edit: My apologies to those of you who may have been offended.

Using what I remembered from modeling in Blender, Wings, and other programs, I have figured out a way to import character (organic subdivided) models from these programs into Animation Master's patch system with much better accuracy.

If you import an organic model directly into Hash, you will usually get an ugly result like this. The points, so use to being averaged by the Catmull-Clark system, are now being left where they are and told to push the edges between them further outward. This causes the organic model to become puffy and ugly.

Click to view attachment

Now, with my method, the points are relaxed before import in a special way that places them where they would be expected exist in a patch model.

Click to view attachment

Click to view attachment

Here is the video tutorial. It is a mov and requires Quicktime or VLC to play.
Click to view attachment

Wow, that is one of the best results I saw till now to get a model from a polyprogramm to hash
Thanks for the explainations too

*Fuchur*

Nice work, Christian! This will please all those people with their catalogue of polygon models!
1 day old in the A:M community and a champ already!
Many thanks!

Very interesting. I wonder how the splines flow in those models? Do they keep the same orientations as in the poly app? Also, there are some modeling methods used in poly apps that don't work in AM because AM uses five point patches. Though I'm not sure about spline modeling in other apps.

Even if these are problems (though I doubt it from the lack of creasing), your method does make things better looking.

Edit: What format did you import with? OBJ? That would bring in texture too.....

Thanks guys!

Yeah, obj is a good format for subdivision.

Here is how you fix a pole. Most of you probably already know, but here it is anyway...
Click to view attachment

update: it appears Hash's obj importer doesn't support ngons. You'll have to fix the poles in Hash.

Ngon.....that's a multiple point surface right? AM only supports up to five points. More aren't needed.

That pole fix will cause a crease. It should probably continue on and hook off at the next spline. In fact, I believe the front five pointer will cause a hole because you can't hook off inside a five point patch. I think the importer has some abilities to convert these.....or maybe I'm thinking of the exporter.

I did some more work on it. Starting the fix from Wings/Max/whatever and finishing it in Hash seems to work.

Click to view attachment

Get rid of the splines that cut through the intended ngons, then recreate a 5 point patch in their place.

I notice three lines meeting in the center. That's usually another no no. It might look ok (no crease?) but it could cause animating problems.

I like the avatar. Did you get him into AM?

How exactly do you make a character model with all quads and all 4 edged points? Why, the model would have to be an irregular cylinder... are edge loops not important in Hash? I think it will be fine when animated if it's rigged right.

I'm actually more inclined at the moment to learn how to model properly with patches than working in some external program and bringing it to Hash. I just needed to figure this method out so I could continue a project in Hash (heh, it doesn't involve Lucy).

Thanks, but Toki isn't in A:M yet.

It is easier to go from a 5-pole to a 5-point patch than the reverse. You can do that by either removing an edge loop or by continuing one edge to a hook.

There is no concepts like edge loops with patches. A spline is a loop in itself because of its inerrent continuity. A spline "loopness" is not determined by the topology of the adjoining edges.

Almost all the solution you found to convert a 5-pole to patches are no-nos. They will all produce creases. Learn about 5-points patches and hooks. In the patch land, those are the topology solving tools.

Wow, simply removing the spline fixes the whole problem. No three edged points. Thank you!

edit:

Ok, now I see what you mean about the spline being a loop in itself. Heh, this thread has turned into a little modeling tutorial.

Click to view attachment

Hey, I haven't tried it yet , but this looks like a useful technique that will expand the potential user-base of A:M.

thanks!

Maybe. It's accuracy goes down the fewer polygons you have, but for most character models this should work fine.

Click to view attachment
Hash's is the blue wireframe.
The white dots are where it matches (also where the points are)
The red dots are where it doesn't match. There are no points here, if there were, it would be better defined and match better.

I am no longer convinced this is a good method of getting characters from other programs into hash. Sure it works, but the models are way too dense. In subdivision you make creases by reinforcing edges with more edges. In hash, you just peak the spline or reduce it's bias. The effort spent cleaning the imported models could be better spent remodeling the characters in hash properly.

Perhaps some judicious pre-thinning in the original app will make for better candidates for importing?

Totally.

edit 3-25-08 : or maybe there is no such thing as too dense... at least when it comes to realistic characters. Pardon my uncertainty, I am still learning.