I am working on an idea to create very nice textures, especially ground textures from 3d-scanned grounds. I can make them big, and cover more space than most conventional ground textures can.  I have experimented with actually baking them in Poser..

I take the 3d scanned ground, and put the camera facing it, setting the ambience node to max, connecting with the texture, deleting all lights. Works perfectly. Then I render it with auxilliary options, normals and depth.   The depth and diffuse works really well, and all I have to do is to take the resulting layered image and make it seamless.

However, the normal maps that come out of Poser's renderer are really strange.. I was thinking it would be nice to dabble with normal maps (combined with displacements), but I am still quite new with these..   The thing is, I get really strange colors from the normals when rendering them with Poser's auxiliary output, and I am also trying to understand this whole deal with tangent space vs object space normal maps.   Tangent space seems to be standard, but I have no idea how to get the right color scheme..The reason can be found in the attached screenshots..  They all give me -different- colors, depending on the camera I chose!  (I have faced the ground prop fully towards the camera in each screen, but each camera yields different colors from the normals)  So which one (if any of them) has the right colors for a normal map?  The Right camera has very typical normal map colors, but is it correct, and why would it be that angle?  Weird huh?

Back camera:

Bottom camera:

Front camera:

Left camera:

Right camera:

Top camera:

Every time you "faced the ground prop fully towards the camera in each screen,", you pointed the prop in a different direction, resulting in different normals.

I think you think you're rendering a tangent space normal map, but you're not. The normals you're rendering are absolute world-space normals. This isn't even object space, it's something else altogether. It's the actual direction in the final orientation, which is of no use to anybody.

Ah thank you for the explanation, Bagginsbill!  I will either drop normal maps for this project then or use different software, like Xnormals for it.  I suspected it was not creating proper normal maps, but I wanted to understand better what was going on. Thanks for the quick reply. :)

Note that it may be possible to do what you want. The light blue one looks right but I haven't tried such a thing to see how you'd go about it, nor tested the results.

On the other hand, in Poser it's easier and more powerful to use a height map (as displacement) than a normal map (which is only bump).

I might try it with the Right camera, but as you say, displacement map may work better, and it was super easy to make too, from the 3d scanned model.

Just my two cents, which may only be worth one, but don't expect any refunds...

The way most people seem to use normal maps with Poser is as a displacement map, using a texture map that has been converted into a normal map, which basically does nothing other than add a sort of bump map conversion of the original texture map. And, given how Poser users these, there's really not much of a point at all to using that particular sort of normal map. You're probably better off tweaking a displacement and/or bump map, as BB suggested.

This is sort of alien to me. I see Poser bump and displacement maps as 2D maps derived from an image file specifically designed for that purpose. A normal map, on the other hand, is something I've always associated as being generated by geometry, not textures. So, when I see people "creating" normal maps by shoving the object's texture map into Photoshop or another 2D app, I sort of scratch my head. "There's a bump/displacement channel for that... what's up with that?"

In the OP, it looks like an attempt to generate "something like" a normal map using displaced geometry and different rendering angles. That's a bit closer to the mark, but it may not work well, as has already been addressed. (I have no idea what Poser's capabilities are above PP2012, so if later versions can generate normal maps or have some more utility there, potentially allowing them to be created... I'd be interested!)

To me, the best use of a normal map is simply taking the original object, increasing its resolution/density, fine-tuning detailed geometry and sculpts, then rendering the high-poly version of the object in something like xnormal, resulting in a high-res normal that you can then place in the normal channel on the lower-res model, getting a potential two-fer :) in memory footprint, but with the added benefits of a normal map.

In other words, I am confused by the idea of generating normal maps from 2D maps, almost always the original texture map image, then trying to plug them into the normal channel as if that was going to do much more than a regular bump map. (Just wanted to clarify - That doesn't appear to be what the OP is doing, since they're combining displacement, which effectively increases the object's geometry, making it more useful in generating a normal map.)

On "Baking" in Poser, which was mentioned in the OP. CAN some standard version of Poser bake procedurals? !! I imagine the game-dev version can or should, at least, but can an out-of-the-box version do it? Again, I'd be intensely interested if it can OR if there is a script/way-to-do-it in Poser. (I know Carrara can do it, with a plugin.)

Oh, on tangent and object space, this is a pretty simple explanation: http://blog.digitaltutors.com/bump-normal-and-displacement-maps/  And, some more technical details in this answer: https://gamedev.stackexchange.com/questions/31499/what-are-the-advantages-of-tangent-space-normals-over-object-space-normals

Actually, I am not generating the displacement/normal map from a 2D map.  It's a 3d scanned ground plane, which means it has high resolution geometry information.  It turns out though that just rendering the depth map information was what was needed.