Mushy sculpty

Okay so I made my first sculpty in Wings3D by hand. It's all hard lines and angles. However! When I upload? I get this:

http://i86.photobucket.com/albums/k109/LephaRotting/buttonissue.jpg

All mushy and gross looking.

Is this an issue with the exporter? Anyone else had this issue?

P.S. How it should look: http://i86.photobucket.com/albums/k109/LephaRotting/shouldlook.jpg

JPG is a lossy file format.

Save it as PNG and upload it as a lossless 64x64 sculpt.

It's actually uploaded at .bmp 128x128. Lossless is checked.

Lemme try scaling it to 64.

Looking at the prim again, I think you need to double-up all the points where you want a sharp edge. Sculpties weren't intended to be sharp-edged, so you need to give them that extra cue to force them to behave that way.

Nope, still funky.

You're probably right. Arrggg.

Thanks!

I think I figured out the issue.

In all the examples I've seen of sharp sculpties (DanielFox's tutorials show some), the mesh on the model has a different pattern. I'm thinking this is the key, and that using the bae sphere model for Wings3d is what is making it 'mushy'. Gonna find templates and experiment. Will post results if anyone is curious.

wings3d's exporter is very notorius on the poly count = smoothness factor.

try a model 16x15 sphere and see that works. thats how i made my gems.

You can get sharp edges without problems as long as the edges run along the X-, Y-, or Z- axis. This is one sculpty with a 32*32 mesh:



You can get LOD invariant sharp edges as long as you take care of modelling in low LOD. In my 3D tool (blender) i typically start with 8*8 faces, then i increase LOD until my model has 32*32 faces. With that technique i can preserve sharp edges and LOD at the same time. Just take care, that no more than 9 sharp edges are possible when you want to keep LOD consistency for standard sculpties. (for oblongs other values apply of course...) This is an example of ibeams created with 12 sharp edges and the standard sculpty mesh (32*32 faces):



So it is NOT (but almost) LOD-invariant. Very useable for all practical purposes though whgere ibeams are needed

You do NOT need to use duplicated vertex lines to force your model to show sharp edges. It is definitively not true that this increases the sharpness. It only decreases the amount of available vertices

There is one situation though under which it makes sense to do that. But this has to do with texturing, not with modelling. See my article about vertex lighting at the machinimatrix blog:

http://blog.machinimatrix.org/2009/04/12/article-about-vertiex-lighting-in-sl/

for LOD you might take a look at this video:

http://blog.machinimatrix.org/2008/06/04/precision_sculpties_lod/#more-14

It shows how to do things in blender, but the principles apply to any sculpty model and thus to any 3D application. I only can tell you about blender, but i hope that this information can help you a bit,

Cheers,
gaia

I usually take the bmp from wings 3d and make it a transparent tga in photoshop for security. Also, from my experimenting, resolution of 64 has never worked. I use 32.
I then upload it using SL Image Upload http://wiki.secondlife.com/wiki/SLImageUpload

During the time where sculpties came up until today i constantly hear the rumour that sculpties of size 64*64 would not work correctly. There is a fraction of people who say, it must be 128*128 pixel images to get better results. Others say it must be 32*32 images. But the truth is, that only 64*64 images work as designed and if they are created correctly, the sculpties will be perfect. All other image sizes seemed to get scaled up or down to 64*64 before applied.

For 128*128 images there is a speciality which is more about how fast the sculpty rezzes at the client side. Many(!) people say that the rez speed is coupled with the image size and for some reason 128*128 pixel images trick out the client so that they rez faster than 64*64 sculpties. I have not tested this behaviour but i believe it is true

When oblongs (sculpties with different sizes) come into play, the image size gets much more important, because it will guide the client what to do. for instance only a 64*64 pixel image will become a 32*32 faces sculptie in the future while a 32*32 image would become a 16*16 faces sculpty! So you see, that the duplication of pixels vs. facecount is on purpose. Now what purpose is it then ?

The reason why we need 64*64 pixels to code 32*32 faces is, that for 32*32 faces we need 33*33 lines of vertices aka 33*33 pixels. More exact: 33*33 pixels need to be provided for planes. spheres, cylinders and torus need less due to stitching. On the other side graphic processors seem to love power of 2 image sizes. Hence in order to encode 33*33 lines with power of 2 image sizes, you will automatically end up with 64*64 pixels as the smallest possible solution...
Maybe there is another reason for power of 2 images:as you know SL always transforms textures to power of 2 ratios. so probably this is also true for sculpties, which indeed seem to be technically treated as textures. So maybe the constraint comes from the image importer rather than from optimization issues. The truth may be found somewhere in the middle

I hope i could explain a bit.

Cheers,
Gaia

At this point I'm just going to have to ditch Wings. The only way I could get a sharp edge was as suggested to go with 16/15, in which case it gets too blocky. There's no middle ground with this exporter (although I'm sure there's some trick of the trade with meshes that would fix this).

I don't think it's honestly an LOD issue, although that tutorial was amazing. I think I'll have to go start learning Blender now.

Okay, I am making some very basic error here.

I just went farting around with sculptypaint and got the same issue. Not only that, but I followed the tutorial to a 'T'. Even when I previewed my upload as a sculpty? Looked sharp. The moment it gets rezzed it turns to mush.

Whaaa?

Rotting, you don't need to give up. Just take a deep breath. Regardless of what tool you use, sculpties in SL are a clever hack and to get the most out of them, you have to understand their limitations.

The particular limitation you are running into is the fact that regardless of how many vertices you make your model with, only 32x32 of them are going to get used in SL. (FYI, this is not the full truth, but it is close enough to explain your issue.) If you use a 64x64 bitmap, 3/4 of the vertices are going to get ignored by SL. That's a problem for your model, because your edges become mushy when the "wrong" vertices get used.

The simple solution if you are using Wings is to start with the 32x32 template. Then, every vertex you see in Wings will actually get used in SL, and 32x32 should be enough points that you can get a satisfactory result.