors involved with that theory, though, and even if that happened the Wii U's models can end up looking noticeably superior. Take the character models for GCN's RE4 vs 3DS RE: Revelations. I'm sure that Leon's model has alot more polygons than the ones for Revelations, but the use of modern normal maps and lighting techniques pushed them visually well above what was seen on the Wii.
10x the detail over current gen is believable?
That's almost how powerful PS4 is over PS3/360. There's no way Wii U is matching that [PS4].
We also don't know what the rest of the game looks like. Do you believe all the enemies and objects will have the same detail too?
Not even the PS4 would do that. If we are talking about polygons, the PS4/Durango's tri-setup are both clocked at 1.6 billion tri/sec, which is only a little more than 3x the 360s. The chances of those machines rendering close to that max will also decrease as you add more effects and other features, and them pushing higher-level processes can reduce it faster than it did for current-gen consoles. In terms of raw geometry, it will not be an order-of-magnitude difference.
In practice, though, most devs will put more detail on playable characters than on the other characters (like the Uncharted series), so it is possible that we will see a significant boost in geometry on them. It is common for non-major enemies, objects, and NPCs to have less detail. We are also at the point where there is harsh diminishing returns in using a ton of polygons for a character model over just using a lower-poly one with shaders, though, so it may be more interesting to use more polygons for the gaming world itself (example: PS4's Killzone trailer).
