jit.bfg filter basis functions - broken?
OK, I give up - I couldn't get any of the jit.bfg filter family to do anything, then as soon as I posted something, I tried some more (I think I set "weight" to 1) and all of a sudden filter.disk and filter.gaussian were working.
And then as soon as I edited the post to say "Ignore post - figured it out", and went back to my patch, it stopped working again.
What is the trick to getting the filter family (esp filter.gaussian & filter.disk) to work?
It looks like the documentation and attrui for the basis attribute are really confusing on this front. The filter.* and transfer.* are not meant to be set as the basis directly but as attributes of other basis functions. If you look at the scnoise subpatcher in the helpfile (noise.sparse.convolution), set the basis as noise.sparse.convolution and then mess with setting the basis' filter attribute to one of the filter.* functions. Be sure to tweak the width parameter, which sets the filter width. If it's not set above 1 you won't see much of anything. I'll try to make this more clear for the next release.
Thanks Wesley. Makes sense, but you're right, it's really confusing and badly documented - I guess then the filter. and transfer. functions shouldn't even be in the top-level umenu at all?
Anyway, I will look forward to having better docs someday.
Hello Wes, Hello Pseudostereo,
It seems that the documentation hasn't changed since the last comment on this thread late in 2011.
I'm trying everything I can imagine to produce the simple gaussian image that one finds in the jit.bfg reference, like this pic:
I have tried Wes' suggestion (the 'scnoise' subpatch of the jit.bfg helpfile) and am able to produce some resemblence of what i need, but it's filtered noise that i'm getting, not a pure gaussian.
Am I trying to use this object for something other than what its intended for?
-ali