mestela

It depennnddsssssss. Some bullet points: It used to be a very binary choice; be self taught or go to a school, nowadays (esp with covid) there's several options in between generally some kind of collaboration component is useful, i often compare vfx to a team sport, the more time you can be 'on the field' the better I taught a 1 year masters degree for a few years, we focused heavily on big teams, lots of collaboration, making mistakes. Our success rate with grads was great, but yes, masters degrees are expensive Other options these days range from free, like joining a discord, to cheap, like paying for mentoring or a patreon, to mildly pricey, like rebelway/cgma, through to full costly tertiary education courses Totally self taught is possible, but hard. It's good to have at least some component of interaction with others, regular feedback, accountability. Know yourself. If you think you can get that level of feedback from a free discord (or forum like odforce), great. If you need more direct feedback via patreon, great. If you need the accountability of paying lots of money and a strict classroom structure, great. Work out whats best for you. DO NOT make the mistake of thinking that paying lots of money for a course will be the fix for everything. Every year we'd have 1 or 2 students who wouldn't put in the work. We'd point out how much money they were paying, what the stakes are, maybe they'd be better off dropping out, they wouldn't listen. They'd finish the course passing by the slimmest of margins, and then be confused why they couldn't find work, while all their peers got jobs. I heard a line once, 'an education is taken, not given', and thats especially true at the teritary level; you can't buy your way into a career, you have to work at it. The perks you mentioned of certain schools (green screens, cameras, farms etc), they're nice, but not essential. You can find cheap or free greenscreen material online. Pretty much all the software worth learning have cheap or free education versions. An iphone shoots better footage than a top of the line prosumer camera from 10 years ago. Gridmarkets and the like offer cheap(ish) farm rendering should you need it. That's all gonna be way cheaper than paying for a expensive course. There's lots of videos around now from recruiters, heads of departments in various big studios, talking about what they want to see on a reel. If often feels like students and beginners misconstrue what skills they need to show, and either set the bar way too high for themselves, or get trapped/distracted showing the wrong things. Find those videos, study them, stay on target. My gut feeling is anyone who is mildly competent should be able to get a okish gaming pc, houdini indie, lock themselves in their bedroom for a year with a helpful discord, and be able to get a job. The big variable there is the discord, finding the right mix of people who know what they're talking about, can offer appropriate advice, honest but not cruel feedback. Ah, I see animatrix has replied already, i just checked it, doesn't conflict with stuff I've written so far. I'll stop writing and hit submit. -matt

This seemed to work; I had a tops @foo attribute array with 1,2,3,4, and in a for loop i could accees it and get the index with evals("@foo."+detail(-1,'iteration',0)) Where the metadata node was wired in as a spare input.

Think of it like a big particle; it just stores position and radius, the surface itself is defined by those attributes rather than by surface triangles. Houdini has several primitives types like this, as a broad rule they're not used much, often you'll convert them to more familiar polygon based representations, but they're handy to have around. More info: https://www.tokeru.com/cgwiki/index.php?title=Points_and_Verts_and_Prims

You're close, the answer is kind of embedded in your question. You already have the pieces defined both via groups and via a @class attribute. Ideally the vellum constraint sop will be able to recognise the pieces via one of these methods, and only make constraints between the pieces, not within them. Turns out this is a built in feature; set 'define pieces' to be 'from attribute', set the piece attribute to 'class', now it will only make constraints between different pieces. The other bit that always cataches me out is that both the group type and target group type drop downs have to be in 'points' mode. grain_deform_02_me.hiplc

I don't know the details of that algorithm, but from what I understand biharmonic skinning is based on a similar idea; bones are resampled into curves with many points, the mesh has a tet mesh generated inside/through it, and weighting data is transfered via the tet mesh in a similar way to heat transfer, which is in turn transfered to the original skin.

Agreed, annoying, it should be a preference. RFE!

I don't think there's a clean way in tops nodes, but its easy enough with a python script top: # get attrib = work_item.stringAttribValue('myattrib') # split attrib = attrib.split(' ') # rejoin with underscores attrib = '_'.join(attrib) # set work_item.setStringAttrib('myattrib', attrib)

There's several volunteer mods that have to approve posts from new users, but we're all fulltime houdini artists too, and sometimes don't get here as often as we'd like.

Not sure, you've not given much info to go on! Whats the starting size? Ending size? Are you storing extra attributes? Is it regular usd, or usda/usd/usdz as input? How are you exporting? Have you tested on a simple box to get a sense of how it goes on a simple case, then a highish tesselated sphere? Is there animation? If so, are you creating uv's per frame unintentionally, blowing out the file size? I've tried to cover the basics here, it might give you more context on your issues: https://www.tokeru.com/cgwiki/index.php?title=UsdGuide

AFAIK, there's no getting around this, you need to generate uv's in sops. A sop modify should let you cleanly roundtrip, so append one, dive inside, unpack, add uv's, when you jump back up to lops the stage should be identical, with the addition of uv (well, st) coords.

Proximity skinning doesn't use tets, the setup is a little simpler: https://www.tokeru.com/cgwiki/index.php?title=HoudiniKinefx#Proximity_skinning

Here's one way, you don't need particles to get the effect you're after (assuming I understood your question). volume_trail_grow.hip

Read the first bit on that solver sop page, it explains it.

Use mops, its great. Here's a way to do it, but I bet its more complicated than the mops method. DynamicBlending_me.hip

Great stuff!