Houdini 20 Rumors

   200445   515   23
User Avatar
Member
85 posts
Joined: Oct. 2018
Offline
Jackatack
hMono
Jackatack
I'm starting to feel that a lot of the challenges Hmonkey is facing with Houdini might just be part of the learning curve. Let me break it down. It's common to not quite get the hang of something and then point fingers at Houdini's viewport.

Hey thanks for breaking it down for me! I guess I was wasting time in houdini for the past couple of years

Two years using Houdini(wow), and in all your posts, you criticize the software you're learning just because it's not like Cinema 4D or doesn't operate like other software... Maybe it's a matter of approaching the learning process differently

I know what you mean, and you missed the point of those posts.

Meanwhile this might help you - https://www.wikihow.com/Use-an-Abacus [www.wikihow.com]

Best regards
User Avatar
Member
64 posts
Joined: April 2022
Offline
Jackatack
hMono
Jackatack
I'm starting to feel that a lot of the challenges Hmonkey is facing with Houdini might just be part of the learning curve. Let me break it down. It's common to not quite get the hang of something and then point fingers at Houdini's viewport.

Hey thanks for breaking it down for me! I guess I was wasting time in houdini for the past couple of years

Two years using Houdini(wow), and in all your posts, you criticize the software you're learning just because it's not like Cinema 4D or doesn't operate like other software... Maybe it's a matter of approaching the learning process differently

I've been intensely learning Houdini for about 3 to 4 years now, and I still feel like a noob. However, if something doesn't work out as expected, I don't give up. I don't blame my limitations on Houdini not being like Cinema 4D or some other software. Nor do I take the easy way out by saying, 'It's not me, it's the software that's faulty.'

About 90% of the time when something doesn't go well, I go back to the beginning of the setup to figure out what's not working right. I search for information and ask questions. Eventually, I manage to solve the problem, and my learning keeps progressing.

I maintain a humble profile while learning Houdini, which I believe is the right way to learn—from a standpoint of 'I might be doing something wrong; let's figure out why it's not working.'

Perhaps over these two years, you've set aside too many things as 'Houdini problems' that could have contributed to your learning. Maybe you think that with 2 years of learning, you can hold authority and claim that your setups are flawless and everything is Houdini's fault. Personally, I prefer to first look for the problem within myself, with the help of experienced experts who have been at this for more than 10 years.

Best regards
3 to 4 years of intensive learning of Houdini didn't make you any smarter. Talk about the obsolescence of Maya tools in studios where it is intensively used specifically for modeling, not just animation. Stop talking nonsense. You're already being discussed on other forum. And this is no joke.
alexeyvanzhula.gumroad.com
User Avatar
Member
64 posts
Joined: April 2022
Offline
About edge loops:
alexeyvanzhula.gumroad.com
User Avatar
Member
403 posts
Joined: March 2014
Offline
... what would be really helpful for modeling in Houdini is a RMB context sensitive dropdown menu.

The RMB can speed up workflow a lot, not only for modeling. But it seems completely ignored in Houdini, it gives the same options no matter what is currently selected.
User Avatar
Member
328 posts
Joined: April 2018
Offline
hMonkey
Keep waiting
I mean, you said yourself that you'd do it, so I was just expecting you to keep your word. It's fine that you chose to not do that. Keep those reports going - that's definitely more useful to everyone.
User Avatar
Member
64 posts
Joined: April 2022
Offline
alexeyvanzhula.gumroad.com
User Avatar
Member
64 posts
Joined: April 2022
Offline
I post these demonstrations to show that everything could have been better

Attachments:
Untitled.png (1.6 MB)

alexeyvanzhula.gumroad.com
User Avatar
Member
85 posts
Joined: Oct. 2018
Offline
eikonoklastes
hMonkey
Keep waiting
I mean, you said yourself that you'd do it, so I was just expecting you to keep your word. It's fine that you chose to not do that. Keep those reports going - that's definitely more useful to everyone.


quote=hMonkey] It might have been that, but subsequent click should've grabbed the foreground geo and readjusted the camera pivot based on cursor which it didn’t, it got locked to bg plane and it resulted in an awkward navigation, which you need to manually fix with f/h/space+z,etc…

I guess you missed it, but lets just agree that you are right and drop it, since the whole thing is platform/observer relative anyway…
User Avatar
Member
405 posts
Joined: Aug. 2019
Offline
alexeyvanzhula1984
About edge loops:

That reminds me of another issue of Houdini.

While Houdini is praised for its proceduralism, some important features are buried inside some mega destructive nodes. How do you slide an edge loop? Uh... Edit SOP can do this, cool. But Edit SOP can also do another 100 things. Once you use Edit for simple tasks like sliding an edge loop, you can no longer check the node network and see which step is which at the first glance.

A similar case: how do you relax out your topology without changing the mesh's shape? Uh, TopoBuild SOP can do that, cool. But (repeat the previous paragraph)
User Avatar
Member
328 posts
Joined: April 2018
Offline
raincole
While Houdini is praised for its proceduralism, some important features are buried inside some mega destructive nodes. How do you slide an edge loop? Uh... Edit SOP can do this, cool. But Edit SOP can also do another 100 things. Once you use Edit for simple tasks like sliding an edge loop, you can no longer check the node network and see which step is which at the first glance.
A procedural slide edge tool does not exist at all in Houdini though, right? The sliding in Edit is destructive (by design).

A Slide Edge SOP would be cool.
User Avatar
Member
405 posts
Joined: Aug. 2019
Offline
eikonoklastes
raincole
While Houdini is praised for its proceduralism, some important features are buried inside some mega destructive nodes. How do you slide an edge loop? Uh... Edit SOP can do this, cool. But Edit SOP can also do another 100 things. Once you use Edit for simple tasks like sliding an edge loop, you can no longer check the node network and see which step is which at the first glance.
A procedural slide edge tool does not exist at all in Houdini though, right? The sliding in Edit is destructive (by design).

A Slide Edge SOP would be cool.

That's exactly what my point was. When I complained about how Houdini's modeling tool is not as good as Blender, some peope might think that I'm trying to push an agenda of having a "Modeling SOP" that does everything from Edit to Sculpt to TopoBuild.

That's the opposite of what I want!

I don't want fewer nodes. I want even more nodes, each of them does only one thing. I even think PolySplit should be split (pun intended) into two nodes. I just hope that SideFX makes sure these nodes work well interactively so I can do most modeling tasks without leaving Viewport. And perhaps a shelf tool for Network view that automatically puts the recent "similar" nodes into a subnet and give it a meaningful name (like "PolySplits").
Edited by raincole - Aug. 11, 2023 10:31:00
User Avatar
Member
4409 posts
Joined: Feb. 2012
Offline
eikonoklastes
raincole
While Houdini is praised for its proceduralism, some important features are buried inside some mega destructive nodes. How do you slide an edge loop? Uh... Edit SOP can do this, cool. But Edit SOP can also do another 100 things. Once you use Edit for simple tasks like sliding an edge loop, you can no longer check the node network and see which step is which at the first glance.
A procedural slide edge tool does not exist at all in Houdini though, right? The sliding in Edit is destructive (by design).

A Slide Edge SOP would be cool.

I wrote one for H12 using Python. The code is quite outdated now, but it still works:



node = hou.pwd ( )
geo = node.geometry ( )

def filterEdgeGroup ( edgeGroup ):
    group = []
    names = edgeGroup.split ( ' ' )
    for name in names:
        try:
            edges = geo.globEdges ( name )
        except hou.OperationFailed:
            edges = ( )
        
        for edge in edges:
            p0 = edge.points ( ) [ 0 ]
            p1 = edge.points ( ) [ 1 ]
            
            try:
                isValidEdge = geo.findEdge ( p0, p1 )
            except:
                isValidEdge = False
            
            if not isValidEdge:
                continue
            
            if len ( edge.prims ( ) ):
                group.append ( edge )
    
    return group


def getPolysFromEdges ( edges ):
    polys = []
    for edge in edges:
        polys += edge.prims ( )
    
    return polys


def getPointsFromEdges ( edges ):
    points = set ( )
    for edge in edges:
        points.add ( edge.points ( ) [ 0 ] )
        points.add ( edge.points ( ) [ 1 ] )
    
    return points


def getPointsFromPolys ( polys ):
    points = set ( )
    for poly in polys:
        verts = poly.vertices ( )
        for vert in verts:
            points.add ( vert.point ( ) )
    
    return points


def isValidEdge ( point1, point2 ):
    try:
        group = geo.globEdges ( "p{0}-{1} ".format ( point1, point2 ) )
    except hou.OperationFailed:
        return False
    
    for edge in group:
        p0 = edge.points ( ) [ 0 ]
        p1 = edge.points ( ) [ 1 ]
        
        try:
            return geo.findEdge ( p0, p1 ) != None
        except:
            return False


def getEdgePointsOfPoint ( point ):
    points = hou.hscriptExpression ( "pointneighbours(\"" + node.inputs ( ) [ 0 ].path ( ) + "\", " + str ( point ) + ", 1)" )
    points = points.split ( ' ' )
    points = map ( int, points )
    edgePoints = set ( )
    for pt in points:
        if isValidEdge ( point, pt ):
            edgePoints.add ( geo.iterPoints ( ) [ pt ] )
    
    return edgePoints



group = hou.evalParm ( "group" )
amount = hou.evalParm ( "amount" )
edges = filterEdgeGroup ( group )

if not edges:
    raise hou.NodeWarning ( "No valid edges were found in the group." )

polys = getPolysFromEdges ( edges )
points = getPointsFromEdges ( edges )
outerPoints = getPointsFromPolys ( polys )
outerPoints -= points

if not outerPoints:
    raise hou.NodeWarning ( "No valid edges were found in the group." )

first = outerPoints.pop ( )
source = { first }
innerPoints = { first }

while source:
    edgePoints = getEdgePointsOfPoint ( source.pop ( ).number ( ) )
    shared = edgePoints.intersection ( outerPoints )
    source = source.union ( shared )
    innerPoints = innerPoints.union ( shared )
    outerPoints = outerPoints.difference ( edgePoints )

goals = outerPoints
if ( amount < 0 ):
    goals = innerPoints

for point in points:
    edgePoints = getEdgePointsOfPoint ( point.number ( ) )
    target = edgePoints.intersection ( goals )
    if target:
        p0 = target.pop ( ).position ( )
        p1 = point.position ( )
        point.setPosition ( p1 + ( p0 - p1 ) * abs ( amount ) )
Senior FX TD @ Industrial Light & Magic
Get to the NEXT level in Houdini & VEX with Pragmatic VEX! [www.pragmatic-vfx.com]

youtube.com/@pragmaticvfx | patreon.com/animatrix | animatrix2k7.gumroad.com
User Avatar
Member
232 posts
Joined: March 2009
Offline
alexeyvanzhula1984

Honestly this is an amazing tool (Modeler). Can’t wait for the next version of this tool and to see it work with Houdini’s next version modernized viewport. I kind of wonder why doesn’t SideFX reach out to the developer and either hire them or purchase this from them. It would solve a lot of problems and complains regarding modeling :-)

If someone could do this for parametric curve modeling and modernization of that toolset that would be amazing. Until then I think we can all agree that Houdinis way of modeling is not for everyone and that’s ok. Everyone works and thinks differently. That’s why there are so many tools out there. We also agree that UX needs work and that Viewport needs to be revitalized. I think SideFX recognizes that and Viewport is being redone. Hopefully in H20. Balancing complexity with flexibility, artists with technical director needs is not easy.

Another thing I think we would mostly agree on is that today as it stands Houdini is not in all areas an artist friendly tool. In many areas it feels like it’s written for technical directors or programmers. It starts with inconsistent naming things (e.g. edges, poly lines, paths), naming of nodes, naming of parameters, nodes that try to do too much, and most importantly - imho - horrible documentation, especially for the older areas of the software. I’m almost inclined to start a thread on documentation hall of shame lol. But that’s a different topic.

The good news is that SideFX development speed is nothing short of amazing and I’m sure they recognize the opportunities. The bad news is like with every software they have to pick their battles and can’t do everything at once.

If I look at the stuff SideFX has done last 3 years it’s nothing short of amazing. PDGs, KineFX, muscles, DOPs in SOPs, Karma CPU, Karma XPU, MaterialX, Solaris. Just go over Houdini 19 and 19.5 feature list and name one 3d software that did 25% of that in 2 years.

Then you have to look at return of investment. Sure, maybe (let’s assume it’s true) Houdini could use better box modeling tools. Would you want the developers to invest their time into that or focus on RT Viewport or core architecture? Being a product owner myself and running a large team of product owners, those are not easy decisions. Everyone always wants everything at once. Every software has opportunities for improvements. You need to balance things.

Cheers.
Edited by LukeP - Aug. 11, 2023 15:16:35
User Avatar
Member
629 posts
Joined: June 2006
Online
LukeP
I think SideFX recognized that and Viewport is being redone. Hopefully in H20.

There will be a new Viewport in H20 based on Vulkan, this task has been started 2-3 years ago(guess). Othere DCC are also on the shift to Vulkan.
Modo has showen a first glimps (very basic), blender should bring end of the year a alpha/beta version.
User Avatar
Member
76 posts
Joined: Aug. 2020
Offline
eikonoklastes
citizen
an 'edge divide' tool that works decently
The Edge Loop command works decently, no? Or are you referring to something else?
It works for its purposes, but I'm referring to 'edge divide' because it's the only one that currently can divide the edges of a partial edge-ring selection. And it's buggy AF. And neither have a "preserve curvature" like this:
Edited by citizen - Aug. 11, 2023 15:12:35

Attachments:
edge_divide.jpg (174.3 KB)

User Avatar
Member
9 posts
Joined: Jan. 2021
Online
I do think that some of the opinions on this thread might have been different if more of the Houdini user base had been exposed to Alexey's plugin. I saw him ask for opinions on his discord server recently about why there aren't more responses to the forum threads he creates, and I am also confused, though maybe it simply hasn't caught on yet. I don't do any type of traditional modeling but I still own a copy of the plug-in and recognize its benefits.

Now, this discussion of destructive versus not in regards to modeling reminds me of the very interesting "recache strokes" button that is/has been found on a few nodes. It is a fantastic, wonderful button. Obviously its results are probably something more akin to an estimation or make use of something like the attribute transfer or fill nodes. But the real intrigue, the mystique of the button comes from the notion of capturing destructive changes as a parametric set of instructions. That could be a really cool technique to look into. I might play around with the idea myself.
User Avatar
Member
89 posts
Joined: Jan. 2015
Offline
Maybe despite the noise we do here, not so many current users are interested in direct modeling in Houdini, so the audience is kinda restricted from the start. Pragmatic people are using dedicated solution for that task already.
Gameloft
User Avatar
Member
230 posts
Joined: March 2013
Offline
I don't know why you guys are so hot for modeling. It's like, anyone who models day in day out as a job, is already using
an app that is clearly years ahead of houdini, and only popping into houdini to run certain modeling aspects that make sense.

Procedural models sure, it can be quite helpful for that too. But guys, houdini's core usage is not Poly modeling.
As far as destructive modeling goes, maybe you never worked in a studio, and weren't taught to simply version up your
scene file as you do modeling stages that may not end up how you'd like?

Pushing the modeling aspect over several pages, and the back and forth about it is a total waste of time. The tools will get
better, the viewport will get better, but about 98% of houdini's user base is not concerned with modeling, I get you are hot
for it, but you're in the 2%, so temper your expectations.
Personally, I would never poly model in it, it just doesn't feel right, Max feels miles better, ditto Maya, ditto Blender to an
extent. And in the real world, assets is one of the only disciplines that is almost software agnostic. Every other part of the
pipeline is far more locked to a DCC, assets is not, so bashing on about houdini becoming a top shelf poly modeler is just
not reading the room at all.


L
Edited by lewis_T - Aug. 11, 2023 23:19:23
I'm not lying, I'm writing fiction with my mouth.
User Avatar
Member
37 posts
Joined: Aug. 2021
Offline
Any news about the siggraph show ?
Do you know when it will be publish on sidefx’s YouTube channel ?
User Avatar
Member
405 posts
Joined: Aug. 2019
Offline
lewis_T
But guys, houdini's core usage is not Poly modeling.
L

And it's not character animation, so bye bye KineFX. And it's not for game dev, so bye bye 50% of Labs and Houdini Engine. And of course COPs should be gone too.

I think Houdini should get rid of all the GUI and provide itself as a solver. Actually not a solver, but a library that helps you write your own solver. libhoudini.so. Because obviously that's what everyone wants. /s


Blender could have shrugged their shoulders and said ZBrush is the clear winning in sculpting and sculptors are <2% of Blender's users. But they didn't. Well actually one of them didn't: there was a guy who single-handed turn Blender's sculpting tools from toys to usable tools. But I digress.

The reason we want a better poly modeling environment is not that we don't have access to Maya or Blender. It's that Houdini's poly modeling is actually like 80% good. It's just that the last 20% seems to take 80% to develop (which is common in software dev).
  • Quick Links