2D Metaballs?

[JaMike]

Let me rephrase my objection - water, fire and smoke don't occur that often in most animations, whereas character animation happens all the time. I'd prefer tools that help me all the time, rather than on rare occasions.

If I need an effect for 30 seconds in a ten-minute film, I'm happy to go outside the program to get it (to 3D or an FBF program). If I need character animation for 9'30 out of 10 minutes, I want to do that in my primary software.

But Selgin is right to suggest it as something to think about. However, if Smith Micro were to spend all year developing it for version 9, I would be disappointed.

[GCharb]

Let me rephrase my objection - water, fire and smoke don't occur that often in most animations, whereas character animation happens all the time. I'd prefer tools that help me all the time, rather than on rare occasions.

If I need an effect for 30 seconds in a ten-minute film, I'm happy to go outside the program to get it (to 3D or an FBF program). If I need character animation for 9'30 out of 10 minutes, I want to do that in my primary software.

But Selgin is right to suggest it as something to think about. However, if Smith Micro were to spend all year developing it for version 9, I would be disappointed.

That's basically what I meant. If a user want to spend time on such a script, I will welcome it, but if Mike C time is spent on that sort of new toys, that time should have been spent on more widely useful tools.

The reason Mike time is precious is because he works at the application level, the root of the software, and this is where the most important tools should reside, not scripts, even great ones.

An exception to that, in my opinion, is the morphdials script.

Just my two cents here.

[Rudiger]

I'm with you, Selgin. The way I see it, 2D vector animation is a burgeoning field, pretty much like 3D was in the 70s. We are only trying to get a similar look to frame by frame animation, but using totally different techniques that make it much more efficient. There should be no rules as to what techniques should or shouldn't be used to achieve the frame by frame look.

Anyway, I had a bit of a look into metacircle/metaball algorithms, but couldn't find any vector friendly ones. Also, in the example you showed, you had arbitrary shapes, whereas metaballs would require the primitive shapes to be circles only.

It occurred to me that it might be better to fake metaballs intead. You could distort points of the primitive shapes based on their proximity with other shapes. You could then use bezier clipping to merge all of the individual shapes into a single shape.

Your post also reminded me to get back to my tool that automatically creates keys for new points based on keys for points either side of it, to help with the "stuck-pin" problem. I had put it on hold when AS7 came out as it was too difficult to model the new way that curves were represented. However, AS8 now gives you direct access to the bezier handles that AS generates, so it should now be possible to support AS8 as well as AS6.

[JaMike]

I can't get Gilles's lavalamp comment out of my head - I think that doing anything more advanced than that using metaballs just isn't going to be worth the effort.

[ponysmasher]

Your post also reminded me to get back to my tool that automatically creates keys for new points based on keys for points either side of it, to help with the "stuck-pin" problem.

I've already done that

Well not exactly, I've modified the split curve script to work on the whole animation: http://animestudioscripts.com/temp/ds_splitcurve.zip

It works if you only select one curve (two points, that is). If you select more AS will crash. Haven't figured out a solution for that yet.

My script, like the original Split curves, uses PointOnSegment to figure out where to put the points, but doing it by calculating the bezier curves would be way more hardcore. Maybe it would be more accurate as well since my script isn't always 100%.

[Rudiger]

Your post also reminded me to get back to my tool that automatically creates keys for new points based on keys for points either side of it, to help with the "stuck-pin" problem.

I've already done that

Well not exactly, I've modified the split curve script to work on the whole animation: http://animestudioscripts.com/temp/ds_splitcurve.zip

It works if you only select one curve (two points, that is). If you select more AS will crash. Haven't figured out a solution for that yet.

My script, like the original Split curves, uses PointOnSegment to figure out where to put the points, but doing it by calculating the bezier curves would be way more hardcore. Maybe it would be more accurate as well since my script isn't always 100%.

Oh, OK. That's interesting. What I like about mine is it preserves the position of the point along the curve and it can also fix points that have already been created.

[ponysmasher]

What I like about mine is it preserves the position of the point along the curve...

So does this. But sometimes between the keyframes you can get weird results.

[chucky]

water, fire and smoke don't occur that often in most animations

I have to disagree, with this statement just watch any Saturday morning cartoon think back to even early anime like gatchaman, certainly limitations always affect the amount of usage of any technique, for example in 3d the development of various lighting methods and also the use of fur and hair were limited until the tech caught up.
I don't think objections are particular helpful in this thread, any advancements are always welcome.
Obviously this meta-ball technique does break the general rule of AS with the point count, so I can see that this potentially great idea could get messy in its application.
I say go for it, there is no good reason not to push the boundaries in any creative endeavour.