Thanks Goldleaf! Def will put those things in tickets moving forward - was out at SIG this week and wanted to make sure I had a solid repro.

Hello, is there any update on the above bug? Or perhaps a timeline for a fix? Just planning ahead for a show in house that could benefit from this. Also, is there any chance that the Hair Procedural could be back-ported to H19 if requested by a studio? Or is there too much 19.5 tech that would make that a non-starter (USD version for example)?

What timing; sorry for the long silence, there have been some tricky issues to navigate, but tomorrow's cut should have the following updates:

• The Capture and Deform parameters should be hooked up correctly.
  
• We've added a 'Skin' Deform Mode
  
• Generator mode should preserve N when copied from the guides onto the generated curves
  

Hamilton Meathouse
Also, is there any chance that the Hair Procedural could be back-ported to H19 if requested by a studio? Or is there too much 19.5 tech that would make that a non-starter (USD version for example)?

I doubt it, there would be an awful lot of work; the procedural relies on more than just USD/Solaris technology.

robp_sidefx

jsmack
There's a few 'dealbreakers' that are currently missing:

Thanks for noting these!

Greetings, I was wondering, is there any 'fiddly' way to make the houdini procedural hair read attributes from the skin mesh(with no knowledge of python)? Also is there more info on how they managed to create that groom using the procedural hair - 'https://www.sidefx.com/products/whats-new-in-195/character-fx/ [www.sidefx.com]' .Maybe all of the attributes for color variation are stored in the guides, and as a workaround for the 'density' they blasted parts of the mesh that would be 0 density? Only such robust solutions come to mind without the ability to read attrs from the skin. All in all, is it a good idea to try using it in complex production grooms in its current state? Very exciting stuff in 19.5 by the way!

Tomorrow's build (19.5.431) has some fixes/updates to the hair procedural, including the ability to drive generated fur density/length with attributes on the skin.

Is the hairprocedural broken in the current daily builds? I get a couple of warnings and errors with the example scene in 19.5.441. The hair isn't visible in Karma even after adding a rest attribute.
Are there any additional steps needed to make the example scene work?

im also trying to use the hair procedural, and seem to be feeding it everything it wants, but it is not showing anything in render!

Yep looks like something is broken, sorry about that; we'll look into it.

Please don't hesitate to log bugs w/ hip files when you find problems. As threads get longer, sometimes responses get missed; logging a bug through support or the submission form gets more eyes on the issue and is trackable.

Thanks

elovikov thank you for the example!

Does somebody know if its possible to have this working with motionblur especially with geo samples. Maybe it is possible somehow to sample the geo before and after the frame to create a per frame "cache" to get motionblur on the points with the procedural?

In the image attached you can see that motionblur doesnt work in this case. the rubbertoy has timesamples on the points.

Regards Alexis

Hair Procedural only supports velocity based blur in Houdini 19.5. The velocities authored on the guides gets transferred onto the deformed/generated hair. We hope to support multi-sampled blur in the future.

Thanks for the info! We were playing around with attaching non simulated hair like peachfuzz to a deformed body. So we dont have any velocity to transfer and thats why i wondered if it maybe existed something like multisampling. But this is good info and we can workaround it for now and hope for it maybe in the future

Hi to all,
it's very interesting topic. I'm trying to use hair procedural but something wrong when I add an animation sublayer. Maybe I'm missing something and I need a recap :S

I would like to use hair procedural to apply static fur to my animated characters, this is what I've done...

1. I create an asset with component builder with hairprocedural to generate fur on cahracter's hat (I created guides in sop, then imported with sop import in stage).
   Everything works fine with lookdev and render, except this warning in render in HtoA 6.2.1.0
   00:00:19 2641MB WARNING | /XXX/geo/GRP_fur/hat_fur_procedural: orientations ignored for mode ribbon, leaving out orientations would save 0 MB
   Anyway, I export my chr.usd
   
   
2. I load my chr.usd with asset reference or sublayer in a new scene, I add a light and Preview Houdini Procedural, everything works fine, with the same warning above.
   
   
3. I import the chr_animated.usd and the curves in viewport don't match the position, they remain in original pose of the chr. However if I render the hat of chr has fur even if with another look, it looks differently combed. In console I have a new warning:
   00:00:05 2757MB WARNING | /XXX/geo/GRP_fur/hat_guides: missing points parameter
   I want to specify that the chr_animated.usd contains skinprim only, the animation file has not the guides
   You must have always guide also in animation? what then would be the benefit? only have more hair in render?
   
   

I certainly have an update problem in render, if I switch from static to animated or vice versa, to see the fur updated I have to save and reopen the file.

Is the procedure I have chosen right?
It is a great advantage use hair procedural rather than the guides in sop import, with usd in solaris?
There's a better way?
Thankyou for support

any ideas?

Hi there,
I am attaching dirt/snow (in the form of curled up curves) to fur using the hair_procedural. It has to be done with the hair procedural because the caches are prohibitive.
Unfortunately this does not work, seemingly only because guidedeform is forcing the root points to project to the skin (not sure if that is truly necessary in any case) as you can see.
Now I thought I would simply make a new version of guidedeform with the move_to_skin_pos wrangle bypassed as shown, and then simply change the reference in the procedural code to the new one, and thusly avoid having to endure the headache of making a new procedural... but I cannot find where that is. Any ideas?
Alternatively adding "Geometry to Deform" so that I could set "Arbitrary Geometry" as well as the "Rigidly Transform Pieces" toggle parameters would resolve the problem.

Cheers
S

I'm beginning to grok what's what & where with the hair procedural thanks to elovikov's helpful explanation, but in order to rewrite the currently embedded "hair_graph.bgeo.sc" with the fix, I need the hip file from hence it came (sadly the sopnet is not included in the hair proc hda).
Would it be possible to share it here please? This would also probably make it a lot easier to add more parameters as we see fit?

Cheers
S

Also, I'm thinking if one really needs to bother with parameter promotion at all?
Why not create a generic hda that has only parameter for file path to the graph.bgeos and save button? Inside would be an editable subnet. So instead of changing parms at lop level we change them in the subnet and publish changes to the bgeo with the save button?

I tried and failed miserably to reverse engineer my own graph bgeo for guide deform...
BUT, just as I was beginning to lose the will to live, I had the idea to edit/hack graph.bgeo that comes with the hair proc hda. In my case set the "flag_bypass" attribute on the corresponding point of "move_to_skin_pos2". Which of course worked fine. Phew...
Anyway, just noting here that that lifeline exits for these unusual fixes or overriding other parms without the pain involved to make it all work properly.

elovikov, amazing breakdown of how procedurals work, thank you!