Problems with Sculpts & Sculptie Maps

Sorry if this question has already been asked. I am trying to learn building, especially furniture. I recenty started making sculpts from sculpt maps and bought some sculpts to create items. Everything looks good from one angle but when the avatar (or the camera) move around, the sculpts distort totally out of shape.

Also when applying a texture to a sculpt cushion or pillow, if the texture is stripped it becomes spiral or circled.

I have altered the settings in Advanced - Debug as recommended and this does solve the distorting problem but even though I now see the object looking okay, others will still see it dstorted. Can anyone help please ?

Sounds like whoever made the sculpt maps you're using didn't take precautions to "LOD-proof" them. The number of vertices in a sculpty is not constant. It drops as the camera moves further away. There are certain key points that always remain, but the ones in between disappear when the camera moves away, and reappear when it gets closer. Those in-between points are meant simply to help add detail when the object is viewed from close up, detail which likely wouldn't be visible anyway from further away.

The problem is people who don't understand how the LOD culling works will often use the in-between (tween) points incorrectly, for corners or for other important structural details that end up killing the shape if they disappear. The only way to avoid that is to make sure your basic structure is made from the key points, not the tween points. Sometimes that will mean making your object out of more than just one sculpty, which is fine. That's how it's supposed to work.

"Pixel pushers", as Qarl Linden affectionately refers to them, often try to squeeze every last detail out of every last point. This is OK if you're absolutely certain the object will only ever be viewed from up close. But if there's any chance at all it will be viewed from a distance, it's best to be more conservative. Throw in a few more prims, and call it a day.

For better or worse, sculpties have constraints, and it's on us as sculpty-astists to make sure we operate within those constraints. Sculpties that appear to malfunction as you described need to be remade, with the right points in the right places.




Remember, no matter how it's been bent or distorted in 3D space, the default topology for a sculpty is spherical. If you put a striped texture on a regular sphere, the same thing would happen. Rotate it one way, and you get concentric rings of latitude. Rotate it the other way, and you get converging lines of longitude.

If you've distorted the sphere into something more cubical, like a sofa cushion, you can't just throw parallel lines on it, and expect them to remain parallel as they would on a real cube. They're going to wrap around the surface, just as they would on a sphere.

You can try rotating the texture 90 degrees. That MIGHT do the trick, depending on how exactly the cushion was made. Chances are it won't work, unless the cushion was made in a very specific way, but it's worth trying, since it only takes a second.

If you want stripes on a sculpty, you have to create the texture in such a way as to counter the distortion from the surface wrapping. You can do this quite easily in any full featured 3D modeling program like Maya, Max, or Blender, or in a 3D paint program like Photoshop CS4 Extended, Zbrush, or Mudbox. But doing it in 2D on a flat canvas will be difficult. It's doable, of course, but it takes an intimate understanding of exactly what's happening with the surface as it wraps in 3D space, and precisely how the changes in 3D shape translate to counter-changes on the 2D canvas.

The Polar Coordinates filter in Photoshop, Paintshop Pro, or GIMP might help, but again, only if the cushion was made in a very specific way.

To see what the surface is actually doing, throw a test pattern on it, like this one: /109/45/214824/1.html

If you make any assumptions as to where those key points are, you are already a pixel pusher by Qarl's definition.

In fact it's not necessary to know which points remain and which ones disappear. All the sculptor needs to know is that each vertex line in a non-oblong sculpted prim has a 50% chance to survive a LOD reduction. Anything beyond that rule is implementation-specific and subject to change. See here: /8/20/362562/1.html

The only way to make a sharp edge LOD-safe in a way that does not depend on viewer implementation details is to double its vertex line. This will make sure that one of the lines survives the LOD change and the edge remains sharp.

It also makes sure that the texture jumps during LOD changes. It'll change between using 1 face worth of textures to using 2 faces worth of texture when the doubled vertex line is removed in the viewer by LOD. I find this more annoying than the smooth shading it tries to avoid

It's all a matter of how you distribute the vertices across the object shape. Of course, if you put very small faces next to very large ones, the texture will jump a lot when the vertex line between those faces disappears. But if you gradually increase and decrease face sizes, the jump will be minimal and hardly visible.

Blender has some tools to support this way of modelling, e.g. mesh smoothing and proportional editing.

I recently tried to make a texture using some fine details with highlighting. At the first LOD Jump i see that some of the details suddenly drop, others get blurred and increase in size. I suspect that this sort of texture jump is not avoidable? Or am i doing something wrong here ? BTW there is not a single duplicated line of vertices in this example. The whole mesh is smootgh. Only near the edges i narrowed the loops a bit to get the rounded edge of the glas stand(?) Probably that is where i need just another line aside to stop the texture to blur up at LOD 2 ?

oh, i think this answers my question... i did not realize that it was allready posted

Try using a diagonal stripey texture to test it. This technique also means that texture baking isn't going to work effectively and you'll have a lot of adjustment work to do.

It's funny that you felt the need to explain Blender's toolkit for me

Hi Gaia,

That sort of texture distortion is caused by uneven faces. It's pretty much unavoidable on anything but very simple shapes. Subsurf modelling to keep face sizes even is usually the best way to minimise the effect. The texture jumps I'm talking about would be if you had placed two rows of vertices on top of each other to give a sharp edge around the base of the glass. The texture distortion you are seeing would then be doubled in size giving a much more noticable problem.

I can assure you that texture baking works just fine with this technique as long as none of the faces has a size of zero.

I'm not saying that this technique magically resolves all the design flaws of sculpted prims, but if your sculpties depend on implementation details such as the way sculpt maps are sampled at lower LODs, they will break when the implementation is changed, and the breakage will be worse than a texture jump.


It's funny that you think I should know who is in need of an explanation and who is not. This is a forum thread, not an email correspondence. Any piece of info we drop here may be useful to many people.

Then you aren't really doubling a vertex line. That would give zero sized faces.



Nah, it's just funny (to me) that a Blender user doesn't recognize my name as the author of the Primstar scripts

If you position two vertices very close to each other in Blender, the sculpt map's 8-bit resolution will effectively double them in SL. So the faces will be zero sized in SL, but not in Blender, and texture baking will work normally.


I don't use the Primstar scripts, so please forgive my ignorance.

Fair enough. I think we've covered the pros and cons of this method now. Personally I prefer to have a sculptie model with fairly even sized faces on the LODs and a more detailed model with the sharp edges for texturing. I use bake selected to active so the sharp edges are in the texture rather than the sculptie.

All of this is advanced stuff though, so not much use to the original poster

Now the average user gets quite curious about what that means... and asks google... and found a partial answer in the blender docs. But as way too often there is yet the next question under the hood... I understand there are 2 objects involved. The low poly sculptie and the high poly "best world" model. So how do i instruct blender to use the high poly model to create the texture for the low poly-sculptie ? which option in which window under which screen in which state is it ?

meanwhile i found this one:
http://www.veoh.com/browse/videos/category/educational/watch/v6379634HgSpwjCM

The crucial point seems to WHERE the 2 objects are located relative to each other! And i also found out how to specify the active object: It is simply the last one you have selected in object mode. So you will need to use the SHIFT key to first select the high poly mesh, then the low poly mesh, then bake. The video still opens up another bunch of questions ...

Hello

Ive tried to find an answer to this simple problem, hoping someone can help.
When image texturing a plane or hemi in blender I can only get the image to apply on one side leaving the othe side clear in world. I have the Double Sided button depressed in the mesh control. How can I get the image to render on both sides of the plane.

thankyou in advance

Michrd

short answer: You cant, sculpties have only one side.

middle length answer: create your sculptie in world with only one side textured. Then create a duplicate of it and either check the inside-out button for the duplicate in world or create an inside-out sculptmap in blender (s x -1 would do it) locate both sculpties at the same position and your inside is textured.

long answer: http://www.vimeo.com/8504087 In that video i show you how you can get rid of the problem by folding your mesh in such a way that it looks like it had double sides. I do not show how you make a "true" double sided sculptie, but how to make a sculptie with an optimze use of vertices where they are needed...

more answers will come

People other than the OP are reading this thread. And even beginners who are trying to learn - or even just thinking about trying to learn, can pick up useful information from a discussion of the advanced, finer details.

It may not have an immediate impact on our lives, but just the vaguest of memories that a subject has been discussed can be useful when trying to deal with the problems and questions that will eventually arise.

@Gaia: Baking geometry to textures is a standard exercise for game content designers. Usually it is done to generate normal maps, but you can also bake a "full render" that works pretty much like a photograph: Anything that is floating below the surface or within a certain distance above it (check out the "dist" and "bias" fields below the "bake selected to active" button) will be rendered onto the texture of the target mesh. The point of view of the camera is always opposite the direction of the surface normal of the face that is being baked to. This is ideal to give textures a 3D appearance that doesn't look "painted". Here's a (rather basic) example:

http://insilico.ning.com/photo/tin-girl

The skin of this robot avatar was not painted but modelled on top of the SL avatar mesh, then baked with ambient occlusion and specular reflection turned on. All the "metal" parts and the rings around the waist were actually separate geometry layered above the mesh. The eye cover is an oblong sculptie whose texture was baked "in place" so that its lighting matches that of the skin. No manual 2D postprocessing was performed on the textures; they came out of Blender just like that. As I said, this is only a basic example, but I guess you can imagine what this technique can do in the hands of more creative individuals than me.

If you are looking for inspiration to make another Blender tutorial, consider geometry texture baking, because it's extremely useful to offset the limitations of sculpted prims.

@Masami: Thank you for the clarification and hints on the texture baking issue!
indeed... thats where i am investigating right now. It might find its way into the wineglas tutorial, but i doubt that it makes sense to use this technique for a wine glas So it will take a while as there is a huge load of tutorial reviews to be done beforehand.

Uh oh... Maybe I shouldn't have mentioned how much wine I was drinking at the launch party.. Is the mini Dom going to be drunk in the new tutorial? Nah don't tell me.. I'll wait to see

So one workflow could be something like this :

1.) Create the sculptie model
2.) Create a copy of the sculptie, there refine the model
4.) relocate both models to the same position
3.) scale up the refined model a bit until it fully encloses the sculptie
4.) goto object mode. There select first the full model, then SHIFT-click the sculptie.
5.) go to edit mode (not needed, but ensures that the sculptie is the active object)
6.) Bake full render

Now the texture generated for the refined model is projected on the sculptie with all the details coming fr0m the full model baked into the image.

Is that a good workflow which i can use as the base for a tutorial ?

----

Furthermore, if i would not "Full bake" but "Bake normals" then the resulting texture would probably later be used in the texture stack of the sculptie. Is it correct that the normal map "simulates" the lighting environment of a high poly mesh and allows the baker to create the shadows and light effects on the low poly mesh "as if it where a high poly" ?

If that is the case, then i could derive a high poly mesh fr0m my sculptie using subsurf, then apply the subsurf modifier, refine the model, then bake the normal map to the sculptie. And finally bake my sculptie texture using the normal map... And that would allow me to keep my sculptie as it is, do NOT add extra subsurf to it just to bake a higher resolution texture ?

If all that is true, then it makes much sense to me to always use two models: The sculptie itself and a refined model for texture baking, then use "Bake sculpt mesh" for the sculptie and bake to active for the sculptie texture ... That would keep my sculptie always correct. No more "temporary adding of subsurf" for the texturing and such ...

Can you confirm this, or clearify ?

I should create a magic flask for the "Dom in the bottle" and keep it always in reach Hey, that just gave me a very magic idea! hmm, it may take a while to get that to work...

It's best to keep both models the same size and adjust the "dist" field for any bumps, ridges or floating geometry in the refined model that stick out of the sculptie.


This is what you would have to do when baking normal maps. In full render bake mode however, all the objects in the visible layers of the scene will be included in the bake (if they are within "dist" range) whether they are selected or not. Just activate the object that you want to bake to.


Normal maps are not really useful here until SL allows uploading them. The only purpose of a normal map is to speed up repeated rendering (such as in animations). It won't help in creating sculpties.


Normal maps affect diffuse and specular shading only. Shadows and ambient occlusion won't be affected because the surface detail added by the normal map is fake; the faces of the mesh remain as flat as they were before.


The only way to bake a higher resolution texture is to use a larger target image for baking. Unlike normal rendering, the bake function does not perform oversampling. Subdividing the UV faces won't change this. To reduce aliasing, bake to 2048x2048 and scale the result down to 512x512 using bilinear interpolation.

I want to thank all of you for taking the time to reply to my question. As a "newbie" builder, there were some parts I didnt fully understand but I am sure a lot of others have read your replies to this post and will have gained valuable knowledge from them.

Thank you once again.