mbPlumage - Feathers for Houdini

[Squizmek]

Hi everyone,

mbPlumage is a digital asset for feather styling and simulation in SideFX Houdini. A 3-day-trial version is available on orbolt.com and runs on Houdini Apprentice, Indie, and Commercial.
You will probably recognize some of the features from mbFeatherTools 3.0 (a similar plugin for Softimage). mbPlumage is much better though, so I recommend you check it out.
For more info, visit http://tools.haybyte.com

[Hirazi Blue]

mbPlumage is much better though

I have no reason whatsoever to doubt the truthfulness of that statement,
but for promotional reasons it might be helpful if you could elaborate on that (why is it better?)
and your views on this might be interesting to some of the user base as well
;)

[Squizmek]

... it might be helpful if you could elaborate on that (why is it better?)
and your views on this might be interesting to some of the user base as well
;)

Hi Hirazi,

thanks for your question. For example, mbPlumage can not only instance feathers, it can also create geometry and deform the feathers using the Deform Cage and with the Feather Simulation, Capture, and Point Simulation. You can also mix the simulation with the capture to reduce intersections. The Feather Cage is a great new way to style feathers, and grooming is now possible, too.
I will soon make more videos showing how to use the tool.

Cheers,
Michael

[FXDude]

How about perfornance?

I recall seeing weight painting on fully feathered (uncached) birds, playing back /interacing with wind / dynamics and things. (from a while ago)

But I guess Houdini is not dead which is of course the most important thing to consider.

(though I understand if you weren't making as many sales, but which is also related to that previous sarcasm.)

Disappointments aside, indeed looks very good.

[Squizmek]

How about perfornance?

Hi FXDude,

there is already one video about performance which I will link below. Weight painting on a fully feathered bird is something I have spent lots of effort optimizing. On my 8 core CPU (i7, 3.40GHz) when painting on a bird with 10,000 feathers, it updates at ~140ms (7fps) when viewing as lines, and at ~160ms when viewing as instance planes. Note that the parrot example in the second video has only 6,000 feathers.





6214 feathers
Simulation time per frame: ~900ms
Render time per frame: ~2,5 minutes

[FXDude]

I don't know, but at the very start of this clip (don't know what system specs you had 2 years ago),


modulating with nulls seems pretty-much "live"

Compared to doing essentially the same thing in your new clip (at 1:25)

In general, XSI (ICE) is known to be somewhat faster,
Although surely no doubt your system is faster than other Houdini effects.
(often only final renders and/or playblasts are shown)

But I guess what bothers me is when I see what seems to be prematutre migration
(before its actually worth it, or only because of little other than what can be "surface attributes")

[Squizmek]

You cannot compare performance if you have different characters, features, and particle count. A performance video is coming soon, however.

[FXDude]

So are you saying that it's actually faster in Houdini? (because that would qualify as a first)

They are perhaps different characters with different feather counts, but they are at least visually (or roughly) similar in both respects.

And if your new simulated feathery balls comes to about 10fps,
here in your vid (at 5:40 .. from -5- years ago)

we see you wiggling around another bird at at least as many fps as your capture can handle
(can also see occluded fps stats that are double digits with the first number not being 0, 1 or 2)
just before wigling another ball with at least as many feathers and just as live.

But aaaanyhoo..

please don't get me wrong, I may be comparing ICE & Houdini,
but absolutely not your new baby (with indeed great new features) with your former baby if that makes any sense.

(and not comparing XSI as a whole (as oposed to just ICE) with Houdini, which are in contrast, much more like apples and oranges or like entirely different things)

[NNois]

this one is very responsive https://vimeo.com/108300207

[Mathaeus]

In general, XSI (ICE) is known to be somewhat faster,

Just for small defense of Houdini - generally, I found Houdini and ICE "core", to be somewhere close when it comes to speed. "Core" = building the network of let's say VOPs in Houdini in similar way to ICE. While is still hard to say some exact numbers, because of different 'preferred layout' of operators.
But, there is something I'd like to call 'simulator's mind' of developers and some users too, where real-time interaction simply isn't considered as important, when it comes to final implementation. As opposite to 'animator's mind' in SI and Maya, where fast interaction is ultimate. I think this thread depict very well what I wanted to say.
SI is not free of these "beauties", too, remember the XSI biped full of SCOPs and so fragile spine (but with built in no-sense of ability to twist the human spine to 360 degree).
Long story short, H desperately needs a veeery good number of tools, created by authors who have on-field, out-of-studio experience. In other word, I think Side Effects alone should buy first 2k or 3k copies of mbPlumage, just to give a proper support.

[FXDude]

this one is very responsive https://vimeo.com/108300207

5.7 fps (non accelerated) playback (at the end), and for the other performance vid, the non-MB-Plumage-enhanced regular point sim sphere (8 min) is just scary, and any of that is apart how ICE can more or less be to Houdini, what SI can be to Maya (in terms of friendliness concerning otherwise complicated things)

As opposite to 'animator's mind' in SI and Maya, where fast interaction is ultimate. I think this thread depict very well what I wanted to say.

In Houdini's defense, (or in criticism of SI) at least there is a Houdini discussion on their non-Softimage (or otherwise) overwhelmed Houdini community forum.

Quickly! move (to Sketch-up, whatever!) before Softimage becomes too slow (or otherwise comparatively less capable) !
(when... 2022? ... 26?)

[Squizmek]

In general, XSI (ICE) is known to be somewhat faster,

Just for small defense of Houdini - generally, I found Houdini and ICE "core", to be somewhere close when it comes to speed. "Core" = building the network of let's say VOPs in Houdini in similar way to ICE. While is still hard to say some exact numbers, because of different 'preferred layout' of operators.
But, there is something I'd like to call 'simulator's mind' of developers and some users too, where real-time interaction simply isn't considered as important, when it comes to final implementation. As opposite to 'animator's mind' in SI and Maya, where fast interaction is ultimate. I think this thread depict very well what I wanted to say.
SI is not free of these "beauties", too, remember the XSI biped full of SCOPs and so fragile spine (but with built in no-sense of ability to twist the human spine to 360 degree).
Long story short, H desperately needs a veeery good number of tools, created by authors who have on-field, out-of-studio experience. In other word, I think Side Effects alone should buy first 2k or 3k copies of mbPlumage, just to give a proper support.

I totally agree Mathaeus. Useability and clean design should be the first priority, and only after that comes speed. What you mentioned about Maya reminds me of premature optimization.

The performance tests I did with mbPlumage were interesting. In most cases it's faster than mbFeatherTools. Video is coming in a few days

[Squizmek]