12/11/07

The anticipated upcoming release of Poser Pro (we don’t know when or at what upgrade price) advertises to include Normal Map support.

Could anyone explain what Normal Map support would mean for the Poser User? I have read through the Wikipedia article suggested by bagginsbill,

http://en.wikipedia.org/wiki/Normal_mapping

Apparently Normal Maps are most often used for real time rendering which is done in the computers hardware (video card). Could the Poser programming team possibly be planning to harness the of the systems video card?

Also Normal Maps seem to be used more for animation in video games, as they help reduce the memory overhead of figures with high polygon counts. Some of us don’t do animations, will Normal Maps be as useful for rendering single frames?

I would have loved to see some comments from Steve at E-F about this, but I think his blog is gone for good. CP-Starlet already has a Content Paradise blog up on the Smith Micro server where E-F and CP have moved their websites. If Steve was going to reopen his blog, it would have already been put back up.

LMK

Normal mapping is mostly game technology.  It doesn't do anything that displacement mapping cannot also do, and Displacement mapping has the advantage of being able to actually raise or lower (i.e., displace) the geometry of the object whereas normal mapping cannot.  I don't see any situation where a normal map is more useful than displacement map, with the sole exception of speed at the expense of quality.

OOOoooOOO! I wonder if normal map support will be seen in the preview? if so, it might replace people using it over displacement maps.

Here's a secret. Normal maps are already supported and always have been. Load one and plug it into Gradient_Bump. Voila.

They are inferior to bump maps or displacement maps. But if that's what you have, use it now.

I'm not sure if inferior is the best term.

My understanding of normal maps, is they can deform an object in all 3 spacial directions (x,y,z), where as displacement maps only deform "up and down" from the mesh's normal facing.

I would call them different,  but not really inferior.

12/18/07

The Gradient_Bump channel does not have the same checkerboard artifact that is caused by the underlying mesh when a noise map is used on the Bump or Displacement channels. This artifact has been discussed before, and the Poser development team has never fixed this serious limitation.

Is there any reason why All three channels cannot be used? The article I read seemed to say that Normal Maps could add detail to low poly meshes, allowing more detail relative to less memory overhead.

The article also mentioned two types of Normal Maps, "object-space and tangent-space normal mapping." Which kind does (will) Poser use?

LMK

Quote - My understanding of normal maps, is they can deform an object in all 3 spacial directions (x,y,z), where as displacement maps only deform "up and down" from the mesh's normal facing.

Normal maps produce a shading effect similar to bump map, but cannot alter the geometry in profile; i.e. they won't affect shadows and won't alter the edge of the model, pretty much like bump map.  Displacement mapping can do both of those.

edit: incidentally normal maps are pretty difficult to create in contrast to bump or displacement maps.

Actually, Normal maps can affect the shape of an object. Look at some of the normal map effects in Unreal Tournament 2007, and in second life.

Second Life uses them to deform simple prim (primitives) into amazing shapes. The same might be possible in Poser pro, and some thing cool, is that tecxtures can be animated, and that might open up a lot of interesting possibilities.

Much depends on how they are implimented though.

maybe it will mean that vickie 5 will be 5,000 polygons (instead of 500,000 polygons), as they can now use normal mapping in the preview to avoid having folks freak out when seeing a blocky low-poly video-game thing.

If they are incorporating state of the art game tech, using normal maps could be a great thing!

I'd love to be able to have some of the backgrounds or props that look killer, like in UT3, Bioshock or Crysis!

Quote - Second Life uses them to deform simple prim (primitives) into amazing shapes. The same might be possible in Poser pro, and some thing cool, is that tecxtures can be animated, and that might open up a lot of interesting possibilities.

SL just uses the wrong terminology. Those arent normal maps, but what is called vector displacement maps. Regular (scalar) displacement like that used in poser and most 3d apps only uses greyscale info to do the 'pushing' of geometry. It is always along one single direction, which is along the normals of the geometry. Vector displacement OTOH lets you specify not only the amount of 'push' but also the direction. Which is what lets you sculpt the object out of a primitive in second life (in a limited way of course)

Regular normal maps as used in games are just baked bump maps. ie a bump is done on a mesh and the resulting normals are stored in a file. This way in game, all you have to do is read the file to get the same effect and can avoid the costly calculations involved in doing bumps. Even in UT3, if you take a screencap and examine carefully at the edges of models you can see where the illusion fall apart. It's just very beautiful ZBrush work so you cant tell immediately.

Dunno which version Poser pro will support though as vector displacement requires some fancy rendering techniques. Would be nice if it did of course as you can do all kinds of cool things with it like scales and hair for example.

Quote - maybe it will mean that vickie 5 will be 5,000 polygons (instead of 500,000 polygons), as they can now use normal mapping in the preview to avoid having folks freak out when seeing a blocky low-poly video-game thing.

No: all objects will now be just be a simple cube and a normal map!

Mike

12/18/07

"is that textures can be animated"

Gareee:

??? I know that textured objects/figures/models can be animated, so I don't understand what you mean that with normals maps, textures can be animated. In what way does Poser not now animate textures?

ghonma:

If normal maps have no more information in them than regular greyscale bump maps, then why do they have three channels (RGB), such that "values of each channel (color) usually represent the xyz coordinates of the normal in the point corresponding to that texel"?

I hate to admit it, but I do not understand the term "baked," which I have seen several times lately.

If Second Life is using the wrong termonology for "vector displacement maps," might e-frontier be making the same mistake?

LMK

Quote -
Vector displacement OTOH lets you specify not only the amount of 'push' but also the direction. Which is what lets you sculpt the object out of a primitive in second life (in a limited way of course)

Hmm. I have not seen the article but this sounds REALLY interesting!

Looks like you can define a morph in a bitmap, using one channel per direction (typically red=x, green = y, blue=z), and then you can:

1 - share morphs between geometries that have same UV mapping but different node order and arrangement.
2 - affect/mix morphs and vary their strength over the object using a shader tree.

In fact, it would be easy to generate such maps from existing geometry/morph definitions.

Hey I am looking forward to this new Poser Pro!

B.

*??? I know that textured objects/figures/models can be animated, so I don't understand what you mean that with normals maps, textures can be animated. In what way does Poser not now animate textures?

*in poser you can apply an animated texture (such as a video clip) to a model, so in theory, you could animate displacement or normal mapping.

Quote - 12/18/07

"is that textures can be animated"

Gareee:

??? I know that textured objects/figures/models can be animated, so I don't understand what you mean that with normals maps, textures can be animated. In what way does Poser not now animate textures?

ghonma:

If normal maps have no more information in them than regular greyscale bump maps, then why do they have three channels (RGB), such that "values of each channel (color) usually represent the xyz coordinates of the normal in the point corresponding to that texel"?

If Second Life is using the wrong termonology for "vector displacement maps," might e-frontier be making the same mistake?

LMK

What I mean, is normal maps can be used for animating. Imagine completly changing the shape of an existing object with a change of a map... not replacing it, swapping in one object for another, but completly changing the objects shape.

You could possibly also do a gradual dissolve from one shape to another, and since you can animate textures already, even the textureing could be changed easily on the fly in an animation.

Imagine a banana slowly changing into an orange as a simple example... not a 2d image replacement, but a full 3d conversion.

Quote - What I mean, is normal maps can be used for animating. Imagine completly changing the shape of an existing object with a change of a map... not replacing it, swapping in one object for another, but completly changing the objects shape.

You could possibly also do a gradual dissolve from one shape to another, and since you can animate textures already, even the textureing could be changed easily on the fly in an animation.

Imagine a banana slowly changing into an orange as a simple example... not a 2d image replacement, but a full 3d conversion.

I'd say, you could do it in Poser today. Looking at Baggins Bill's columns created from a cylinder
using displacement maps.
As you can animate almost any value in Poser's material room, it should be possible to change 
shapes using math nodes in the displacement channel.
 

ghonma and pjz99 are right, there is some confusion between Normal maps and Vector displacement maps. modern Normal maps are simply a diffrent type of bump map, the big difference is that while bump maps give the illusion of pushing the surface out in only one direction, normal maps give the illusion of pushing the surface out in different directions, by controlling normal directions, giving a better illusion that the shape is changed than bump maps. but its still not adding any actual surface detail like Displacement or Vector Displacement maps do, so it looses the illusion along the edges of the object. Thats not saying that if done right it isn't a step up from standard bump maps. I tend to doubt though that poser is adding Vector displacement maps, if they were they would make a lot bigger deal about it, becouse even a lot of the high end suites dont have vector displacement yet.

Quote - Here's a secret. Normal maps are already supported and always have been. Load one and plug it into Gradient_Bump. Voila.
They are inferior to bump maps or displacement maps. But if that's what you have, use it now.

Yes, unfortunately I disagree with Baggins on this one. I think that normal maps are better, if not superior to bump maps. It's best to render side by side to tell the difference but I think for detailed work it's noticable.
Really I think it's best to use displacement and normal mapping together to capture the full effect.

Yes, it would be killer if Poser could adapt real time normal mapping in it's preview window, could be pretty tough for the little proggie to do though. We shall see.

12/18/07

"incidentally normal maps are pretty difficult to create in contrast to bump or displacement maps"

pjz99:

The link listed in the first message of this thread references an application, xNormal, a "free and very complete normal mapper by Santiago Orgaz,"

http://www.xnormal.net/

In another thread that touched on this subject, stonemason suggested a program, CrazyBump, which makes bump/displacement maps and normal mapes from greyscale images,

http://www.crazybump.com/

I have not yet tried either of these applications.

LMK

a note - if you never take the time to actually work with normal maps and see what is involved in making them, it's hard to judge how good the results will be that come out of one of these auto-normalmap makers.  Not to say they are bad, but you simply must take some time and experiment with doing it by hand or you'll never really know whether it works the way you think it does.

Let me clarify.

A normal map defines a purturbation of the normal  - i.e. which way the surface faces. It records this information as a unit vector - which for purposes of storing as an image turns into RGB values.

But 256 levels is all you get for each, because each channel has only 8 bits. So you end up with vectors that are not perfect in the sense that magnitude is not always exactly one. Also, because the resulting normal depends non-linearly on the original dialed-in bump, you can't just willy nilly change the amount of bumpiness as easily as you can with a bump map.

In particular, I can disturb a bump map with other math functions, modulating it so that I can make it more or less in different areas (or at different times during an animation). This makes bump maps potentially much more flexible and dynamic than normal maps, which are essentially frozen-in-time snapshots of the resulting normal after applying a bump map.

So bump maps give me more control and I can generate them on the fly using math, giving very high resolution.

A normal map must be baked and stored in 8-bit-per-channel images. Not so good.

In other words, if normal maps are hard to make, cannot be dynamically changed using math/masks, and are limited in resolution, they are inferior to bump maps, IMO.

And BOTH are inferior to displacement, which as some have pointed out, I have used to create 3-D sculpted objects from otherwise featureless primitives. Vector displacement (where I could give the actual direction) would be even more sweet.

As far as I've been able to see (because I have it for Beta), Poser Pro doesn't have any of this, other than the Gradient Bump we've always had, which is a Normal map data interpreter. I don't remember what style it is. But somebody else posted a normal map built in some other tool used in Poser plugged into the Gradient bump channel. Try it.

Quote -
Yes, it would be killer if Poser could adapt real time normal mapping in it's preview window,.

.. would it really?..initially I thought it pretty cool that Poser might have real-time bump effects,but that's purely from a gimmicky point of view,in practice I can't think of any good reason to have it,& I can't think of any occasion where a real-time bump preview would be of any benefit.sure it might look cool in the preview window but how often does one render using the preview engine?

I agree with Baggins..stick with greyscale bumps & displacement where you have much more control.

Cheers
Stefan

12/19/07

stonemason:

Normal Maps are available in Poser on the Gradient_Bump channel. Does the gradient bump channel only render in the Preview Renderer, but not in FireFly?

LMK

I believe he meant Open GL Preview.

yep,,'preview window' being the Open GL display

& I believe gradient bump works in both p4 renderer & firefly

Quote - Could anyone explain what Normal Map support would mean for the Poser User?

In a nutshell, Normal Mapping would replace Bump Mapping because it's higher detail and shows more depth than Bump Mapping can produce.  Most 3D apps now can export displacement textures as Normal Maps which can be imported to other 3D apps and applied to objects.  Displacement Mapping actually generates more polygons for its detail whereas Normal Mapping is still flat like Bump Mapping is.

Quote - In other words, if normal maps are hard to make, cannot be dynamically changed using math/masks, and are limited in resolution, they are inferior to bump maps, IMO.

Well for modern renderers at least this is not a problem, because they support 32 bit per chan float precision in shaders. Even firefly renders internally in float format, it just clips it down to 8-bit when writing the file (IIRC). So it shouldn't be too hard to support them in poser pro. Also some renderers use what are called 'delta' normal maps. These are merely the difference in the actual gemoetry normal and the calculated bump normal. By using them, not only can you store the map in smaller precision (since the range is much smaller) but you can also change the intensity of the 'bump' by changing the magnitude/values of the delta. With some limitations of course. Rez has no limitation only what your RAM limits.

And normal maps arent hard to make, you just need the right software for it. ZBrush does them easily, as does XSI and MAX. Modo too i believe. It's simply a matter of either sculpting your detail and baking it out as normals, or painting a bump and converting that. There's also a very nice free tool for working with them:

xNormal

and an nVIDIA plugin for photoshop:

Photoshop normal mapper

Quote - I hate to admit it, but I do not understand the term "baked," which I have seen several times lately.

Baking simply means taking some calculated value in a render and writing it out as a texture. That way you only do the calculation once per image/animation and not for every single pixel in the image or every frame in the animation. You can bake out stuff like lighting, shadows, complex nodes or IBL, AO, GI etc. So eg if you have a scene in which only the camera is moving and the lights etc are all still, then there is no point in calculating the lighting and shadows again and again. You just calculate it once, store it to texture and reuse it for each frame.

Can save a lot of rendering time if your renderer supports it (dont think FF does though)

EF (or SM if they wanna be called that) must have some good reason for adding the feature in. Guess we'll just have to sit back and wait to see what advantage/disadvantage it has to offer?

Quote - Here's a secret. Normal maps are already supported and always have been. Load one and plug it into Gradient_Bump. Voila.

Well, yes and no. :) Normal maps completely replace the original surface normal, the gradient bump is being added to the original surface normal. The idea is similar, but they're not exactly the same. And gradient bumps are only fully supported in the Poser 4 renderer.

Quote - EF (or SM if they wanna be called that) must have some good reason for adding the feature in. Guess we'll just have to sit back and wait to see what advantage/disadvantage it has to offer?

For one, compatibility. Many game engines use normal maps, many modelers export normal maps and there are more and more standalone tools to generate normal maps from high-res geometry. For example this will open new paths for creating Poser content or using content that was created for real-time applications. And it's always good to have new options, isn't it?

  The displacement alters the geometry. So strong that it exploded the box. Bump and displacement always have artifacts, even with higher displacement bounds in render settings.  Normal map gives high detail, but it would not show this detail from the side view in the same way like displacement does.

 Anyway, that would be great to see that real time shading in the preview window of the next Poser.

12/18/07

"And gradient bumps are only fully supported in the Poser 4 renderer"

Stewer:

To what degree (in what way) is the Gradient_Bump channel supported in the FireFly rendering engine? There should be no problem in acquiring an answer to this as Poser 7 is long on the market.

If it will not cause you difficulties, I would like some information about Poser 7 Pro. The PDF file does not directly mention Gradient_Bump, only that the 7/Pro/Base versions support Bump maps. This implies that there will be no change in the Bump Map features (Bump and Gradient_Bump) as they are presently implemented.

Does anyone know, or can anyone say how Normal Maps will differ from the Gradient_Bump implementation of Poser 4 - Poser 7?

Will Normal Maps have their own plug in channel in addition to Gradient_Bump, or will it replace Gradient_Bump altogether?

LMK

Quote - > Quote -

Yes, it would be killer if Poser could adapt real time normal mapping in it's preview window,.

 

.. would it really?..initially I thought it pretty cool that Poser might have real-time bump effects,but that's purely from a gimmicky point of view,in practice I can't think of any good reason to have it,& I can't think of any occasion where a real-time bump preview would be of any benefit.sure it might look cool in the preview window but how often does one render using the preview engine?
Cheers
Stefan

Would it be practical?
Absolutely not.
Would it look cool?
Absolutely.

Here's an example of a low poly normal mapped model.

Happy Holidays

Pat

mulllltipass!

" I can't think of any occasion where a real-time bump preview would be of any benefit"

I can.

There is a strategy for an animator whose look-and-feel is 'stylized realism." It consists of getting the software and video card to produce the best possible representation in the viewport at all times. Then........you render the viewport playback as animation.

Since "the system" has been optimized for fast response while engaged in posing, etc., you are talking about less than 2 seconds to "render" the viewport. Perhaps less! Add anti-alias and your frame render rate goes perhaps to 4-5 seconds.

Now....you can't go for photorealism. Shadows are a little bit of a problem. But.....if your ultimate "magic" is in postprocessing, guess what....you've got some pretty damn good raw footage at the cost of 4 seconds per frame. Open the sequence in AfterEffects and any 'issues' due to the fact that your raw was not actually "fully rendered" can be addressed in a stylized way.

Frankly, Poser is pretty cool for this technique. I've been using it. The new 2008 version of 3DSMax has advances in viewport realization which feed this paradigm.

This casts a whole new light on video card power, OpenGL implementation and feature power -- such as viewport bump/normal/displacement  in the viewport. Basically, it's hardware-assisted render of raw frames.

::::: Opera :::::

               "you are talking about less than 2 seconds to "render" the viewport. Perhaps less! Add anti-alias and your frame render rate goes perhaps to 4-5 seconds."

do you have any examples of this..seem like a pretty basic way of rendering as your missing out on lighting & all the material effects,personaly I could never render like that,
I could render a preview like that..but with a preview I dont really care how the bumps look

the real time shadow display in max2008 coupled with dx shaders is indeed good,& that's the kind of real time effect that will aid me in creating images

if your into very speedy realtime rendering..have you looked into using game engines?using Cryengine I can render normal,displacement,sss,ambient occlusion,reflection,soft shadows etc..all in real time :)
the engine comes with the game Crysis & you can get it for less than the price of Poser

I'll post a link tomorrow..no time tonight. yes, Max now has shadows in the viewport.

All animators are after fast render time. But I want the high-poly/textured assets like V4, dynamic hair and cloth....so game engines do not spring to mind. I'll think that thru.

::::: opera :::::

being hi-poly isnt a problem in recent game engines(no dynamic hair or cloth though):
http://forum.daz3d.com/postimages/origimage_1_1047266.jpg

This discussion got me curious about something slightly sidestep to this thread.

I wonder why it is that various Poser content vendors don't appear to take as much advantage of detailing their models with bumps, displacements and gradient mapping.

I'm making couple of pieces right now, and have a fair amount of detail done as bump maps. (Fine clothing wrinkles, seams and similar minor details). The feedback I'm getting from some people is as if this is some sort of a super novel idea (it's not).

The reason I'm asking is, if I'm overlooking a major disadvantage of bump maping, I'd like to know!  One plausible reason may be that for a lot of vendors, good part of their target audience is not comfortable with bump maps and/or uses the lowest render settings?
Any thoughts?

Quote -
               "you are talking about less than 2 seconds to "render" the viewport. Perhaps less! Add anti-alias and your frame render rate goes perhaps to 4-5 seconds."

do you have any examples of this..seem like a pretty basic way of rendering as your missing out on lighting & all the material effects,personaly I could never render like that,
I could render a preview like that..but with a preview I dont really care how the bumps look

the real time shadow display in max2008 coupled with dx shaders is indeed good,& that's the kind of real time effect that will aid me in creating images

if your into very speedy realtime rendering..have you looked into using game engines?using Cryengine I can render normal,displacement,sss,ambient occlusion,reflection,soft shadows etc..all in real time :)
the engine comes with the game Crysis & you can get it for less than the price of Poser

I don't understand what you mean by missing out on materials and lighting effects.  Preview already shows approximate diffuse and specular values, with regard to scene lights.  There's already an option to use the GPU to more closely approximate transparent and procedural materials.  What I think a number of us would like to see added to preview would be bump/normal maps and cast shadows.  This is mostly an animation thing, as long per-frame render times are more acceptable for high-res stills.  In the end it comes down to what you are trying to do, but for animators working independently being able to use our GPUs to instantly render graphics of the quality of Gears of War or Ratchet and Clank: Future would be a great time saver and in many cases all we'd need to produce acceptable visuals, especially in non-photoreal styles.

Quote - I'm making couple of pieces right now, and have a fair amount of detail done as bump maps. (Fine clothing wrinkles, seams and similar minor details).

sounds awesome. that's not something commonly seen. seems that most people, if they put in seams either model them in (rare) or paint them into the texture.
 

Quote - The feedback I'm getting from some people is as if this is some sort of a super novel idea (it's not).

You would think it's not but compaired to many of the products out there now, it really is.

I picked up a product the other day that had baked in highlights and shadows for details in the texture and  used a desaturated version of the texture as a bump.

Quote -  The reason I'm asking is, if I'm overlooking a major disadvantage of bump maping, I'd like to know! 

I know there is potential for problems with displacement, but i have never heard of anything in bump maps causing problems.

Quote -  One plausible reason may be that for a lot of vendors, good part of their target audience is not comfortable with bump maps and/or uses the lowest render settings?
Any thoughts?

I think most of the target audience doesn't know how to use bump maps properly and when they decided to make something and sell it, they don't use them or do them incorrectly.

Quote -  I don't understand what you mean by missing out on materials and lighting effects.  Preview already shows approximate diffuse and specular values, with regard to scene lights. 

What that meant to me is that if one is using render settings so low that all it takes to render a frame is a few seconds, they're missing out on the potential image quality and appeal.

When I do some of my architetural viz animations, sure, I can turn the quality of the render down to render things at seconds per frame, but what I get is unacceptable presentation quality. For that target audience.

When I use more refined lighting and textures, the render time per frame is counted in minutes and tens of minutes or more (in othe older days), rather then seconds.

Quote -  I think most of the target audience doesn't know how to use bump maps properly and when they decided to make something and sell it, they don't use them or do them incorrectly.

 
Seems like it's not just the users, but a fair number of vendors too.
I can see it with displacement maps, you actually have to turn them ON in render settings (default render setting has them off)

It's possible I missed something, but I didn't see the way to globally turn bump maps off - short of disconnecting the material node or setting the dial to 0.  heh, well, that'll be a learning experience once the thing is out.

thank you, bagginsbill !  It is awfull, I never knew about this feature in Poser. It works so fine! 

If I only knew :( everything would be different.

Merry Xmas and Lucky '08 to you, bagginsbill 

!

moogal >>I don't understand what you mean by missing out on materials and lighting effects<<

I meant cast shadows and bump maps and certain shader effects, which are not actuated in the viewport at this time. Sorry I didn't explain that more accurately.

Connie, I am not talking about reducing settings on Firefly in an attempt to get the time down to seconds....I am literally speaking of NOT rendering Firefly at all: just rendering Preview.

Rendering out to lossless image sequence with Preview gives me a rendertime per frame between 1 and 5 seconds (the higher time is with anti-alias turned on.) Since I am going for realism (stylized), I want as much quality in the viewport (preview) as I can get. You can't have cast shadows in the viewport in Poser at this time (3DSMax v2008 just got it). But since I am also going for 'stylized' realism as opposed to the more demanding photorealism, the preview result is sufficient, since I can apply filters and effects in AfterEffects, which might add another 5 seconds or so per frame there.

If I can get a look that works...hey, these are great and necessary render times for the small studio/artist/filmmaker.

moogal >> There's already an option to use the GPU to more closely approximate transparent and procedural materials.  <<<
What option is this please?

::::: Opera :::::

Quote - > Quote -  I don't understand what you mean by missing out on materials and lighting effects.  Preview already shows approximate diffuse and specular values, with regard to scene lights. 

 

What that meant to me is that if one is using render settings so low that all it takes to render a frame is a few seconds, they're missing out on the potential image quality and appeal.

yup,that's what I meant..the quality of an ogl preview render would probably be in line with a Poser 3 render(transparency & color maps are about the only material effects you could achieve,(did poser 3 have transparency?))..but hey,if that's good enough for a 'stylised' render then it's all good.

re bump & displacement..I dont buy any content but I'm pretty sure everyone uses bump maps these days,displacement seems to be getting wider useage also.
I've been using displacement in commercial models for a couple years now but one thing that drives me mad is the end user making a point of either turning it off..or not turning it on in render options.or not using the firefly engine.or even rendering in another app & not setting up displacement correctly.

you should include it though,when done right it's miles better than a regular bump map

Cheers
Stefan

Attached Link: Click to download short animation, 4 MB Quicktime

Notice that that are clever ways to get dynamic cloth AND dynamic hair actuated, even with Preview render!

my two rendering stations are busy for the next few days....so I can't show a GREAT example of stylized realism achieved at less than 10 seconds per frame combined raw and postprocessing time.

This little red dress animation is a "hint" of what I'm shooting for. Probably too much "film grain" but still, I like it. 

Working with this paradigm does several things: it invokes realism as opposed to Toon; it avoids the Uncanny Valley; fast render; no limit to creativity in post processing.

I am sinking into Max v2008 now and plan to really push this idea. I still like Poser and Carrara for it, as well.

::::: Opera :::::