Moho Forum

So, I've been getting into some complex stuff, and although I'm using a deadly fast machine, I feel like I'm starting to hit a wall with what MohAS can and can't do. The more complex my rigs/scenes become, the slower the program gets - just manipulating the bones and camera becomes a chunky chore. I'm trying to figure out why this is, and how to fix it. Is it the underlying code? Are my methods insane?

Example 1 - I've been importing 3D objects from MAX. In MAX, I can freely rotate the object and/or change my view of it, at lightning speed. It's totally fast and easy to work with.

But - when I import the exact same object into MohAS, it's another story. Rotating the object or changing the view becomes a very chunky, clunky operation. What's the diff?

Example 2 - I've been building some extremely complex puppets, and have found that at a certain level of complexity (ie, lots of bones, many of them with dynamics or control-bone relationships) Moho really starts bogging down, to the point where it's quite slow to pose them, and playback in workview is pretty much impossible... It seems that the more bones you add, the crunchier the program becomes.

So, I'm wondering if these things are related, and if there's some fundamental issue that makes it hard to work with the program after a certain point. The end results are awesome, and I hope to be able to show y'all some samples soon, but I fear it may take a million years to get there with the app running so slow and chunky.

FYI, I'm using a dual-core 3Ghz machine with 2gigs of RAM...

Lost Marble - any ideas why this is happening? Is there indeed a wall or threshhold which should not be passed, and if so what are its parameters?

Perhaps Mike can answer this better, but my guess is that because AS is multiplatform, it doesn't use platform-specific hardware acceleration. I would prefer if there were a platform-specific hardware support, although this might be difficult to support and debug for essentially three platforms (Windows, Mac OS X and KDE Linux).

Rasheed wrote:Perhaps Mike can answer this better, but my guess is that because AS is multiplatform, it doesn't use platform-specific hardware acceleration. I would prefer if there were a platform-specific hardware support, although this might be difficult to support and debug for essentially three platforms (Windows, Mac OS X and KDE Linux).

There are cross-platform libraries that can use hardware acceleration -- OpenGL is one (2D and 3d), as well as higher level layers like SDL and (I think) Allegro. Apps like Maya and Softimage|XSI are all cross-platform and high performance.

Yes, you're right and software vendors don't have to license OpenGL to their customers. Hash Animation : Master (Mac OS X and Windows only) uses OpenGL as well, I saw on the openGL website, although that product isn't as robust as you would like (do a Google search on "animation master crash"), but that's beside the point.

I would welcome any stable hardware acceleration to Anime Studio, after the current bugs in AS have been eliminated.

What is you file size?

I have found the same problem when working with large project files.
On one 47 meg vector file, the jump to rendering is fine, no delay. But if I call the Export Animation, after I select the destination bin, I can go have coffee and a three course meal before the "render as uncompressed" window comes up. The program is doing some kind of auto-conform or data checking between the routines - I'd be happier if it did this after I have selected the destination path and the render foramt required - then I don't care how long the render prep is, I can do something practical (like sleep). For Mike's record, the actual delay time can be over 1 minute, though normally between 30-40 seconds. Once it never came back and I rebooted the system.

As we get more crazy with project complexity, these little procedural things which are inpeceptual on a 70K file, become monsters when you start handling 70Meg master scenes.

BTW: I now have a habit of rendering the first and last frame of every scene and check the output by flicking between the two images in a viewer like Irfanview ... it tells me quickly if there are any colour pops, things moving which should be, stationary objects stationary, etc. When you have huge data files with many timelines, groups and levels, it is very easy to put in a wrong key accidentally or not start it at frame 1, etc.

By rocking and rolling between the two frames, I have twice caught stupid errors, saving reshoots on scenes which take 3-4 hours to render. Tonight I have 3 big renders on batch render - I hope this pre-checking will pay off.

The delightful news is the 4 hour render is only for the Standard TV output - I have to render the HDTV files later.

Rhoel.