1920x1080

houdini_-_sae_paris.jpg

๐Ÿงต Houdini General

Anonymous No. 874422

I wonder why these are never that popular.

Anyone got some resources on more advanced procedural modeling especially when it comes to urban and environment modeling.

Anonymous No. 874473

>>874422
>I wonder why these are never that popular.
because Houdini is for nerds

t. nerd

Anonymous No. 874491

>>874422
Blunder is already too hard for 3Chanlets.

1659x1644

Houdini Pasta.jpg

Anonymous No. 874585

>>874422
https://www.youtube.com/watch?v=afHVjiNeH7A
https://www.youtube.com/watch?v=Wen2rDTnCGU
>>874473
and for women too, there many women who are houdini artists
>>874491
no

Anonymous No. 874588

>>874585
>and for women too, there many women who are houdini artists
biological?

Anonymous No. 874609

for some reason i've built a detail array (vector 2) edges in an object (with no repeats).
it seemed like a good idea at the time i was doing it but now i'm not really sure what to do with it.

Anonymous No. 874611

>>874585
>women

Anonymous No. 874827

>>874585
Thanks but I have seen these. Same for Anastasia's tuts on the Lake House course. But is there anything tying all these together to make a basic city prototype and be able to change the streets layout at your will?

I watched Dokai's tutorials in the procedural roads but if feels too basic.

Anonymous No. 874839

>>874588
the type of "women" that ack themselves at a high rate

Anonymous No. 874854

>>874609
Detail wrangles are super slow (try making a 1 megabyte array lol) so scrap it, use half-edge functions in a vertex wrangle to work with edges.

Anonymous No. 875509

>>874422
you need at least a10k workstation to use Houdini properly
only rich kids can participate

Anonymous No. 875526

>>875509
cope

Anonymous No. 875538

>>874473

houdini is for retards who cannot code. change my mind.

Anonymous No. 875557

>>875538
>vex
>python
>opencl
>cpp

Anonymous No. 875619

>>874854
t bh i mostly do small, lolsoabstract, stuff for fun so running into serialised bottlenecks isn't too much of a problem.

that said, i just looked at my code again and i'm retarded. lel.

Anonymous No. 875629

just discovered that:

i@group_rotate = 1;
if (i@primnum == 3) i@group_rotate = 0;
if(i@group_rotate==0) i@test = 1;

this will work as expected but:

i@group_rotate = 1;
if (i@primnum == 3) i@group_rotate = 0;
i@test = inprimgroup(0,"rotate",@primnum);

this will not, unless you put the test in a different wrangle.

why.

Anonymous No. 875630

>>875509
>you need at least a10k workstation to use Houdini properly

you're very very wrong

Anonymous No. 875632

>>875538
you won't go very far with Houdini if you can't program
it's not CS501 of course, but you still need to know a couple things to do interesting work

Anonymous No. 875635

>>875629
inprimgroup(0, ...) checks on the incoming geometry. the wrangle (by default) works on a copy of it instead

Anonymous No. 875636

>>875635
interesting, and i suppose that makes sense now that i think about it.
i'm trying to update a prim attrib and then run a check on it from a point wrangle, which is why i needed the function.
i guess i can just pull stuff into a local variable for the time being. is there anyway to actually get access to the in memory geo from those sorts of functions?

Anonymous No. 876277

What's the proper/best workflow for using houdini and another program such as blender or maya? I want to create and animate a creature and use houdini's fluid simulations for water.

Do I just model, rig, and bake the creature in blender/maya, then export the fbx to houdini? Do I just re-learn how to model and rig in houdini and just do the whole project there? Or is there a better way to do it?

Anonymous No. 876284

>>876277
If you want to mess with the animation:
https://www.youtube.com/watch?v=7vg0bnc5R7Y

Or basic alembic will do.

>Do I just re-learn how to model and rig in houdini and just do the whole project there?
Don't do this.

Anonymous No. 876287

>>876284
what if, for example, i had a character and a walk/run cycle animated and I wanted to bring it into houdini, where I would make terrain and water. could i adjust the run cycle in houdini to match the terrain that the creature is running on? Or should I import the terrain from houdini to blender/maya, adjust the animation there, then bring it back to houdini again for water?

Anonymous No. 876327

>>876287
I don't do character work personally, so someone else can chime in. The tools are there:
https://youtu.be/cbhMQLUX9Pg?t=1798

But like most things in houdini, the simple stuff can be a PITA to do. I'm assuming that's the case for traditional animation as well but I could well be wrong.

The new rigging toolset in houdini is called KineFX if you want a keyword to help you start looking. It's quite new. Here's the big overview:
https://www.youtube.com/watch?v=Z04t_OxBshA

Anonymous No. 876341

>>875636
>is there anyway to actually get access to the in memory geo from those sorts of functions?
I don't think so. you can experiment with the "Compute Results in Place" option in the wrangle, but it doesn't let you update attributes, only read from some and write into different ones

Anonymous No. 876525

>>875526
no
>>875630
no I'm not

1920x1080

2022-01-22 19-24-38.webm

Anonymous No. 877445

newfag here. how do i fix the flickering? boolean starts to work as it's supposed to only on ~90frame

Anonymous No. 877484

>>876327
>But like most things in houdini, the simple stuff can be a PITA to do.
Just why is Houdini so bad at basic modeling? I looked around and to get it to Blender's level I'd need to buy a 100$ plugin. The annoying thing is that the issues are 100% UI related.

Did it not occur to any of the devs that using handles for everything is a bad idea? It literally has everything under the sun in it, so why not sensible modeling workflow?

Anonymous No. 877499

>>877445
try setting the boolean A input to "treat as surface"

Anonymous No. 877505

>>877499
thank you! works like a charm

Anonymous No. 877528

>>877484
>issues are 100% UI related
It's more fundamental than that. Houdini doesn't do neat object hierarchies like other software, the automatic tools invariably make a mess of your graph so much of the time is spent double-checking and tidying up after the scripts, sometimes (mostly because of object level transforms) tools just break and there is no reliable way to cleanly nuke the history. This state of things makes me very sad.

Anonymous No. 877533

>>877528
>Houdini doesn't do neat object hierarchies
technically, you can do that in Solaris. but I see how it can be annoying or even harmful during modeling

>the automatic tools invariably make a mess of your graph
if you aren't doing procedural modeling this shouldn't matter, just build the graph inside a subnet you can ignore

>there is no reliable way to cleanly nuke the history
you can store the geometry on a node and delete all other nodes above it

1002x773

leaf.png

Anonymous No. 877625

>>874422
I created this leaf in Blender, exported is an an .obj and imported it in Houdini. But I have this weird checkerboard pattern on the stem primitives. Why is this happening and how do I fix it?

Also, I am not sure if it is related to this, but the leaf object is a part of a bigger one, and when I merge it all together I get a warning related to the shop_material attribute in the merge node. What do I do about that?

I haven't actually assigned any materials to the objects, so I find it strange that there would be an attribute mismatch.

Anonymous No. 877626

>>877625
>Why is this happening and how do I fix it?
Blender must have added some spurious UV coordinates. delete the `uv` attribute to fix it

>I get a warning related to the shop_material attribute in the merge node. What do I do about that?
that attribute links to the material assigned to each primitive. it's what you set to tell the render what to use. the merge complains because its inputs have mismatched attributes. you can delete it safely. Houdini won't add it out of nowhere, so it must have been put there by Blender if you just loaded the file

Anonymous No. 877632

>>877626
Thank you. That fixed it.

Anonymous No. 877738

What are some good beginner tutorials for houdini? I'm on CGPeers so paid is welcome too

Anonymous No. 877803

>>877738
the 'start here' and 'basics' courses here are pretty good:
https://www.sidefx.com/learn/getting_started/

1085x153

file.png

Anonymous No. 877875

>>877803
I tried those but the Pajeet voice... And then the comments like pic related

Anonymous No. 877876

>>877875
>>877803
For reference I ended up going by this tutorial which was pretty okay
https://www.youtube.com/playlist?list=PL2SMrYpOIl0McqZHMV7dONrmIpJTY2WpH
And now I'm starting P2

Anonymous No. 877896

>>877876
Yeah that one's fine as well. Rohan's accent is very thick but he's a decent instructor. But the nine between course or the tim van helsdingen ones would be fine as well as introductions.

Intro Houdini courses have an impossible task tho imo. Theres just so much to cover. Each of these courses will have different stuff in them.

Check out the applied Houdini stuff for dynamics when you want to dig a bit deeper - they're very good at laying foundations. You'll pick up a bunch of general lessons along the way.

393x360

hmmm.jpg

Anonymous No. 879400

Can someone tell me how the Python scripting compares to the VEX scripting in Houdini?
I'm learning with VEX but today I had the tought that if you can pretty much do the same things with Python, it would be smarter in the long run and for other softwares to learn Houdini using Python.

Anonymous No. 879505

>>879400
I'm just learning Houdini now, but I already konw python.
From my short experience what I love about Houdini is that scripting/VEX is a first class citizen, while in Maya/Blender it feels like it's bolted on.
Like in other software it's a modeling application + coding, here it's a coding environment that can do modelling

Anonymous No. 879550

>>879400
python in houdini is single threaded. when you're doing, for eg, look ups on a few million points every frame this becomes a problem.
you can do geometry operations with it but you're not encouraged to.
python's still nice to know though for the boring shit - cleaning up networks etc.

Anonymous No. 879567

>>879505
>From my short experience what I love about Houdini is that scripting/VEX is a first class citizen, while in Maya/Blender it feels like it's bolted on.
in short, you dont know what you are talking about.

Anonymous No. 879873

>>874422
Is it possible to have viewport navigation focus on a particular object like when using Blender? I find it really annoying how in Houdini after a while the viewport would get unhinged and I have to refocus manually by pressing G.

1175x170

1630578995039.png

Anonymous No. 879878

>>879873
you can use a shortcut to set the view pivot where you want

Anonymous No. 879899

>>879873
space + f frames up the selected node.
i guess it might lose focus from time to time, i just bash it autistically and derive pleasure from the act.

536x216

1618922705714.png

Anonymous No. 879901

>>879873
>>879878
consider also using these settings, they work well with custom pivots

1984x676

50mmlens2.jpg

Anonymous No. 879924

A project I've been building to physically simulate a camera in Houdini, still very much a toy. Any ideas why the highlighted rays miss the focal point? The ray tolerance is already pretty high, looks like despite a successful hit the refracted vector is sometimes calculated incorrectly.

1072x757

prism.jpg

Anonymous No. 879926

>>879924
cont.

Here's another case with a basic prism, doesn't look right.

672x378

84861186.jpg

Anonymous No. 880117

To the Houdini wizards in the thread: how can I learn this magic software?

le youtube peoples said I could be hired anywhere if I knew how to use it

1920x878

fence.png

Anonymous No. 880137

>>874422
I did the fence by doing an inset on a top of a box, resampling and the copying the fence segments on the resampled curve. Then I did the little opening using a box boolean.

I am not happy with this. First of all, I do not know how to get rid of those corner segments. One option is to just blast them away, but that leaves too much of a gap. I also did not want to do a box boolean for the sake of creating an opening.

I did a similar scene in Blender, and the way I did it there was to select all the edges, rip them apart, and then scale them individually. This gave me just the right amount of gap in the corners. Also instead of using a box to carve out the front entrance, what I did was subdivide that specific edge, chopped off the middle and then scaled the two points towards each other until the opening was just right. Right now in Houdini, I've tried selecting an edge part of the resampled curve, Del'd it, but that ends up erasing the whole fence. The dissolve operation gets rid of the whole mesh which is obviously not what I had in mind when I pressed delete.

1) How do I rip the edges of a mesh into individual ones?
2) Should I then use a foreach edge loop to scale them then?
3) How do I erase just a part of the side? Blasting an edge ends up connecting the removed ends which I do not want.
4) In Blender I also deleted the top face leaving only the edges, how do that in Houdini?

Anonymous No. 880169

>>880137
For separating curves into individual edge segments I think there's a SOP that already does it but I don't remember what it is. The logic in VEX is simple:

//in a point wrangle
int neighbours[] = neighbours(0, @ptnum);
foreach(int point; neighbours){
if(point > @ptnum) addprim(0, "polyline", @ptnum, point);
}

You could also do it in a vertex wrangle and loop over half-edges connected to the vertex, if the half-edge is primary create the segment.

Anonymous No. 880172

>>880137
(edited)
For separating curves into individual edge segments I think there's a SOP that already does it but I don't remember what it is. The logic in VEX is simple:

//in a point wrangle
int neighbours[] = neighbours(0, @ptnum);
foreach(int point; neighbours){
if(point > @ptnum){
vector neighbourpos = point(0, "P", point);
int newpointA = addpoint(0, @P);
int newpointB = addpoint(0, neighbourpos);
addprim(0, "polyline", newpointA, newpointB);
}
}

You'll want to group the original primitives beforehand and blast them after the operation is done.

You could also do it in a vertex wrangle and loop over halfedges connected to the vertex, if the halfedge is primary create the segment.

Anonymous No. 880193

>>880137
>1) How do I rip the edges of a mesh into individual ones?
Convert Line SOP will turn each segment of a curve into its own primitive. Follow this with a Unique Points SOP to break them apart.
>2) Should I then use a foreach edge loop to scale them then?
Not necessary. You can use a Primitive Properties SOP to scale each primitive from its centroid.
>3) How do I erase just a part of the side? Blasting an edge ends up connecting the removed ends which I do not want.
Won't be an issue now. If you want to make this more procedural you could use a Delete SOP with a bounding box area to determine what gets deleted. That way if you futz with your resample values you won't have to go back and re-select your gate geometry.
4) In Blender I also deleted the top face leaving only the edges, how do that in Houdini?
Not sure if this is what you mean, but you can group and delete by normal direction. In a Delete SOP disable the Number tab and go to the Normal tab. Set the direction to 0,1,0 and use a low spread angle. That will delete all the upward facing polys. The Group SOP has the same functionality.

Anonymous No. 880271

>>880193
>Won't be an issue now. If you want to make this more procedural you could use a Delete SOP with a bounding box area to determine what gets deleted. That way if you futz with your resample values you won't have to go back and re-select your gate geometry.
I'd like to use the boolean subtract to get rid of the corners. If this worked, this would have the advantage of cutting off only up to a specified length. The Primitive SOP works, and this is a similar solution to what I used in Blender, but that would result in the scaling magnitudes being proportional to the size of the box. I'd instead want them to be ~0.5m from each corner.

Right now the boolean subtract does not work on the output of Convert Line SOP.

The Delete SOP just outright removes the points which is not good for two reasons: I have to put in the bounding sphere by hand, I can't just copy them to the 4 corner points and pass them into the SOP like in Boolean Subtract. The other reason is that I'd need to resample it twice since it just removes the points, and I'd need to use a large resolution the first time otherwise it might remove too much from the corner.

I thought that maybe using the Group SOP and passing in the bounding object to it would work, but I get an error.

At any rate, is there are more powerful node for deleting points from poly lines?

Anonymous No. 880273

>>880271
>Right now the boolean subtract does not work on the output of Convert Line SOP.

Install the Labs tools and use Boolean Curve SOP.

Anonymous No. 880278

>>880273
Perfect. Thank you.

1920x883

pool.png

Anonymous No. 880305

>>874422
In Blender, after I did the pool frame, I took the convex hull of it, and bool subtracted the original to get the water. I see that Houdini has a convex hull SOP, so I could go down that route again, but I was wondering if I could do something smarter than that in Houdini. I can imagine non-convex pool shapes that would get broken using such a method.

Anonymous No. 880319

>>880305
I have a metallic construction frame here above the pool, and I want to add a bunch of crisscrossing beams for stability. After that I'll sneak some giant vines through this and distribute the flowers on it. If this was Blender, I could select all the faces and triangulate them. That would turn the quads into tris. This would get me a part of the result, but I'd like a cross shape. So an idea that I'll try here is to loop over each of the primitives and subdivide it. Maybe that will give me what I want.

I'd like it if the main beams in the screenshot stayed as they are, but the newly added edges have a smaller radius.

This is not a huge deal, as I could do a polyframe for the main beams and then the subdivided mesh with a smaller radius which would cover them up, but I'd like to know what the most elegant solution to my problem would be here.

1920x912

frame.png

Anonymous No. 880320

>>880319

1574x643

polyfill.jpg

Anonymous No. 880418

>>880319
One way to turn quad into crosses is to use the Polyfill SOP set to "triangle fan" run over each primitive, see pic.

Anonymous No. 880508

so i was orienting copies yesterday and ran into a problem. solved it, but i'm not sure i understand why it works:

start easy enough:
make a matrix3 from N (maketransform or just get the other vectors manually)
rotate matrix3
convert to quaternion and set as vector4

problem started when i wanted to do another rotation, but use the normal in its post-orient position as the new axis:
tried qmultiply() the N with the orient - didn't work
tried converting the orient back to matrix and multiplying - didn't work
tried building a completely different identity matrix at the beginning, multiplying that with the rotated matrix and then multiplying N with that - didn't work

what ended up working was qmultiplying {0,0,1}

why?

Anonymous No. 880509

so i was orienting copies yesterday and ran into a problem. solved it, but i'm not sure i understand why it works:

start easy enough:
make a matrix3 from N (maketransform or just get the other vectors manually)
rotate matrix3
convert to quaternion and set as vector4

problem started when i wanted to do another rotation, but use the normal in its post-orient position as the new axis:
tried qrotate() the N with the orient - didn't work
tried converting the orient back to matrix and multiplying - didn't work
tried building a completely different identity matrix at the beginning, multiplying that with the rotated matrix and then multiplying N with that - didn't work

what ended up working was qrotating {0,0,1} / multiplying {0,0,1} with the rotated matrix.
why?

Anonymous No. 880515

>>880418
This works, but ended up with a more points (after Fuse) than my Add solution. Thanks anyway.

1920x856

flowers.png

Anonymous No. 880516

>>874422
How do I prevent the flowers from phasing through the construction frames? I scattered a bunch of points, but they are in the wrong place. I suppose I could try painting a mask myself to get a better distribution.

Anonymous No. 880519

How is Houdini's performance when working with dense/high poly models?
Does Mantra use the gpu?
I can't find anything on these questions, that's not at least 5 years old.
And also, how's mantra compared to vray?
If anyone knows and wants to answer this stuff, that'd be great. Cheers

Anonymous No. 880525

>>880516
you could check for intersections the scatter point normals and remove them before the copy:

https://www.sidefx.com/docs/houdini/vex/functions/intersect.html
https://www.sidefx.com/docs/houdini/vex/functions/removepoint.html

won't be perfect,

if you didn't already know:
variables in the vex reference that are preceeded by '&' means the function is writing to those variables, so you need to create those vars beforehand and the feed them into the function.

alternatively, create density masks from attribute transfer or mask from geometry.

or for more accurate, slower checks, you can do volumesample checks post-copy.

>>880519
>How is Houdini's performance when working with dense/high poly models?
fine for me, but i'm a casual user

>Does Mantra use the gpu?
no.

>And also, how's mantra compared to vray?
slow af. vray has a houdini port that's pretty feature rich now.

also, karma is the new houdini renderer. i don't know if mantra is in maintenance mode now, but the bulk of development will be for karma.

karma is also not the fastest. faster than mantra, but slower than everything else.

karma gpu however, is fast, but is in alpha and missing core features like any form of scattering (subsurf or volume). i can't remember if they gate an ETA of mid 22 or 23 before it's production ready.

1993x572

delclosepoints.jpg

Anonymous No. 880530

>>880516
Try deleting points that are close to the beams?

Anonymous No. 880549

>>880525
Oh, man, thanks. This helps a lot.
I'll be in good hands with vray and dini. Thanks a bunch!
I'll give karma a go before.
Cheers!

Anonymous No. 880553

When working with houdini in a pipeline what would you rather spend on :
>$1000 cpu/mobo, doubling your previous single threaded cinebench and quadrupuling your multi cinebench
>or
>$1000 videocard, 12x faster than your previous gpu with rtx enable

1920x858

fence segment loa....png

Anonymous No. 880557

>>874422
I did the fence segment in Blender, exported it as obj, and whenever I open the Houdini file I get this weird error in the message box. Why is it happening and how do I fix it?

Also I know that the middle 3 beans are reversed, that part was easy to fix using the Reverse SOP.

Anonymous No. 880558

>>880557
probably a bad attribute?
run it through a clean node, wipe the attributes.

Anonymous No. 880564

>>880558
Strange that this would be the case, since I made sure to only export the mesh and not the normals or UVs. And looking at its info, that is the only attribute it has.

I tried running it through the Clean node, but that does nothing.

Anonymous No. 880565

>>880564
> that is the only attribute it has.
I meant to write it only has P.

Anonymous No. 880568

>>880553
Anyone?

Anonymous No. 880570

>>880564
not sure then, lad. if you want to share the obj i can have another look.

Anonymous No. 880582

>>880570
https://file.io/gM8n4PjTlCJ3

Let me give this a shot.

Anonymous No. 880583

>>880582
not sure what the issue is but if you just save the obj out of houdini without doing anything to it, it gets fixed.

just right click the file node you use to load it in > save > geometry

Anonymous No. 880584

>>880583
Yeah, that fixed it. Thank you.

Anonymous No. 880589

>>874422
Given a spline, how can I select just the last point of it?

Anonymous No. 880590

>>880568
>>880553
no one has any feedback on whether dollars are best spend on cpu or gpu in houdini and how the gpu accelerated solvers like axiom or opencl accelerated built in nodes factor into this equation? Please help me

Anonymous No. 880596

>>880589
if it's just one primitive:

group range node:
group type: points
start: 0, end: 1
invert range

Anonymous No. 880602

>>880590
Unless you're doing a lot of GPU rendering, the better CPU is more important in Houdini. OpenCL / Axiom are baby shit because you're gonna be VRAM constrained, I wouldn't pin my simulation dreams on them.
Don't skip on RAM either like a retard if you have any interest in doing sims.

Anonymous No. 880610

>>880602
Is 64gb ddr4 enough? The mobo im thinking of getting can support a maximum of 128gb

Anonymous No. 880611

Bare minimum for doing solves IMO. If you go for 64GB I would get higher capacity DIMMs so you still have room to upgrade to 128GB.

Anonymous No. 880654

>>880583
>>880584
Blenders exporters are pretty bad. I was having difficulty getting a cached out alembic to play. Loaded it into Houdini, saved it out and suddenly no issues.

Almost makes me doubt the party line on FBX issues.

Anonymous No. 880657

>>880519
>How is Houdini's performance when working with dense/high poly models?
pretty good.
>Does Mantra use the gpu?
No, but Karma can/does (which is the new renderer that is gonna replace Mantra eventually - but its still a WIP).
>And also, how's mantra compared to vray?
In what aspect? Mantra can look absolutely gorgeous, but it isn't fast. Far from it actually. But it is competent.

Anonymous No. 880714

>>880657
>Mantra vs V-Ray
I was wondering about the capability. (Example: caustics). But it doesn't matter, since there's V-ray for Houdini.
>High poly
30mio.+ polygons per scene, possible without suicidal thoughts, or better not?
My guess is, that it can handle it, since Houdini is the non plus ultra in simulation. I am more interested in models imported from zBrush, for a quick retopo, unwrap and bake.
Apropos baking: is the baking accurate, or does it make a lot of problems with artifacts or strange shadows etc.
I've been using maya until now, which gives me very good results.
Thanks for the answers, lad.

Anonymous No. 880730

>>880714
For a general idea on performance, this classic video is informative:
https://www.youtube.com/watch?v=4VmTBCFZE6Q

But in your use case
>models imported from zBrush, for a quick retopo, unwrap and bake
I think Houdini might disappoint you. The internal remesher is old and strictly single threaded.
Remeshing heavy objects (10M+ polys) can take up to a minute.

No strong opinion on unwrapping, I guess it's competent? A Rizom bridge is available if Houdini's own tools don't cut it.

Baking with the Labs tools is good and quite fast.

Anonymous No. 880772

>>874422
Why do subnets have 4 inputs? Is it possible to change this so they have more or less?

Also, I want to factor some of the functionality that I've made, do I have to turn those parts into HDAs? HDAs end up being separate files, while I just want some functions (in programming parlance) that I could easily reuse. Having a file for every function would create a mess.

I suppose I could try reusing subnets, but I am not sure how to do that from different objects for example.

Anonymous No. 880774

>>880772
>Is it possible to change this so they have more or less?
it is when you convert them to an hda

right click > type properties, is where you can set this on an hda. on a subnet it throws up a warning and says you should really make it an hda first.

not sure what your objection to hdas is, esp vs subnets. you don't have to load all your hdas in with every project. asset > install asset library will let you bring in hdas on a per project basis.

if you're speaking strictly in terms custom functions you've written for vex, you can save them as snippets, or have them as separate libraries that you can just #include

Anonymous No. 880778

>>880714
TopoBuild node is decent enough for manual retopo. From what little experience I have with them in both packages, UV's are a lot more painful in Houdini than Maya but doable.
>>880657
>but its still a WIP
Only on the XPU end..

Anonymous No. 880780

>>880778
>Only on the XPU end..
even the regular cpu version is slow as molasses, even compared to free renderers like cycles. How can you say its not a WIP.

Anonymous No. 880781

>>880774
>not sure what your objection to hdas is, esp vs subnets. you don't have to load all your hdas in with every project. asset > install asset library will let you bring in hdas on a per project basis.
Imagine if you were programming in Python and had to have a different file for each function. I don't yet understand how to deal with HDAs properly.

```
File "<stdin>", line 1
{ 'event_type': hou.hdaEventType.LibraryInstalled, 'library_path' : 'E:/Art Projects/YYY/XXX_Flower.hdanc' }
^
SyntaxError: invalid syntax
```

Also, for some reason I got a Python error when turning a subnet into a HDA. I obfuscated the YYY and XXX, but otherwise it is the same. Should I be worried about this?

Anonymous No. 880782

>>880781
This error seems to have something to do with me changing the default input labels. I can't give the inputs informative names for some reason.

Anonymous No. 880784

>>880780
Because speed aside it's feature complete and stable unlike XPU. And if it's significantly slower than Cycles CPU I'll have to take your word for it since I don't render animations so speed isnt important to me.

Anonymous No. 880786

>>880784
>Because speed aside it's feature complete and stable unlike XPU
It doesnt even have material x lama equivalent, just plain old fashioned material x which isnt good enough and is a dealbreaker.

Anonymous No. 880789

>>880786
"The MaterialX Lama project represents ongoing R&D into materials at ILM, and represents one facet of the larger MaterialX project."
I'm still hopeful they'll open source and we'll get it in a release or two.

Anonymous No. 880791

>>880789
In 2 years - if they choose to release it - there will then be the next thing that they dont have for x amount of years and is exclusive to renderman

Anonymous No. 880792

>>880782
Actually, it works just fine if I ignore it.

Anonymous No. 880819

>>880786
By that definition, every renderer aside from Renderman is incomplete. Obviously, that's a goofy take. On the Karma CPU side there are the gazillion traditional VEX based VOPs, not just the new MaterialX nodes.

>>880789
>>880791
The plan is for the MaterialX and MaterialX Lama standards to be "100% aligned and compatible in the future". From one of the MatX devs. How long that takes is anyone's guess. I've actually really enjoyed working with the MatX nodes in H19.

Anonymous No. 880993

>>874422
I have an HDA which uses another. Is there an easy way to bring the parameters of that child HDA onto its parent? Right now, it seems I'd need to drag 'n drop everything by hand.

Anonymous No. 880994

>>880993
Ah, it turns out I can just drag the whole node and it will do what I want. At times like these I love Houdini.

1920x886

branch.png

Anonymous No. 880998

>>874422
What do you do in Houdini when passing geometry to an HDA is not enough? I want each of the smaller flowers to have variation, but I can't think of a way to pass a geometry generator.

If Houdini was a functional language I'd pass a higher order function as an argument.

Anonymous No. 881001

>>880998
if i'm getting this right:
>you have an an hda that makes flower variations
>a different hda that's copying these flowers

and you want some way to pass in the flower variation hda (by connecting it to input 0 for eg) so when it's being copied it can use some point attribute to generate variations?

i'm not totally sure you can. these days you can't really send data back up the network in houdini and do things that way. the old copy stamp node sort of worked like that. the dopnets kinda work like this but not really.

i can think of a convoluted way of using python to basically rebuild the node connected to input 0 of the hda inside the hda itself and then connect that up to wherever you want. i've never done this, but i remember video about using python to save and rebuild entire networks:
https://www.youtube.com/watch?v=SY6NyECDwpk

targeting nodes connected to specific inputs is possible as well as traversing contexts within python, so you should be able to do this if you really want.

you could re-consider your approach to the problem though.

Anonymous No. 881003

>>880998
I wonder if it is possible to pass VEX scripts via channels? I can't see a type for that in the param window.

Anonymous No. 881018

>>880998
I think you could use a for loop, make your HDA so that it uses a seed variable (you'd carry that in a geometry attribute), and then loop the HDA over the outputs you want as a base for further variation, taking different seeds. look in particular into the metadata node in for loop blocks

Anonymous No. 881019

>>881003
you can write VEX in a string parameter and then channel that back (`chs('path_to_parm')`) into a VEX wrangle
note that every change prompts a re-compile, so it can be slow

Anonymous No. 881029

>>874422
I know how to get the first point of a curve using a Group by Range, but how do I use that as the pivot point of the transform? The Primitive Properties SOP transforms around its centroid, and if I use a Foreach primitive loop, I do not know to connect the initial point to the pivot of a Transform node.

Anonymous No. 881038

>>881029
https://www.tokeru.com/cgwiki/index.php?title=HoudiniVex#Rotate_prims_around_an_edge.2C_alternate_version

Anonymous No. 881042

>>881038
```
int points[] = primpoints(0,@primnum); // list of points in prim
vector p0 = point(0,'P',points[0]);

@P -= p0;
@P *= chf("factor");
@P += p0;
```

So something like this then. Thanks.

Anonymous No. 881043

>>881042
not really

788x960

1637748569996.jpg

Anonymous No. 881044

Brehs, I have an ignorant question. I'm trying to make some modular environments for Unreal and was wondering if making the modules in Blender and importing these files into Houdini to make procedural buildings/houses is a good idea instead of doing it all in Houdini?
I feel it could give me more artistic freedom than what Houdini provides

Anonymous No. 881045

>>881043
Well, I only needed to scale it for my particular problem. But this leaves me wondering how I could reuse the transform node for this task. Once I drop down to Vex, I have to do it all in Vex.

Anonymous No. 881047

>>881044
Yeah probably, destructive modelling is much quicker in Blender than Houdini. Question is if you intend to use Houdini Engine in Unreal, and how familiar you are with Houdini in general. Might be better to just use Blenders Geometry Nodes or create your own tools in Unreal.

Anonymous No. 881053

>>881047
Well I started 3D with Houdini somewhere around last year but I only did some beginner stuff like a procedural scaffold and house, some basic VFXs and just stuff that resulted from some tuts.
But yeah I also plan on using it for a procedural terrain that would also have some interest point scattered (also placed programmatically) and some other stuff but what really got me into this software is the potential of making procedural architectures/roads/terrains.
I bought the two book of Junichiro Horikawa thinking it would help but only the tiling ones will serve me.
I also got the Anastia course on a procedural Lake House but it's too detailed for me, I don't really need to build even the brick from scratch. So yeah in the end, I plan on using Houdini a lot.

If I had to say it more clearly: I need some resources that'll teach me to make a house generator with modular pieces. The pieces being for example, walls, roof parts, doors, windows, vents, air conditioner, etc, ...

Anonymous No. 881058

>>881053
you know there's a building generator in labs, right?

Anonymous No. 881069

>>881058
Yes even tough I never used it but my goal is to have as much as freedom as I can when deciding for the layout or style of my house. I don't want necessarely a building but I'd like also to be able to make some traditional japanese house or whatever

Anonymous No. 881070

>>881069
you know you can just double click that node and see how they built it right?

Anonymous No. 881072

>@numpt instead of @ptnum
>forget to set normal map to Vector3
>half an hour to get a node to work, only to find the solution by reading the entire documentation on it instead of only what I thought was relevant.

Maybe I should have stuck with crayons.

558x614

file.png

Anonymous No. 881080

>>881072
some day in the future you're going to spend a day wondering why your code isn't working, only to realise that houdini will assume all your custom attributes are floats unless you're explicit about type.

get into the habit of doing that now, even with the default attributes.

1920x911

vines with branch....png

Anonymous No. 881221

>>874422
I am happy with how this came out, but under the hood I messed up the architecture. I imagined it as big vines -> branches -> flowers, in other words as an object hierarchy. But instead what I should have done is each of the the phases produce a group that the later ones can distribute on, in other words vines -> vine_stalk_group -> branch -> branch_stalk_group -> flower -> flower_stalk_group

This would have brought out the performance benefits of parallelism and made the node tree easy to deal with. I won't bother redoing it from the start, but I am going to have to look into whether Copy To Points can merge point groups of the objects it is copying. I expect it should since this is such an obvious approach to take given Houdini's design. I just did not realize that Houdini is a first order language until yesterday.

Anonymous No. 881225

>>874422
Is there a way to prevent Scatter and Align SOP from tagging the points with the 'scatteralign1' string. Does that serve any kind of purpose?

Anonymous No. 881231

>>881225
it picks it up from the "Tag Name" field.
you can always delete it later, but if it really bothers you, you can edit attributewrangle14 inside the node.

Anonymous No. 881237

>>874422
I am trying out Orbolt for the first time, and it is messed up. The Launch In Houdini option does nothing despite me turning off UBlock, and when I try to log in from the help pane using the internal browser it does not accept my password despite me pasting it from Chrome settings.

> [11652:11288:0209/164254.450:ERROR:device_event_log_impl.cc(208)] [16:42:54.442] Bluetooth: bluetooth_adapter_winrt.cc:1060 Getting Default Adapter failed.

I also get this error when I using Orbolt from the help pane. Some docs say that I should use the Asset Browser pane, but that thing is just a still image that straight up opens Chrome.

https://www.orbolt.com/asset/theis::furnitures

Leaving that aside, I decided to download the above asset manually, and I've installed it, but I can't see the nodes anywhere. Just how am I supposed to use this?

Also, why is the extension '.otl' instead of '.hda' like I expected it would be?

Should I have clicked Install and Create instead of Install while installing it?

Anonymous No. 881239

>>881237
Ok, it seems like it works on the object level and I missed it.

Anonymous No. 881241

>>881221
I have no idea what I'm looking at but it's cool to follow your progress Anon. Groups are just boolean attributes on the geometry so yes, they will be treated as one big group after the copying.

Anonymous No. 881254

>>874422
How do you use a channel on something like Group Combine SOP fields which take in string inputs?

Anonymous No. 881418

>>874422
```
i@variant = detail("../foreach_count1/","iteration",-1);
```

I have this inside of an Attribute Wrangle inside loop block and it does not work for some reason. When I try to put `detail("../foreach_count1/","iteration",-1)` inside an Attribute Create it starts working. To make things really confusing, `detail` is ignoring its third argument which is supposed to be the default when it cannot find the input.

Why is this happening?

Anonymous No. 881420

>>881418
> The last argument is always ignored. It is just there so you can change a prim/point/vertex call (which each have an element number argument) to a detail call by changing the name without having to change the arguments as well.
That last argument is not the default argument like I had assumed. It says in the docs that it is always ignored, but the wrangle breaks when I set it to anything other than 0.

Anonymous No. 881422

>>881420
```
i@variant = detail(1,"iteration");
```

I got it to work like this. I connected the metadata loop node to the attribute wrangle which allowed me to use it as an index. Why did passing a path not work?

Anonymous No. 881424

>>881422
i think it would be reading it as a string.
there's some stuff about using backticks (`) to force hscript expressions to evaluate but i have no idea if it works in a wrangle.

Anonymous No. 881619

>>874422
Is there a way to produce several outputs from a for loop?

Also, the Scatter and Align SOP complains really loudly when I pass it geometry that has mixed regular and packed inputs. Every time I click on it, I get a Python exception saying that `Advanced Drawables only support polygonal geometry`. Is there a way to deal with this, it is very annoying.

1920x927

finally done with....png

Anonymous No. 881665

>>881619
I think I finally understand how to organize the node tree. For getting multiple outputs out of a loop, the best way is to pack the geometry and put it in a singleton group. When geometry is packed it is easy to merge it together with other kinds without all of them having to have the same types of attributes. Initially I thought that packing is just good for sharing geometry, but it good for organizing as well.

This way of doing things has the disadvantage of requiring splitting and unpacking down the road, but the extra bookkeeping is worth it. The node tree ends up being clean enough, and the Scatter and Align SOP is no longer complaining.

I actually had something similar done a few days ago, but back then it required 4gb and a minute of baking for only half the geometry. This bakes in a snap and takes only 37mb.

It is a relief to finally be done with these vines, being able to procedurally generate them is why I picked up Houdini. I didn't like how inflexible Blender was when it came to generating instances. Now I just need to roll up my sleeves and finish the scene. I feel like I am starting to understand Houdini.

Anonymous No. 881845

Does anyone have experience with character animation in Houdini?
I'd like to switch from maya to Houdini, Mari and Zbrush only.

480x360

b96.webm

Anonymous No. 881848

>mfw surfacedist function

Anonymous No. 881850

>>874422
What does the second input to the Edit SOP do? The example provided in the docs literally changes zero after I cut the wire.

Anonymous No. 881853

>>881850
Also is there any way of making the edits local?

I'd really love to be able to put an edit node and a transform node after it so I can position the object, but that does not give me the ability to edit the upper node. The handles end up being positioned where the original geometry was which makes working with them impossible.

Anonymous No. 881854

>>881853
It seems it is possible to make the edit node edits local by plugging in the original geometry as the second input. This is nice and actually makes working with edit node possible. But it needs to be input before making the changes.

1920x925

hook (bad).png

Anonymous No. 882020

>>874422
Note how in the bend, the hook mesh intersects itself. How can this be fixed? The Sweep SOP is not going a good job here.

Anonymous No. 882281

>>877896
> Yeah that one's fine as well. Rohan's accent is very thick but he's a decent instructor.
> https://www.iamag.co/free-tutorial-creating-art-in-houdini/
This free course by him is quite decent.

927x805

slack_tAsGw2I4QU.png

Anonymous No. 882611

https://www.disneyanimation.com/resources/moana-island-scene/

A USD version of the Moana island scene has been posted.

Anonymous No. 882612

>>880305
Clip the top, use the Group Expand to a do a flood fill selection of the inner primitives, Blast the outside part, PolyFill the top. Reverse + face Normals after that to fix the normals.

Anonymous No. 882664

>>882611
what server is this?

Anonymous No. 882687

>>882611
Great...but in the end you're messing with a stupid island from a stupid movie.

Anonymous No. 882697

>>882664
Academy Software Foundation's Slack server

Anonymous No. 883145

thought i'd save u boys the trouble of getting some of the from cgp.

CGCircuit โ€“ Advanced Particles 1: Geometry Based Simulations
https://www.bilibili.com/video/BV1ZT4y1X7gw

CGCircuit โ€“ Advanced Particles 2 : Demon Fire Trails
https://www.bilibili.com/video/BV1f5411o75U

CGCircuit โ€“ FX MAYHEM 102 โ€“ Smokeless Fire & Flames
https://www.bilibili.com/video/BV1Qq4y147AA

CGCircuit โ€“ Houdini Elements โ€“ Smoke
https://www.bilibili.com/video/BV1km4y1d7sY

CG Forge - Destruction (all 3?)
https://www.bilibili.com/video/BV11Y411L7jW

VFX Grace โ€“ Advanced Tsunami Simulation
https://www.bilibili.com/video/BV1Sb4y1j7tW

Sapphire FX โ€“ Ship Destruction Weeks 1-6
https://www.bilibili.com/video/BV1434y1y7r7

Anonymous No. 883172

>>883145
thanks dude - perfect timing. I just installed the Axiom solver.

Anonymous No. 883487

>>882687
This is the tech-art equivalent of 'I don't look at reference lol'

Anonymous No. 883493

>>882020
too many fucking points on that curve, bruh
you even have a resample there, what are you doing?

Anonymous No. 883538

>>874422
is this good for Houdini?
https://renderro.com/pricing/

I don't have 8k to buy a pc do learn houdini

Anonymous No. 883547

>>883538
you can learn houdini on a potato.
go away and stop making excuses.

479x790

file.png

Anonymous No. 883558

started a little set up that spits out a nicely kine-fx rigged leaf/petal starting with a line+ribbon sweep

:)

Anonymous No. 883589

>>883493
It loses too much detail with less. At any rate, the way to fix this is to pick the corner point, fuse it with the original curve the second input and once the corner points have been dealt with do a sweep.

One other option would be to do less points and the subdivide it, but I do not like the result of that.

Anonymous No. 883709

>>874422
Why is everybody learning Houdini now?
What happened?

Anonymous No. 883754

>>883709
Why aren't YOU learning Houdini, anon?
In all seriousness tho who is everybody?

Anonymous No. 883764

So does an attribute wrangle create the attributes only after it's finished running so I you have to work with variables or make a new node?

So in the example of creatting a point with addpoint() and then reading it's position with point() it would return null if on the same attribue wrangle?

Anonymous No. 883772

>>883764
So I ran into similar issues lately and you're confusing a couple of things. I'll start with your example:

The point() function (and similar functions) explicitly query the input the you're specifying.

The geometry being processed inside the wrangle is not the same as specifying input 0 on these function. As far as I can tell you cannot access the in-memory geo using these functions.

In your example the point you've created exists on the in-memory geometry, but when you call input 0 from point() it's not there.

eg: a single point (any geo will work) connected to a point wrangle

i@newvar = 1;
// note that this line looks for newvar on geo coming in from input 0, not the in-memory geo
int newvar = point(0, "newvar", @ptnum);
i@newvar += newvar;

Doing ^ means newvar is just going to be 1.

But your first question:
>So does an attribute wrangle create the attributes only after it's finished running so I you have to work with variables or make a new node?

is actually different and the answer appears to be that you can create and update variables inside the same wrangle:
i@newvar = 1;
i@newvar++;

newvar = 2 here.

Anonymous No. 883774

>>883772
Thanks, I get it now.

Anonymous No. 883888

>>883772
>create and update attributes*

1920x937

bad normals.png

Anonymous No. 883892

>>874422
Why are the normals messed up in the the stage context? This looks just fine in the obj context.

Anonymous No. 883901

>>883892
You need to explicitly add a normal attribute to your models in the stage context. In objland the scene viewer is automatically adding some default normals for display, which the Solaris viewer isn't, for whatever reason. Easiest way is to just slap a Normal SOP down at the end of your network, but obviously that will give all your models the same normal angle, which may or may not work out.

Anonymous No. 883903

>>883901
You're right. I thought that I had the normals on the offending object, but I didn't.

1920x931

bad uvs.png

Anonymous No. 883905

>>883901
That fixes the normals, but now I find that my UVs are messed up. They look fine in the regular and Karma CPU mode, but show up like this when using the Karma XPU renderer. Would it be possible to fix this?

Anonymous No. 883907

>>883905
Not sure what's going, but in case you weren't aware, XPU is a tech preview and won't be production-worthy for at least another couple Houdini versions.
Is that object set to render as subdivision surfaces in /obj by any chance? That looks like UV stretching you would see with SubDs sometimes.

Anonymous No. 883910

>>883907
>Is that object set to render as subdivision surfaces in /obj by any chance?
No. Maybe it has something to do with the way the object tree is structured. I have packed prims and another object that does not have UVs, but I do not want to start taking things apart to figure out what is wrong.

I think at this point I have two options:

* Do the lookdev in Houdini using the regular rendered and let the Karma CPU grind away at the finished scene for a while.
* Find a way to export the scene into Blender and do the lookdev there using Cycles.

Yesterday I watched the Entagma guys talking how easier it is to do the lookdev in Blender, and I am mostly interested in Houdini for generating procedural environments rather than doing simulations, so that kind of workflow might be a good fit for me. The reason I decided to try out Houdini is to generate those big vines.

What are my options for exporting given that I have the non-commercial license? The regular export options are grayed out, but I saw a video on exporting geometry into .obj using the File SOP. I've never tried it so I am not sure whether that would deal with packed instances correctly. There are about 20k of them in the scene and I do not want them to be put into a collection, let alone output as separate geometry.

Anonymous No. 883913

>>883910
>let alone output as separate geometry.
Tried this and it generates a 2gb file. No, I'll have to pass on going to Blender.

Anonymous No. 883915

>>883910
How is Blender's USD import and conversion to native data? I can't really think of another format that will translate instances/packed prims.
The other option would be to buy/pirate Houdini Indie + Redshift or Octane or VRay. All of those will have capable GPU rendering.
Or just use Karma CPU for the time being like you said. I'm personally doing that, but I'm not currently rendering animations.

Anonymous No. 883922

>>883913
The Mtlx Standard Surface shader tricked me into thinking it works with the raster renderer. It seems I can consider the Mtlx SOPs being in the Alpha stage as well.

1170x947

First Principles.jpg

Anonymous No. 883972

>>874422
What are the First Principles of Houdini?
What should I learn?

Anonymous No. 884052

>>883972
>What should I learn?
how to google

Anonymous No. 884062

>>874422
I am playing around with color nodes and am trying to make passing point attributes into Attribute Adjust Color SOP. I've set pattern type to Noise and its type to Simplex. Changing the Element Size directly in the parameter window works, but setting @elementsize to anything in the wrangle does nothing.

This is the first time I am trying to control a parameter using an attribute, so I might be wrong, but shouldn't this be possible? I've seen it work in the shader context.

Anonymous No. 884251

>>884062
Try hscript point() expression

Anonymous No. 884256

>>884251
It worked!

>>879901
Tumbling works well, but scrolling with the mouse wheel ends up unfocusing the viewport unless I hold L. L is far from my left hand and is hard to press correctly. Is there a way to set it up so the mouse wheel zooms on the object instead of the cursor?

1920x932

bad volume lighting.png

Anonymous No. 884258

>>874422
https://youtu.be/WTg_pVZoac0?t=625

Why does the noise pattern looks so much better in the video? For some reason I am getting extremely dark and light spots. It does not feel like I messed up the settings.

Anonymous No. 884259

>>884258
Nevermind the question, clamping fixes the noise being in the [-1,1] range. I guess the author got lucky and did not get negative noise in his particular shot. It seems like in Vex, there is a Fit node in the VOP context so that might be a better option.

Anonymous No. 884262

>>884259
Why does the ramp not work in the VOP context. I set its name to $OS, but that is not the problem. Moving the point does nothing, and changing its Ramp Type type to spline just deletes the volume. I am trying to manipulate the volume's density.

Anonymous No. 884272

>>884262
the ramp parameter node?
there's a weird quirk to it - you can't tweak it from inside the vop. go up a level and you'll see the interface exposed on the vop parameters. by default it's set to zero. don't ask me why.

Anonymous No. 884293

>>884272
Thank you. Another question, what does the debug flag on the VOP nodes do? I tried Googling it and looking in the docs, but I can't find an answer to it.

How do you see the output of intermediate VOP nodes?

Anonymous No. 884741

what's the best way to learn vex?
I don't remember the code I'm learning

Anonymous No. 884803

>>884741
It is basically C with a library and some language specific extras bolted on it. You are going to need to learn how to program in order to use it.

If you don't want to dive into that, you might have an easier time using VOPs which have the same functionality, but in a node based manner. For long programs, text is more efficient than stitching nodes together. But I feel that VOPs have less footguns. Vex inherits a lot of C's bad traits as a language.

Anonymous No. 884866

>>874422
In a CHOP2 network, is it possible to get a standard color ramp without having to make one in a VOP node?

Anonymous No. 885082

>>874422
How much Maths do you need to know to master Houdini?
Any good Maths classes on the tubes?

Anonymous No. 885084

>>885082
college level math

Anonymous No. 885159

>>874422
Has anybody else managed to get the pirated version of Houdini 19 working?

I tried it 1.5 months ago and could not get the XFORCE generated keys to get accepted, so I went for the Apprentice version, but now I am far enough in my work that I either need to get it working so I can export the scene to Blender and do the lookdev there, or ditch it entirely. Houdini just sucks at lookdev so badly, I can't imagine doing it in Mantra. Karma XPU would have been good enough if it actually worked, and the 3rd party renders are banned by design.

Anonymous No. 885161

>>885159
>Has anybody else managed to get the pirated version of Houdini 19 working?
Nwm, I got it working. I think I just fucked up the way I put in the keys last time. Phew.

Anonymous No. 885204

>>885161
But even if I can pirate Houdini, it is useless to me. Blender can't import USD instances yet so I'd get the scene taking 100x the memory even if it was not wrecked due to the upgrade, and I am can't find a 3rd party render. I am 95% sure that the crack for Redshift I am looking at is a virus and I can't find anything trustworthy. There are a bunch of old Octane/V-Ray/Reshfift versions for C4D and 3DSMax that I can find, but nothing for Houd. I phew'd too soon.

Did anybody have luck pirating the renderers themselves?

Anonymous No. 885205

>>885204
Arnold/htoa works. If you're on python 3:

> Arnold 6.2.1.1 HtoA 5.6.3.0 - Houdini 18.5.596 Python 3 /

you can use the crack from there, but grab the installer straight from solidangle

There's a crack for v-ray as well but I haven't tested it.

Octane is a no-go. Redshift is crack is from a few years ago.

Anonymous No. 885338

>>885205
> There's a crack for v-ray as well but I haven't tested it.
Using the Launcher, I've tried updating the pirated Houdini 19.0.383 to 455 and it passed. Now I just need to download V-Ray for Houdini 19.0.455 from a certain pirate site and I will be set. Just why is V-Ray so damn large? It is 4gb spread across 4 files. One of the places I am downloading has 20kb/s speeds. Good luck with that. With these speeds, I am actually at risk of the connection breaking and having my downloads wasted.

Anyway for anybody wondering, I managed to find downloads for Houdini 18.5.499 + V-Ray for Houdini 18.5.499, though they are on different sites. Probably the easiest way to pirate Houdini without all this hassle is using the Rutracker site. There is a bundle of Houdini + 3rd party renderer of various kinds, it goes up to 18.5 though. I haven't tried it, but if what I am doing now fails me that is what I will go for.

Sigh, I was determined to go back to Blender, but despite the many flaws that Houdini has, I am really into the procedural workflow. I just can't resist it. I'll buy it for real when I got the point where I am making income from this.

Anonymous No. 885346

i'm doing the usual curve > carve > sweep thing but i want to scatter points on the swept geo and have it stick.

unfortunately attribute interpolate doesn't quite work because the prim count is changing. anyone have any ideas?

Anonymous No. 885357

>>885346
nvm, worked around it
:^)

Anonymous No. 885378

>>885338
Got it all installed! It turns out 3.45gb is the material library, and 360mb is in an useless zip file that is there for manual installation. The exe file does not use it, and will also ignore and (I conjecture) try to download the asset library from the net despite it being on the hard drive. The Houdini launcher did that kind of stupid thing to me when I got the ISO file, the data was there and the launcher ignored it in favor of downloading it from the net.

The actual V-Ray is only 400mb, so it is not too extreme. I haven't actually tried rendering, but I'd be very surprised if it failed at this stage. I am really happy that I can continue learning Houdini. Blender is not bad by any means, but I like doing it the Houdini way the best. I'd have been happy to stick with the Apprentice version had Karma XPU had been production ready, but getting past the tooling obstacles is a part of the path of being an artist.

Anonymous No. 885384

>>885378
Someone on cgpeers did mention they had trouble during an animation render but that could just be the build or any number of other things. I'm pretty sure the installers are available to download directly from chaosgroup and you can probably continue to use the cgauth.dll for newer builds.

Anonymous No. 885418

>>885384
> you can probably continue to use the cgauth.dll for newer builds.
Thanks for the tip. I'll keep this in mind if I run into trouble.

Anonymous No. 885461

>>874422
I am trying the .rat HDRIs that come bundled with Houdini, and when using the V-Ray renderer they are pure black when passed in as textures to the dome light. Regular .exr HDRIs work fine though. Any idea why?

The .rat HDRIs show up fine in the regular viewport.

Anonymous No. 885464

>>885205
why is HtoA not as snappy as C4DtoA

C4DtoA has the best Arnold integration, even better than Maya's native Arnold

Anonymous No. 885625

noob here.
i have a bunch of parameters which are scattered across different nodes, is there a way to create a separate menu and add those params to it to conveniently edit them all at once? I think some time ago I saw a guy doing this in some tutorial, but can't find it anymore

217x216

cringe.jpg

Anonymous No. 885629

>>885464
>even better than Maya's native Arnold

Anonymous No. 885636

>>885625
For any node, click the gear icon next to the name and Edit Parameter Interface. That will allow you add the parameters. You can also do things like dive into the object and drag it into the interface.

Anonymous No. 885643

>>885636
thanks

Anonymous No. 885724

>>885625
Make a Houdini Digital Assett...

1280x720

metal piece.png

Anonymous No. 885754

>>874422
Playing with noise and I got this thorny pattern. Somehow it took me over 4h to do the geometry for the piece, when I could have done it in 5m. Next time I'll just do it in Blender. I've learned my lesson, it is not worth playing around so much when I should be studying shading.

V-Ray is pretty nice. The first day of using it, it kept crashing like crazy. This gave me a negative first impression, but now I've learned how to get around where that is likely to happen. I like how the nodes are designed and the docs are well written (with the exception of UV mapping nodes.)

1280x720

metal piece disp ....png

Anonymous No. 885853

>>885754
I had a lot of issues with displacement maps and I finally figured out how to deal with them. Trying to mess with render settings in the material just lead to slowdowns and crashes. I had better results with geometry. To get a quad topology (like with Blender's remesher) you can convert to VDB and back with high adaptivity and that will give you very smooth topology. But this ended up increasing the memory burden by 100x. The original gizmo was 180k, but after converting to VDB and back it would be 20mb which is way too much. I played around with the Remesh SOP and got really good results. As it turns out you can just put in * into the Hard Edges Group and it will preserve all of the original geometry.

In the screenshot the huge planar ngon left after boolean cutting the hole has seams running towards the top right corner. This is due to the triangular subdivision V-Ray does under the hood to place the displacement maps running into them.

(1/2)

1280x720

metal piece remes....png

Anonymous No. 885855

>>885853

Here is the result after remeshing the geometry with all the old edges left intact. The uniform target size was high and only increased the memory load by 2x. Pretty nice. I am glad I could find such an elegant solution to this problem.

1112x600

curvature.jpg

Anonymous No. 885948

How do I interpret mean curvature form the measure sop, what's the range and how does it correspond to surface angles? Gaussian curvature seems to be just the angle defect from a flat surface so the range is between -pi and +pi with zero being flat.

Anonymous No. 885949

I've heard and seen some good stuff about houdini. Is it worth learning as an indie game dev? Can you do any real time stuff with it or is it just long ass renders?

Anonymous No. 885986

>>885948
Good questions. I am playing with that SOP right now and it is going bonkers. The scale normalization does not work and the visualizer just gives up depending on what I put in. This kind of buggy behavior makes studying it really difficult.

487x543

file.png

Anonymous No. 885989

>>885853
there are labs nodes for instant meshes (free) and exoside quadremesher (paid/pirate). instant meshes node is pretty barebones - the exoside one is pretty good.

at the end of the h19 keynote they did say that there was some major quad remeshing stuff coming for the next major release. hopefully regular remeshing gets an overhaul as a consequence as well.

>>885948
>>885986
pic-related

>>885949
there's plenty of game-dev talks on a variety of subjects on houdini channel on youtube

the new project titan thing they're doing is intended as a ue-tools show case i believe.
https://www.sidefx.com/titan/

Anonymous No. 885991

>>885986
To make use of its output, what is really needed is to remap it to [0,1]. For that what is needed is to first turn on the visualizer by checking one of its options and bake it into the output. Then you can use it as a color in the later nodes. I can't really figure out what the purpose of Gaussian, Mean and Principal is, but the last one seems to actually calculate what the curvature should look like.

Anonymous No. 885996

>>885989
>there are labs nodes for instant meshes (free) and exoside quadremesher (paid/pirate). instant meshes node is pretty barebones - the exoside one is pretty good.
>at the end of the h19 keynote they did say that there was some major quad remeshing stuff coming for the next major release. hopefully regular remeshing gets an overhaul as a consequence as well.
Good to know

>pic-related
I read the docs, but even with that information it is really hard to understand what the purpose of all these measurements is and how to use them effectively.

1920x910

gaussian curvature.png

Anonymous No. 886005

>>885991
For the Gaussian curvature, its values are all mean centered, so if you remap them so that the area around 0.5 is 0 and increases outwardly then it gives good results.

You can see that in the visualizer, if you make the mid black and the rest white, it comes out well enough. It looks quite similar to Curvedness, but if you look at the distribution of values in the histogram it is a normal distribution, while the Curvedness is a log-normal. I guess that is why it is called the way it is.

Let me ask a question. Is it possible to blur the color attributes? I tried the Attribute Color SOP and it only worked on position.

1475x767

curvature2.jpg

Anonymous No. 886016

>>885986
>>885989
Seems like the range of principal curvatures (when scale independent) is between +pi and -pi as well. It is clear how that relates to surface angles, a 90 degree bend is pi/2. But why does the value jump from the max column to the min column depending if the surface is convex or concave (pic), in this case bending a flat plane should only change the maximum curvature, no?

No idea what the fundamental algorithm is either, would be nice if the docs explained it. In comparison to the Labs curvature node the results look most like Taubin's curvature but the principal directions are flush with the polygon surface, not tangent to the normal plane like is convention.

Anonymous No. 886022

>>886005
Use the Attribute Blur SOP, just change the target attribute from P to Cd.

Anonymous No. 886023

>>886016
I do not think the range has anything to do with Pi. Rather the normalization is done using the L2 norm, by dividing using the standard deviation. After that is done, the individual values can go above 1.

The docs say the curvature is calculated using the principal components based off the normal. I am guessing the first principal component is the tangent, and the other is the cross product between the two. Or maybe it is using something else. The Labs Measure Curvature SOP has a bunch of ways of measuring it.

Anonymous No. 886024

>>886022
It doesn't work.

Anonymous No. 886027

>>886023
>I do not think the range has anything to do with Pi
I'm bending a plane and looking at the raw numbers in the speadsheet, the values seem to be just radians. Note that this is with scale invariance turned on so surface area has no effect. Normalization must happen after the fact.

>The docs say the curvature is calculated using the principal components based off the normal. I am guessing the first principal component is the tangent, and the other is the cross product between the two

Yeah, as the node can output curvatures on both prims and points that's probably how it's done (build a matrix of surface derivatives with triangle as base, find it's largest eigenvector & eigenvalue with PCA (max curvature and direction), then look in the orthogonal direction to find min) but the docs don't explicitly say what their super secret algorithm is.

Anonymous No. 886028

>>886024
Is your color on a texture? If it's a point attribute Attribute Blur should work.

Anonymous No. 886039

>>886028
>Is your color on a texture? If it's a point attribute Attribute Blur should work.
No, it is just a point attribute. Right now, I am looking at the docs, and it does seem like Attribute Blur/Smooth should work. Weird.

Anonymous No. 886040

>>886039
Ah, I get it. Previously I tried using it on a primitive attribute which is why it failed. Thanks.

3114x1834

houdini cutie mac....jpg

Anonymous No. 886172

>>874422
we mac now

Anonymous No. 886370

>>874422
When packing, is it possible to transfer the detail attributes? Not being able to do so really makes dealing with V-Ray materials a chore as I can't just unpack, subdiv and then repack some of my objects like before, but have to make separate versions.

Anonymous No. 886393

>>886370
Could you Attribute Promote your detail attrib to prim temporarily and then promote it back to detail afterwards?

Anonymous No. 886405

>>886393
Yes, but if I had to do that I might as well use prim attributes for the material paths all the time. Which is what I've decided to do instead of splitting hairs about it.

1280x720

flower v1.png

Anonymous No. 886434

>>874422
My first unbroken attempt at rendering. If I was good at this I could have done it in half an hour instead of 12, but I'll get there. It is not bad at all when it works, but given all the bugs V-Ray has it boggles my mind that it is a paid software costing 80$ a month. If I did not know better, I wouldn't have believed it is out of beta.

As an example of one of the bugs I've run into - the floating power operation literally does not work. I had to use the vector power instead. Just why do the float/vector/color ops feel like they've been passed onto the janitor? If it this bad now after it has been out for over two years, I can hardly imagine how it was before.

1280x720

flower v1 (closeup).png

Anonymous No. 886526

>>886434
I meant to post this yesterday, but I was too tired. I'll finish the actual scene in a few days. The leaves are a bit plastic looking and I am not sure how to improve them. Any ideas?

Otherwise, I want to ask about other renders. How stable are Redshift and Octane?

Its is impossible to find reviews on them. The one Youtube review that I saw did not mention the extreme buginess of V-Ray at all. If I had to buy a subscription, I would be thinking of switching to something will less pitfalls.

Anonymous No. 886616

>>886526
Here is V-Ray when it works well. I had to render it for an hour to get rid of the noise, and this image should make for a decent wallpaper. It is half complete at this stage, I still need to add the girl and make the camera below the surface of the pool. There are about 300 branches with flowers on them and 5000 flowers in the scene. Not all of them were captured in this particular shot. Not something I've want to attempt painting by hand. Houdini attracted me due to its ability to scatter points and instantiate the objects. Blender has started to get the capability to do this with its geometry nodes, but it nowhere near Houdini's level. After I do the above as well as finish the rest of the scene, I will have completed my first mastery challenge. I hope I can get faster at making these kinds of illustrations.

1920x1080

vines and shards v2.png

Anonymous No. 886618

>>886526
Here is V-Ray when it works well. I had to render it for an hour to get rid of the noise, and this image should make for a decent wallpaper. It is half complete at this stage, I still need to add the girl and make the camera below the surface of the pool. There are about 300 branches with flowers on them and 5000 flowers in the scene. Not all of them were captured in this particular shot. Not something I've want to attempt painting by hand. Houdini attracted me due to its ability to scatter points and instantiate the objects. Blender has started to get the capability to do this with its geometry nodes, but it nowhere near Houdini's level. After I do the above as well as finish the rest of the scene, I will have completed my first mastery challenge. I hope I can get faster at making these kinds of illustrations.

Anonymous No. 886623

>>874422
How does Houdini compare to Clarisse?

Anonymous No. 886624

>>886623
anon, i....

852x754

rubbertoy.jpg

Anonymous No. 886683

Do you guys have ideas how to fix/filter "bad" normals? Pic is from inside the rubber toy model, the corners of the mouth are very janky. I tried the Smooth SOP but it also slightly disturbs perfectly fine normals elsewhere on the mesh if the filter is low or doesn't smooth enough if the filter is set too high. The ideal solution would be something that's very light to compute as I need to run this inside a solver.

Anonymous No. 886861

>>886623
I've been trying it for a few days. My conclusion is that it is a vastly better option for lookdev than V-Ray in Houdini. I am going to try exporting that scene and redoing the texturing in it. Trying to do lookdev in Houdini is torture.

https://youtu.be/i2QYOnqTc4M?list=PLOZhw3kXDRQelts3VZKzhPySQi2JAofUf

Houdini already starts to struggle with 5k flowers on the screen, but in this video he scatters around 30k trees of 720k polys each with no appreciable slowdown. I've tried it myself and Clarisse's ability to handle large amounts of geometry is magical. Despite being mostly a CPU renderer, it is actually faster than V-Ray with the GPU. This piece of software definitely deserves to be more widely known.

It can't be used to create geometry on its own though, so you have to use in conjunction with another 3d package that can do that like Houdini.

Anonymous No. 886949

anyone using qlib and aelib?
i tend to see the former installed on quite a few people's set ups.

what are they used for primarily? how much legacy-bloat is in 'em? are they worth having around for a hobbyist or do people have em for pipeline dependency reasons?

Anonymous No. 886956

>>886861
>Despite being mostly a CPU renderer
not anymore
You should get the early access version of Clarisse 5.5 which has the new renderer called Angie which is running on GPU or CPU (or both).
I already tested it out rendering hundred billion polygon cities.
Angie is a monster, there is no renderer as fast as Angie when it comes to render absolute massive amounts of Polygons.

Anonymous No. 886961

>>884256
> Is there a way to set it up so the mouse wheel zooms on the object instead of the cursor?
Zoom is something else. Check the dolly instead.

Anonymous No. 887082

>>874422
How do you USD export groups in Houdini? I want to have different shading groups for some primitives.

Anonymous No. 887091

>trying to learn how a kinefx set up works by manually rebuilding it
>rig vop node throws up error
>double check
>error
>go to reference hip and try to replicate vop there
>error
>create an empty vop and literally just copy paste the nodes from the reference vop into the new vop
>error
>comb through top-level settings on vop to see if i've missed something
>literally just copy and paste the refrence vop
>error
>start randomly toggling nodes on and off
>everything starts working

kinefx is so jank sometimes

Anonymous No. 887092

>>887091
it's this set up if anyone's curious:
https://cmolfino.gumroad.com/l/zZvAo

the animate along path rig vop in particular

1184x924

537236709084.webm

Anonymous No. 887094

>>887091
it's this set up if anyone's curious:
https://cmolfino.gumroad.com/l/zZvAo

the animate along path rig vop in particular

>copy paste error'd vop
>suddenly works
>webum related

Anonymous No. 887133

>>887082
Use primitive groups to select them, prefix the name with something like s_, and in Import Data check Subset Groups and put s_* there. This works for packed and nested packed primitives.

1855x557

file.png

Anonymous No. 887149

so i've managed to extend this set up a little into allowing for an arbitrary number of legs, which is fun. unfortunately this also means terrible performance because the for each loop.

the terrible performance is also totally inconsistent? if i hit play it'll play a couple of seconds fine and then run at 2 fps, but if i scrub the time line manually it's totally fine?

anyone know a better way to procedurally set up ik chains so it multithreads properly?

397x400

1639223782065.gif

Anonymous No. 887351

Hey guys, I'm getting trough the basics of Houdini and I'm enjoying it so far but I got into it for a very specific reason.
I'd like to use it to create a smal procedural urban map wich should contain the following:
- Terrain
- Trees
- Roads & intersections
- Pavements and some points scattered that will serve as placeholders for buildings/houses/streets/interest points (by streets I mean appartment buildings that are next to each others)

My questions is: Is this a realistic goal for a solo modeler? I'm not scared of VEX or Houdini as a whole, I got the basics even tough the learning curve is steep but it doesn't seem impossible for me to learn it well.
But since it's my only goal (for the moment) I lack the experience to realised the amount of work needed to do what I want to do.
Can anyone help make this more clear?

Anonymous No. 887354

>>887351
Stick to a square grid with no verticality if possible. Going for realism in procedural urban environments is a bottomless pit of corner cases, real city planners are paid big bucks and yet many of their decisions in real life seem ad hoc if not plain retarded. (Look at highways in any major city, it's like someone spilled a bowl of spaghetti on the map.)

Anonymous No. 887357

>>887354
Yeah I realise that but it's not like I'm going to go for an overcomplicated environment nor HDA's.
I'm planning on keeping the scope of the map as small as possible and same goes for the details of each buildings/houses/roads (and thus their intersections).
I'm not going to make that Spiderman's Manhattan by myself, it's just not possible. Neither I'm going to go for an architecturally accurate city.

Anonymous No. 887361

>>887351
>My questions is: Is this a realistic goal for a solo modeler?
Yeah, for sure. My recommendation is to get as familiar with SOPs as you can in the broadest sense possible, not just looking up tutorials on procedural cities or scattering trees or whatnot. There is going to be all kinds of problem solving along the way, so it's best to understand the tools available to you so you can build efficient networks and avoid getting stuck.

Anonymous No. 887366

>>887351
I don't regret studying Houdini for the past two months, but if I had to do it again, I'd learn proper layouting first with Clarisse. Houdini's built in renders are very slow, Solaris is a buggy piece of crap, V-Ray for Houdini is so bad it should not even be beta. Right now, I am having a hell of a time trying to export the scene I was working into Clarisse. I am actually going to give up on it and move on to working what I wanted instead.

If you are like me and just dive into Houdini, at some point you will get satisfied with the geometry and want to actually make it look good, and then get wrecked because Houdini's lookdev capability is crap. Even if Solaris was not garbage, it still would not have the capability to display the sheer amount of polygons that Clarisse can which you'd need for large scale environments.

As an aside, I haven't looked into urban environments, but Houdini has nodes for importing OSM data and turning it into geometry.

https://www.sidefx.com/learn/world-building/

It is one of the tutorials here. I'll definitely look into it when I start making cities myself.

Anonymous No. 887367

>>887366
>Houdini's lookdev capability is crap
if you're the one who posted those v-ray renders above, i'd stop blaming the tools, m8. lol.

Anonymous No. 887377

>>887367
As long as I put in the effort, nothing good would come from blaming myself. In this situation, it much better to blame the tools. I am sure you'd agree with me if you had any experience with them.

Tools are the one true way to power. If you put in effort you'll reach your peak in a particular domain eventually. Since humans cannot upgrade their brains, tools are the one aspect that can be controlled to exceed one's limits. The surest way to get better at anything is to use a better tool.

Anonymous No. 887426

>>887377
lmao, alright dude.

Anonymous No. 887442

>>887361
Thanks for the tips.
> not just looking up tutorials
So what do you mean by that? Should I read the documentation or just train with as many SOPs nodes as possible and try to mess with them ?
>>887366
I mean I don't really care about Houdini's built-in renders and lookdev, my goal is to model for video games so as long as I end up with a coherent mesh created from and HDA, I'll be more than happy with that. After baking these mesh, I'll take care of the rest in Unreal.
> Houdini has nodes for importing OSM data and turning it into geometry
Yeah I saw that. I still didn't watch any videos on this but I'll definitely will. But I'm wondering if I can create (and if there's a guide for that) my own OSM data to layout the terrain, it would be cool.

Anonymous No. 887792

Who the fuck sneaked in the 4chan logo inside the MILL LDN Houdini as chan import?
Once I tried to sneak in Taiwan and Hong Kong flags in a Chinese advertisement and got caught.

Anonymous No. 887846

>>887792
on the hyena video?

Anonymous No. 887903

>>887792
at what time?

485x409

1449150664715.png

Anonymous No. 887979

What's the best VEX course out there? No need to give me the easiest/noob friendliest course/resources, I want to straight up get familiar with the most important functions and nodes. I'm a programmer so I'm not afraid (Particurlaly interest in SoPs and VFX)

Anonymous No. 887983

>>887979
>joy of vex
https://www.tokeru.com/cgwiki/index.php?title=JoyOfVex

>indie pixel (this one moves relatively quick and assumes a little programming familiarity)
https://www.youtube.com/watch?v=8xxBxUgjD2c&list=PL5V9qxkY_RnKsmnem8A7opUt6OYjk4FCU

>exhaustive coverage of the fundamentals
>can move very slowly
https://www.youtube.com/playlist?list=PLzRzqTjuGIDhiXsP0hN3qBxAZ6lkVfGDI

>lots of very cool stuff in here but a little haphazard in terms of difficulty spikes. maybe not for true beginners
https://www.youtube.com/playlist?list=PL2ErPsQ-O7_cOC87ja4ATTinrIJz9zGJE

>older but still good
https://www.bilibili.com/video/BV1Hz4y1U7fq?p=1

>very old but has good stuff. don't do this before you're familiar with h19, because some things have changed and you may get lost
https://www.bilibili.com/video/BV1V441167kg?p=2


but like most programming material, no course is going to teach you everything or give you a sense of the stuff you'll be using 90% of the time.

vex is everywhere in houdini so most houdini courses out there will have some vex in them and you'll pick up techniques here and there. some places are more vex heavy than others e.g. entagma, junichiro's algorithmic live, some john kunz' archived live streams etc.

applied houdini and rebelway course will have very pragmatic approaches. applied houdini courses are very good general intros to houdini as well. rebelway ones are a little dirtier but more 'realistic' for production.

512x384

1605886656292.jpg

Anonymous No. 887985

>>887983
Thanks a lot anon. I wish you the best in your life
> but like most programming material, no course is going to teach you everything or give you a sense of the stuff you'll be using 90% of the time
Yeah don't worry about that, I know. I'm just after a general sense of the syntax, general rules and practical examples of using VEX for modeling.

Anonymous No. 887987

>>887985
you'll probably enjoy the stuff on indie pixel's channel.
if you want more abstract neat stuff definitely look at juni's algo live series.

if you're familiar with a c-style language already you'll feel right at home. might take a second for you to adjust to dealing with attributes vs variables and some of the quirks of the language.

best of luck.

Anonymous No. 887989

>>887987
You the best. I wish /3/ was filled with people like you.
> juni's algo live series
You know what, I already watched some of his stuff and even bought his two books on Gumroad. But one thing I don't get from his videos are the usability of what he shows. It's so abstract that I don't see how you would use most of his stuff on a project. But then again, maybe I'm not experienced enough to get it.
His books are cool as fuck tough.

Anonymous No. 887990

>>887989
yeah that abstract stuff is mostly useful in design-focused things like mograph/advertising etc. i think his background is in parametric architecture. but there is always value in seeing how someone approaches different problems etc. but pick n choose, depending on what you need to do.

if you want environment modelling stuff have a look at adrien lambert's channel and anastasia opara's lake houses series (it's on billi billi)

Anonymous No. 888154

I feel like I am too stupid to use Houdini. Thoughts?

Anonymous No. 888155

>>888154
You don't need a high IQ to use Houdini. You just need a PC worth $4k-$5k at minimum.

Anonymous No. 888161

>>888155
that's only true because of GPU prices, and even so it's exaggerated. you can work with a less than ideal system, provided you have sufficient RAM/VRAM/storage; it's just going to be slower.

Anonymous No. 888163

>>888161
You need a 12900k, 20+ tb of hdd, multiple NVME ssd, 128gb ddr, gtx 3090, monitors that can display true hdr 1000+, new case, 1200w psu, thousands in software licenses for complimentary software and on top of that you will want to build a second identical machine so you can take advantage of the fact that houdini gives you up to 3 licenses for $270 a pop. We're talking a big initial investment.

Anonymous No. 888211

>>888163
I use Houdini on a 10 year old quad core with 16 gigs of ram and a 3 year old 1080. My rig was a mid-high end consumer pc in 2013, spent ~2k on it at the time and I bought all peripherals, too, which I am still using at this very moment.
None of what you're going to be realistically doing as one person is ever going to actually make use of all that computing power at once. If you do, you're most likely retarded.

You just fucking suck.

Anonymous No. 888223

>>888211
>I use Houdini on a 10 year old quad core with 16 gigs of ram and a 3 year old 1080
What's it like to be so poor you are using a 1080 in 2022 and additionally so ignorant that someone else can actually do this and pay the bills enough to afford modern workstations. You know what? Don't answer that - I dont want you to over exert yourself, old man.

Anonymous No. 888255

>>888223
I unironically have a better job and a better career than you ever will.
The workstation I use shits on the babby specs you posted and costs about ~$50k, but the shit I do there far exceed the scope of whatever I can do on my own in my spare time.
My 10 year old pc, despite showing its age, still chugs happily along for now.

Maybe I have a job because unlike you I can work around technical limitations and make due with what I have instead of complaining about not having the latest, most expensive toys like a little bitch.

Anonymous No. 888257

>>888255
>I use Houdini on a 10 year old quad core with 16 gigs of ram and a 3 year old 1080.
>The workstation I use shits on the babby specs you posted and costs about ~$50k

you are fucking senile

Anonymous No. 888262

>>888257
no, you're just retarded.
here, let me spell it out to please your autism:
>my "work" workstation->(the workstation I use at work)->(the machine a multi-million-dollar company bought and pays me to use)
blows your babby shit out of the water

>my 10 year-old pc->(the one I used to get the job in the first place) ->(the one *I* bought)
is rapidly becoming a piece of shit, yet it still got me through the door

get it now?
do I need to draw pictures?

Anonymous No. 888263

>>888262
i work for a multi million dollar company as well. Whats your point? Let me guess, you think you have high iq or something?

Anonymous No. 888265

>>888263
no you don't, cause you bitch too much. you wouldn't last a week.

Anonymous No. 888267

>>888265
whatever you say, boss man

Anonymous No. 888335

nice new write up on cgwiki to look through:
https://www.tokeru.com/cgwiki/index.php?title=PopsSplitAvoid

Anonymous No. 888343

>>874422
what's the best renderer to use with Houdini?

Arnold is a stuttering mess, can't use it for lookdev, everytime I move something in the viewport with the IPR running it gives me a spastic response, fucking shit.

why is redshift and octane not pirated? fucking shit.

Anonymous No. 888345

>>888343
>begging
>choosing
>complaining about every little thing
suck it up, bitchboy

Anonymous No. 888387

>>888343
>why is redshift and octane not pirated? fucking shit.
incidentally, now that Maxon has blocked all of its Russian customers, I'd expect a crack for Redshift somewhat soon

Anonymous No. 888391

>>888343
If you don't mind exporting from Houdini, Blender is actually a good choice for lookdev. While Cycles is not the fastest render engine, it is pretty fast when used with the GPU and more importantly is very stable and won't crash on you all the time like V-Ray for example.

It can't import instances via USD unfortunately.

If need that or if your scenes have a large amount of geometry, I'd just go with Clarisse. Though Clarisse crashes frequently, at least it is fast to restart unlike Houdini. Going this route won't lock you into old versions of Houdini.

>>888387
This is good news for me. I've never tried Redshift, but V-Ray is so bad I can't imagine its competitors being worse.

I am guessing V-Ray was made for 3ds Max and works well there, but the Houdini version got passed on to the janitorial team. It is really a pity as it has a lot useful nodes, and if it were not so buggy it would be worth its price.

Anonymous No. 888402

is there a cleaner way of getting all the hedges in an n-sided polygon than this? i'm a bit retarded.

primwrangle:

int hedge_array[] = {};
int start_hedge =primhedge(0, @primnum);
append(hedge_array, start_hedge);

int nx_hedge = hedge_next(0, start_hedge);
while(nx_hedge != start_hedge){
append(hedge_array, nx_hedge);
nx_hedge = hedge_next(0, nx_hedge);
}

Anonymous No. 888417

>>888391
vray works really well with max, rarely have any issues with it, but I don't do anything really elaborate

Anonymous No. 888444

>>888402
Maybe primvertices() followed by a foreach loop where it check if vertexhedge() is valid and appends the hedge if so?

Anonymous No. 888476

>>888444
Clever. I was a little confused by criteria for validity, because the Houdini docs don't specify:
>any valid halfedge mesh must be manifold and orientable
Saw this elsewhere and not entirely sure what orientable means.

Anonymous No. 888658

Any way to visualize the actual mask number values for a Highfield mask?

Anonymous No. 888723

>>875632
Any tips?

Anonymous No. 888764

Can karma compete with other render engines? Like redshift or renderman?

Anonymous No. 888932

>>888764
No. Its a toy renderer.

Anonymous No. 888964

>>888764
Karma has gotten better. It still has some maturing to do, but it's getting there. Redshift can't be beaten for speed but I would much rather use Karma than Renderman.

Anonymous No. 888969

>>888932
>>888964
Sounds grim.
From what I've read, houdini's vray is full of bugs.
So there's only redshift?

Anonymous No. 888982

>>888969
people complaining about lack of stable rendering options in houdini are clowns.

Anonymous No. 888985

>>888982
Could you elaborate or tell me about your experiences with it?

Anonymous No. 888995

>>888985
the plugin ones are about as stable as they are in other dccs.
v-ray just came out of beta, but it was fine when i tested it out briefly.
mantra is fine but slow. karma xpu is very promising, but early. it'll be great in a year.

Anonymous No. 889087

>>888995
Thanks a bunch. I'll have to see v-ray for myself then.

Anonymous No. 889140

>>889087
If you are going to go down this route, don't do what I did which was messing with shader nodes, and instead do all your texturing using something like Substance Painter, and plug in the results of that into V-Ray. If you use that workflow you should be able to get around most of V-Ray's bugginess. I've only started learning Pt now and it was not obvious that this was the path I should follow a month ago.

Anonymous No. 889146

>>889140
That's what I usually do. I'm not a shader wizard.

Anonymous No. 889307

>>888995
>mantra is fine but slow.
anon, mantra is one of the worst renderers imaginable, by all measurements

Anonymous No. 889326

>>889307
works on my machine :^)

Anonymous No. 889327

>>889326
nobody said it didnt load, esl

Anonymous No. 889342

>>889327
bofa deez nuts

Anonymous No. 889453

>>889307
>mantra is one of the worst renderers imaginable, by all measurements
that's news to me, would you be so kind to elaborate on it, I thought its only slow and that was its main criticism, but I am willing to learn some more.
I am only using Houdini for simulation stuff, so I usually bring the result to Maya (or Blender even) and render it there....

Anonymous No. 889472

>>889453
it's basement dweller opinion. Most game cinematic trailers are done in Mantra. It's very widely used in production.

Anonymous No. 889473

>>889472
>Most game cinematic trailers are done in Mantra
anon...

Anonymous No. 889476

>Last Labs update is from December
Is SideFX Labs kill? The lead dev is now freelancing, what's the future of Labs tools?

Anonymous No. 889478

>>889476
They just hired two new people for Labs and also published a Labs Roadmap site. Phone posting so no links at the moment, sorry.

Anonymous No. 889492

>>889476
>>889478
https://twitter.com/sidefxlabs/status/1507452885671497729?s=21&t=_m0eVepxKyFiDwuqxjnDXQ

https://twitter.com/sidefxlabs/status/1505886055538507780?s=20&t=W7PtogPDLsmseIMp4gPTyA

Anonymous No. 889722

https://space.bilibili.com/9239877

Has some entagma kinefx videos that aren't on cgp yet.
I think the timj stuff isn't on cgp either but not sure.

Anonymous No. 889723

>>889492
> building generator
Are they going to let your generate whole cities from the labs tool or what? I guess they're going to release every tool we saw on the project Titan video

Anonymous No. 889779

>>889722
Oi! Thanks a bunch, mate!

Anonymous No. 889808

New to houdini but would anyone be able to recommend whether it's necessary to learn python/vex to use houdini properly? I'd like to do so anyway but I just want to make sure I'm not wasting time doing so.

Sorry if there's a post already

Anonymous No. 889809

>>889808
Python is not mandatory but VEX will help you get the true potential of Houdini. It's not mandatory either but you miss a lot of cool stuff if you do only node stuff

Anonymous No. 889810

>>889809
This is correct. But it's worth noting that you don't need to be fluent in VEX right off the bat. You can start off utilizing little snippets of code here and there, and gradually beef up your VEX knowledge as you learn the rest of program.

Anonymous No. 889812

>>889809
>>889810


That makes a lot of sense, appreciate the answers

I suppose I'll just pick up bits and pieces as I go

768x733

1615946471567.png

Anonymous No. 889921

Do you know if this video is the only SIGGRAPH video on video games and Houdini or if there are more on the subject?
https://www.youtube.com/watch?v=7X-r_XarMLI

Anonymous No. 889940

>>889921
i think there's a mortal kombat one. why siggraph specifically?

Anonymous No. 889942

>>889940
Because they have some really interesting stuff on other subjects but they way they organized their Youtube channels/titles/playlists is stupid. You have hours long videos but you don't know what they're about.

Anonymous No. 890101

so i don't have a programming background and have run into a problem when iterating over arrays in a wrangle.

i think i know how to work around this now, but just want some confirmation.

if i foreach over an array, i'm guessing that modifying the array (removing elements usually) i'm iterating over is the thing that lands me in trouble?

is the solution as simple as working on a separate copy of the array, because that's how i've worked around it.

Anonymous No. 890120

>>890101
well yeah, I mean are you talking about removal to an end case i.e. iterating until parts of an array are removed until there are none left?

Or just manipulating parts of an array that are already being operated on in an existing foreach?

if it's the latter, it will cause problems and I'd say you want a separate copy.

If it's the former I would suggest checking if you have properly set a base case for your iteration (so that it will stop trying to remove shit once there's nothing left)

I don't actually know houdini that well yet but I've learned a few programming languages, hopefully this makes sense

531x471

1413517975983.jpg

Anonymous No. 890122

>>890101
>>890120
> reddit spacing
Please stop. Not even bc of Reddit but bc it hurts the eyes

Anonymous No. 890137

>>890120
the last time i ran into this problem i was trying to make unique pairs from an array of points.

>make copy of pt array (copy array)
>foreach on pt array
>make sure iteration pt exists in copy array
>pick random point to pair with from copy array
>*do thing*
>remove iteration point and its pair from the copy array so they aren't paired with anything again

it's usually cases like that, where the array needs to be modified before the next iteration.

>>890122
sorry bb grill

Anonymous No. 890144

who got the ODTools and MOPS+?

Anonymous No. 890149

>>890137
Yeah, if you use foreach() the number of elements in an array shouldn't change, if you must add or remove things from the array during iteration use while() instead, look into how recursion is implemented for a practical example.

Your solution to the unique point pairs problem seems overcomplicated, did you consider just randomizing the point order and pairing every even number point with the next odd number point? No need to check if some value exists in an array, this is very expensive.

Anonymous No. 890150

>>890149
>did you consider just randomizing the point order and pairing every even number point with the next odd number point?
thank you for making me feel like an idiot.

Anonymous No. 890342

https://bilibili.com/video/BV1DS4y117WC

R'way - advanced water. Not on cgp yet I think

Anonymous No. 890354

>>889722
much obliged
how did you find this?
search engine on bilibili is a mess

Anonymous No. 890355

>>890342
thanks

how to get a baidu account?

Anonymous No. 890356

why is karma xpu so bad?

Anonymous No. 890393

>>890354
Just showed up in the related videos.

>>890355
No idea.

Anonymous No. 890611

R'way - advanced veefx rise

https://www.bilibili.com/video/BV1Va411i7BP
https://www.bilibili.com/video/BV1PS4y1P73y

you begin to wonder what the point of cgp is when they keep nuking torrents for rule infringements and no one bother to fix them

Anonymous No. 890811

UE5 matrix city generation talk
https://youtu.be/usJrcwN6T4I

All the hip files are also available for download along with the rest of the assets

Anonymous No. 890834

>>890356
It's better than mantra

Anonymous No. 890843

>>890834
Because it supports gpu and is faster?

Anonymous No. 890844

>>890843
Well yeah

>am poor bad cpufag

Anonymous No. 890886

Anyone know how to export multiple KineFX animations in a gLTF or FBX at a time?