blender tutorials (text and video) for sculpties and more ...

Ok, ok... five minutes more and i found the answers myself...:

i found the shape sliders and can manipulate the avatar mesh ...

1.) in the outliner window: select "avatar"
2.) in the 3D view right click on the avatar element, you want to shape
3.) ensure, that you are in object mode.
4.) selecte the edit buttons (F9)
5.) In the far right you find the shape keys. click on the arrow keys to open the shape slider.
6.) at the bottom of the outliner window you find the recording system. drag it up, so that you can see it better.
7.) now click in to the canvas of the recording window. You will set the keyframe. At each keyframe manipulate the shape sliders to your taste.
8.) After you have set several keyframes, each with different shape slider values, you can play your animation (hit the play button) Watch what happens in the 3D window.

... ok, thats amazing... thats wonderfull ... Now everything i ever wanted to make with
an avatar mesh in blender can be done yeahhhh!!! ...

Now... what is the intention ? Making animation videos ? making animated sculpties ? or just knowledge transfer ?

where to go from here ?

My very crucial first and biggest question now:

How can i export the body parts as sculpties ?

If that is solved, heaven is near

The record button is on the timeline window, which is at bottom left as you found out

At this point, there's only limited functionality in the avatar.blend and it's really only useful as a better base for avatar texture baking. This will change as i continue to work on the import script.

So to use it for this, you don't need the full timeline, just press the red record button and in object mode start selecting shapes and adjusting their influence. As you are recording all changes to frame 1, you can blend different shapes to approximate the outfit in SL.

I want to start spreading the knowledge on this early as there will be an incredible amount of options in the final version. This is really a separate project to the sculptie scripts, but I do expect them to be used together. Once I fix the attachment point positions, then clothing makers can attach a sculptie to the avatar and edit it in place as part of the outfit.

At some time in the distant future it should be possible to import the avatar, sit them on your sculptie sofa, edit their bone positions and export a bvh file for the sit target.

I think the SL photographers might also want to use it for studio renders and of course the machinima possibilities will be astounding

As far as avatar to sculptie is concerned, that's still a UV mapping task. I've no plans to automate it at this point.

OK, i see... I will create an early birds video on this topic. But i want to put some more sugar candy into it, although i dont yet know exactly, what that will be... (my secret dream is still to get the "mesh to sculptie" done, at least for the head-part of the avatar)

Anyways, please can you give me a more detailed list of planned/proposed/thinkable
features of your final product?

Keep at it, the journey will be rewarding.. Mine was Suzanne, the video was a very early attempt, that I wasn't really satisified with. It took a lot of experimenting to get the final unwrap that I used, and even then I had a bit of cleaning up of the sculptie to do afterwards. What it did do is teach me a lot about the uv mapping features of Blender. And lead to Add - Mesh - Sculpt Mesh being added to my scripts because I decided I really didn't want to have to teach people the hard way to do things



Even if I limit myself to the avatar script, it's a very long list

Also until I get to doing them, it's hard to say whether a feature is viable or not. But basically if you think full avatar support in Blender you won't be far off. So avatar shapes, textures, animations & attachments are planned. That's the short way to describe it, watch the website for specifics as each bit goes in..

Got another question, this time texturing related.

I'm using your Snap-to-SL grid setup to construct a fairly low-res model. It has worked out in such a way that to get the slopes/angles I want (and to be resistant to LOD), I would have to compress several levels of vertices together.

the attached pic shows a side-view the offending area. At low LOD, the columns are just one face. I have yet to do the sides of the mesh, but I am certain that they will not come out as well due to the curvature of the whole mesh. The easiest solution for me would be to move all the circled verticies vertically to one plane.

Buuut will this prove to be a poor choice when I go to texture this? I'm worried that by compressing those vertices, I'll also be restricting myself to using only a third of the UV map for that area, and therefore the texture applied to that section will be tiny,scrunched and lower quality after baking.

I looked over your texturing tutorial, and at first glance, it seems like the projection map would allow me to reserve a large portion for that particular part of the sculpt texture.. but if I understand correctly, the projection map only translates my texture to fit over the sculpty... and will ultimately result in using only those faces I did not compress...

Sorry if that's a bit hard to follow.. I wanted to ask now, since it would be a nightmare to separate those vertices after flattening them down the road.

[Edit] Okay... so the forums are messed up and I can't post the picture... I'll try posting it again later. Rrrgh, sorry :/

Hi, Fenix;

I am afraid, you have jumped into a trap. (apologize, we should have adressed this topic
in the LOD tutorial, but at the time, we created the tutorial, we where not aware of this particular problem):

Somewhere in another thread there was a discussion about pro and con concerning the method, we have shown to make sculpties LOD resistant. One of the major cons is: as you already expressed in your question, texturing will become a problem, because the faces, which have been smashed together on the sculptie will not be taken into account for the texture, or if you don't take care, they even can lead to texture jumping (just another LOD effect at the end. after we realized this problem, we even could see texture jumping in the tutorial, if watching carefully) and they may lead to bad quality due to the mapping calculations...

So the very best bet is always to avoid using zero faces (which have been shrinked down to 0 to force LOD preserving)

That's all i can say here.

Yeah... it is a problem. I have run into it. I tried the LOD control with a Spanish hat I built. I wanted to texture it and see how it all worked together. I have the hat at my shop. You can see it change with LOD and see the texture changing. It works ok to about 5 meters.

http://slurl.com/secondlife/Minkartah/108/163/30 - hats are on the support post to the right side.

Or you can IM me in game and I'll send you (Fenix and Gaia) copies of the hat. I have a leather one and a silver one (colorable - tint). The leather one has a name inside on the top and another on the brow band. You can see the 'black' triangle problem destroy the name... On the Silver one I had to move the name outside that row of faces.

On the next one I'll handle the top and bottom vertices differently and may be avoid that 'black' triangle problem.

You can also see that creating circular or round objects has other LOD problems. While I can hold the shape, I lost the curves at low LOD and the texture changed, which I had not expected. I suspect building intricate shapes from a single sculpty is very limited. But I'll keep experimenting.

I think, the black triangle problem is fully solved by now. Domino's sphere sculpt-map already is opened at the poles, so that during texture rendering the black triangles should not be seen anymore. And still, if the are seen, just select the poles, smooth once or twice and bake again. So far, that helped in ALL cases, where we saw black triangles. The most recent tutorial about texturing sculpties adresses this particular topic (within the last minute of the tutorial)

Of course i am interested to see your hat and examine it in more detail. You also may consider to send me an LM so when i get back to SL the next time i immediately receive a reminder

Hi;

the blender beginners tutorial is published.

http://blog.machinimatrix.org/2008/08/24/the-blender-primer/

We have created an 18 minutes video about how to use blender - a basic tutorial. The tutorial addresses newbies in 3D content creation and of course everybody, who is scared about the blender GUI but not too scared to watch the tutorial

We are not showing any nifty techniques, or complicated stuff by any means. It covers the most basic stuff, like how to select vertices, move,scale, border select ...

After the video is finished, we wondered, how we could create such a long video just for the basics... But we felt, we should show the basics in full detail, so we not only cover transform manipulators, but also keyboard shortcuts.

If you have the feeling, it is still too long, or needs other enhancements, or things are missing (definitely there are things missing ... ), please tell us...

Great job!

This is one of the best, if not the best, beginner tutorial I've seen.

Decideing what to include and omit was probably not easy.

I have a suggestion for our consideration. I think this is important and difficult to find in most tutorials. That is the border-deselect. B-RMB

The lasso select (Ctrl-LMB) and deselect (Ctrl-Shift-LMB) are also important. However they probably do not beleong in this video. Nor does the BB-LMB circular select.

Of course those are my pet opinions. I understand you have various constraints and respect your decisions.

Again, great job.

Nalates, thanks for your valuable input!!!

We thought adding some infos about border-deselect would be fine. But at one point in time, we also thought, the tutorial is already way too long... But after one day of relaxing and viewing the tutorial again, i think, we can add border deselect easily near the end. I now know, where it would fit exactly.... So we will probably make this change, maybe tomorrow.
Concerning Lasso-select/deselect... now THAT was new to me ... so we never stop learning, dont we ? ...

I think I fell into another trap, but I'm not entirely sure...

Once again, using your snap to sl grid setup, I tried scaling my sculpty up to give myself more room to avoid using zero faces. In the tutorial, you mention that the only possible values are mapped to a range from -128 up to +127.

Are those referring to the grid intersections?

I'm confused because after scaling the default sculpty mesh cylinder up to 2.5, the vertex positions do not appear to be at those coordinates in edit mode's Transform Properties window. And of course, if it's a percentage, 2.55 would be way over...

How large can I scale sculpty meshes up on the grid before it exceeds the baking limits?

Also, great job on the Blender intro. May I recommend talking about the hide function if you do an intermediate blender video. That really helps when you need to see through the clutter.

Let me try to give a comprehensive answer here. It took myself a bit of time to understand what exactly happens. I was thinking too complicated at first I might be telling mostly things you already know, but maybe remembering them helps a bit to "see" what exactly happens. So here we go:

1.) each pixel in the sculptmap logicaly corresponds to a 3-dimensional vector in space.
2.) each pixel in the sculptmap technicaly contains 3 bytes

so, technically the 3 coordinates x,y and z of each vector are mapped to the 3 bytes of its corresponding pixel.

3.) a byte contains 8 bits
4.) a byte can encode the numberspace from 0-255 which makes exactly 256 numbers where the corresponding decimal to binary map is in first place:

dec binary
000 00000000
001 00000001
002 00000010
003 00000011
...
254 11111110
255 11111111

Exactly this mapping is used, when we are dealing with unsigned numbers (positive numbers only).

5.) But in our case the binary numbers are mapped into signed decimal numbers instead. In computer technology the leftmost bit of a bite is (typically) used as a marker for positive/negative values: If the bit is "0", then the number is positive, if it is "1" then the number is negative. Furthermore it was decided to map the bits of a byte differently as one would expect in first place. Instead of just using the rightmost 7 bits as number from 0 to 127 (that is what you encode within 7 bits) and then evaluating the leftmost bit as sign (+/-), the mapping has been choosen as follows:

--- dec bin
(-) 128 10000000
(-) 127 10000001
(-) 126 10000010
...
(-) 002 11111110
(-) 001 11111111
(+) 000 00000000
(+) 001 00000001
(+) 002 00000010
...
(+) 125 01111101
(+) 126 01111110
(+) 127 01111111

So we do have 128 negative numbers, 127 positive numbers and the 0 in the middle. This makes 256 numbers in total.

Now lets turn to 3D and think about a cube of dimension 2.55 in each axis.
Lets furthermore say we have setup a grid with gridspace 0.1
So a quick calculation tells us, that we can equaly distribute 256 vertices along each object-axis with a distance of 0.1 (just because we do have exactly 255 edges between 256 vertices along one axis)

So, where are the pixels located along one axis, when the object center is exactly at 0 ?
they are NOT positioned at the gridpoints. they are shifted by half of the grid-width in each dimension. Why? think back to what i explained above. How are 256 numbers mapped to the numberspace ? ah... 128 negative values and 127 positive values (plus the 0) ...

Ok. We see, that we need to shift the whole object exactly by 0.5*gridspace in each dimension in order to get the object vertices exactly positioned on the grid.

In fact, if you watch our tutorial "The Arch Example", we have mentioned this fact. We have shown, that the vertices are not places on the gridpoints in first place, but shifted.
and later in the video we have shown, that we moved the object by half the gridwidth "to the left".

Now you could ask, what happens during the transformation to the sculptmap... What happens, if vertices do not lay exactly on the grid ? In first place, nothing happens. The transforming algorythm simply does not know about a grid!!! It takes the boundingbox of the object and maps this boundingbox.

So the easiest way to make precise sculptobjects is to scale up the object, such that it has the dimensions 2.55 in each axis and then take care, that all vertices are exactly positioned at the gridpoints (or all vertices are shifted away from the gridpoints in exactly the same distance, which is mathematically equivalent, but makes positioning way more complicated in the 3D tool...). Besides putting the vertices on the gridpoints, it is mandantory, that your object boundingbox is 2.55 all the time! Otherwise you will destroy your precision immediatly giving unforseeable distortions (also know as "bouncy/rough sculptie surface and often misinterpreted as "compression problem in the SL uploader"

Remember, that a sculpty always is rezzed within a cubic boundingbox! The reason should be clear by now...

I think, you can scale up your object as much as you want. I started using a gridspace of 1 in my early days, but that turned out to be unpractical. So i switched to use the dimension of 2.55 in each axis and use the 0.1 grid, when it comes to precise sculpties.

As i mentioned, there is so much more to say. Since we only have shown a simple cube yet, there was no reason to talk about what happens, when many many vertices are shown on the screen. That sort of tools will be explained in a follow up tutorial. But thanks anyways for you mentioning it. We will put it on the list

Wow, ‘intermediate tutorial’… I do hope you consider it, especially for snap and other selects.

Playing with advanced snap I get confused. There are just things happening I don’t understand and have not taken the time to sort out. But finding a good explanation of snap, basic or advanced, has not gone well. I can sort of use it and get by. Weak tutorial: http://wiki.blender.org/index.php/Tutorials/Snap



I understand, I think, what you are saying in that post. However I think you mean 127 positive values and 127 negative values, not 255 each, as in the above quote. (not counting 0)

Also, the half-space offset, does not make sense to me. I may be missing something. But since we have a 0,0,0 location in the sculpty and that location would most easily be represented by a 0,0,0 location in Blender, why would we do a half-space grid offset? Or would we? Do you mean it just comes out that way if one simply scales up?

But even the scale up should not create a half-space offset. Remember. This is a lines verses spaces thing. There are 255 spaces or units but 256 vertices (units of length + 1). The maximum value is 255 because of 0. With 127 vertices on each side, 0 will always have a vertices. Think of it this way. An object 2 units wide made from 3 vertices. Assuming we have a vertices for each possible position of the 256 locations, we are in the 2 unit 3 vertices scenario. So, we always have a vertices at the zero point.

Domino’s scripts place a sphere at 0,0,0… and it seems… if all the vertices are at whole number locations (in this case of a grid taken from 1.0 to 0.1 real numbers on even tenths would convert to whole numbers), it would not matter where on the grid one locates them, as long as vertices are on the grid intersections.

Also since the conversion to sculpty works off a bounding box (b-box), would the center of the b-box have more effect on precision or may be the outer edges?

Also, it seems any object less than 255 units across (lines or spaces idea) or at least any none even increment of 2 (minus 1), would lose precision as the sculpty scaling converts?

But, how much difference does that make? I made some quick pillows and ignored scale and units and just got a proportional look. The original import looks all squished as it is jammed into a .5,.5,.5 cube. Scaling them in SL results in nice smooth pillows. So, I have not seen the distortion from math round off hitting the sculpty. I would assume that kicks in with lager objects measuring over 255 units or objects close to 255 units sized to none even increments.

Outch! Indeed an error slipped through... i corrected this in the post above. BTW: The correct wording is "128 negative values plus 127 positive values plus 0"

it is the 0, which makes the headache We have 256 vertices, but only 255 edges.. If the object center is placed at 0, we have 127 edges in the negative direction and 127 edges in the positive direction. PLUS one edge around zero... this edge is half in the negative and half in the positive , hence all vertices are shifted away from the grid by half a grid point, if the edge-length is equal to the grid width (which it is, when the object bounding box is 2.55 and the gridwidth is 0.1)... You can see it immediatly:

- create a grid width 0.1
- then create a unit cube of (size 1) at <0,0,0>and scale it up to 2.55 in all dimensions.
- Then zoom to any vertex...

You will see, that all vertices lie exactly in the middle between the gridpoints. shifting the object by half gridwidth in all dimensions makes all vertices line up with the grid points...

But i also agree. All i said only matters, when it is important to place the vertices at exactly defined points in space. It does not matter for organix shapes and for sculpties, that do not need to match exactly.

OK, I see what you are saying. I think.

Because I think of a zero point as nuetral, I don't really account for the condition of zero being a binary positive number. A kink in binnary math. But for your explaination I would have missed it.

So, any symetreical shape is likely to have the half-space-grid problem. Having 128 numbers/points (0-127) and 127 lines on one side with 127 numbers/points and 127 (yeah 127 - had to think on that*) lines resulting in 255 lines. So, no true middle-point on a vertice.

Darn. - Thanks Gaia.

* that 0 to -1 line only has 1 end point of which we already counted one end of the positive side

Guess who....

I get a little further each time. Now I'm trying to texture my sculpt. I think I actually found an answer to my question, but I'd like to impart what I found.

mesh -> UV unwrap -> project from view...

This works so long as I have every face mostly on the same plane I project from. But what if I have some folds in the mesh that are perpendicular to the view I projected from? For example, if you take the default cylinder and close off the ends (without using zero faces).

Using the method above from your tutorial, any faces perpendicular to the view you initially projected from will show up as a flat line in the image editor window.

I've poked around at the other unwrap options and found that cube projection was much better suited for my mesh in this case.

I suppose this is a no-brainer for experienced sculptors, but I think it would be worth mentioning in your tutorial that the other unwrap options might be better in different circumstances.

Fenix; you are absolutely right. Using project from view is not always the best choice. In our tutorial we only adressed the situation, where a projection from view makes sense. All other projections would need to prepare the starting images to some extend (you will need an image with an already applied cylinder-projection when you want to use it to texturize a cylindricaly unwrapped object...)

But (hopefully) the good news is: I just am about to remake the texturizing tutorial after i found a way to apply multiple images to selected faces, thus avoiding the intermediate step that involves an external image editor. after i understood, how that works, the rest became rather easy and simple. i wonder, why it took us so long to find the easy way to go

This is a brief description of the new process:

You can use "partial unwrapping", where you just select those faces, you want to texturise, then either use the current unwrapp, or simply unwrap the selected faces again... Just do this, while your "projection-map" (as we called it in the tutorial) is active. Take care to leave the sculptie map intact. Otherwise your final texture baking goes messy...
After you have worked through all faces, which you want/need to texturise, you can use the "consolidate into one image" function (available in the UV-editor) This will create one single image for you, which is perfectly aligned to the current UV-texture. And this image can be used as the starting texture, from which the final SL-sculptie texture is derived (automatically) ...

Well this is in very short words the way, we are going to show in the updated texturizing tutorial. i guess, things become clearer, when we show how we do it.

I still search for a good way to blend several pictures into each other and thus make seamless textures. i now know, how to do that in principle, but i search for an easier way to go. just give me a few days...

The new tutorial (texturizing with multiple images) is nice. You continue to improve our skills. Congratulations on a well done tutorial. The image is sharper or at least seems so. The highlighting you do of the things you are clicking on is nice.

I learned several things about Blender that I did not previously know. Which is very nice.

I really looked hard to find anything I thought would make this better. So, far I have no good contribution.

Thanks!

Thanks a lot for the very useful tuts you did. The video + the text is a great PLUS.
Sure I am looking forward for more tutorials of the kind, such as Blender node editor ( Part 2

I tried to learn Blender some time ago using other tutorials and got no where fast. I decided to try some other 3D programs for SL sculpties and had nothing but trouble finding info and non corrupt scripts.

I was about to give up on sculpties, then I decided to give Blender another try and found your Mechinamatrix tutorials. This time I think I will stick with it and I would just like to say they make learning blender much easier. Thank you

However when I got to the part about making the helmet on this page: http://blog.machinimatrix.org/2008/05/12/blender-sculptie-creation/#more-11
all the images below the video we are to use do not show. Instead there is small boxes with the red "X" indicating missing images. Show picture does not work. PLEASE get the images back up so we can continue on the tutorials. We will need them for the texture mapping tutorial.

How do I move the camera around and zoom in and out in edit mode while editing(moving, scaleing, rotating)? Every time I try to use the + - it changes my edit back to the starting point. Middle mouse/shift doens't work then either.


Is it at all possible go get a 1024 vertice cube script to use with Blender. This would make an excellent starting point for making cube and rectangulat type shapes plus much more you can't do with normal prims.

Ups. i thought the pics where all uploaded to the blog host. but apparently they are not. So we just restarted our local image server and now the pics are available again. BTW: What texture mapping tutorial are you talking about ?

It may well depend on your OS - Windows/Max/Linux, and also your mouse driver. I'm on Windows XP and the middle scoll wheel on the mouse works for me.



You need to download the scripts from Domino Marama's site for doing things with sculpties: http://dominodesigns.info/second_life/blender_scripts.html
He's done all the hard graft and provides support for the appropriate 1024 face meshes. He's my all time hero!

First of all great turorials.
I have blender 2.48 and the newest Domino scripts.
I'm at the video Surface Textures for Sculpties with Blender and when i go to BAKE RENDER MEHES»TEXTURE ONLY.
And then i get this message ERROR IMAGE NOT FOUND TO BAKE TO.

If someone can help me with this i would apreciate.

Well, i can not yet tell exactly, how and when in the whole process the problem raises, but this is what i found to get rid of the problem:

Short version (when you are sure, you did everything correct):
========================================

before baking:

go to edit mode and select all vertices in the 3D view.
then select all vertices in the UV editor.
now

"render" -> bake render meshes" -> "texture only"

Done!

If for some reason, you still end up with "no image found to bake to", try
to check your .blend file as follows:


Long version(when you can't get through the short version):
======================================

Before you do the final bake:

Be sure that:

1.) you got a material and the material got a texture with the correct projection view (the one you just have created, when you followed our tutorial) You will see your texture on the
lower left part of the "shading pannel" in the material buttons.

2.) You have correctly set the map input value (take care for typos in the texture name. Something is wrong, if you see the name on a red background)

3.) you use the sculptie uv-texture for "rendering texture" and as "active texture" (as explained in our video tutorial)

If everything is correct as explained above:


1.) go to edit mode
2.) select ALL vertices in the 3d-view.
3.) select all vertices in the UV-editor.
4.) create a new image

The new image should contain a yellow grid (and all vertices selected)
And you should also see all vertices selected in the 3D view.

Still in edit mode, do:

"render" -> bake render meshes" -> "texture only"

And your Sculptie surface texture should be magically created.
You should store yopur .blend file now. From now on, (as far as i have seen) you can render in any mode as long as you got vertices selected.

======

If you still dont succeed, send your .blend file and i will take a look at it

good luck!!!