https://vimeo.com/286242378 [vimeo.com]
I hope we can get a bit of a discussion on this new feature going.
My immediate thoughts are:
1. Absolutely all Topobuild functions should be available as separate fully procedural nodes.
2. In Houdini I want to be able to make my own variation of the topobuild node and polydraw node from nodes and script.
Maybe this would require a new context. But imagine you could add and remove your modelling actions and bind them to context sensitive shortcuts.
- I realize this is going into the realm of utter insanity, but I have point with this (see point 3).
3. Via Houdini Engine we can get a full modelling tool into Unreal/Unity! This would be a really really really big deal.
4. The “black sheep” (edit,polybuild,polydraw) nodes are as bad as any destructive workflow in normal modelling applications (like Maya, Max, Blender, Modo etc.).
They are useful, but in a way they are currently a step backwards. Or rather Houdini is just playing catch-up instead of finding its own way.
I think the better way forward would be a better more fluent way of using fully procedural nodes. Currently, to edit multiple procedural nodes at the same time in an efficient way, one basically has to build an HDA. Now this process is time consuming and it feels wrong, because it would be insane to build dozens and dozens of complex HDAs for a smooth modelling experience. Rather than that, I propose to make it easier to select and edit multiple nodes at the same time. So you build things in a procedural way, then temporarily assign hotkeys (1 through 5 or something like that) and then you jump between the edit modes of those procedural nodes. And the edit modes of HDAs should be reworked to work more smoothly. For example they should not require the enter key for confirmation. One has to be able to edit all nodes with just one hand on keyboard and one on the mouse, without moving the hand on the keyboard a significant amount.
Well, those are my thoughts. Fight me. XD
Polydraw node discussion (based on sneak peek siggraph 2018)
5281 6 4- DASD
- Member
- 453 posts
- Joined: 2月 2013
- Offline
- Alexey Vanzhula
- Member
- 538 posts
- Joined: 12月 2006
- Offline
I wish:
1. Add ‘Conform to Surface’ option to TopoBuild SOP. It allows to move vertices in space or through the surface
2. Add TopoBuild options to control projection plane: position, orientation, shape (like simple plane, lens etc.)
3. PolyDraw tool will create TopoBuild SOP without conformation option
1. Add ‘Conform to Surface’ option to TopoBuild SOP. It allows to move vertices in space or through the surface
2. Add TopoBuild options to control projection plane: position, orientation, shape (like simple plane, lens etc.)
3. PolyDraw tool will create TopoBuild SOP without conformation option
Edited by Alexey Vanzhula - 2018年8月24日 06:39:58
- nicoM
- Member
- 74 posts
- Joined:
- Offline
From my thought to say procedural tools are foward thinking and non-procedural are backward thinking are not very useful.
Both concept have their pros and cons!
What I would find very usefull is to find some kind of bridge between the two methodologies.
Lets say I block out some basic shape of a 3d Model with the destructive toolset - some kind of enhanced version of the presented poly draw - more like Zbruh Zmodeller. And have some kind of analytics tools which rebuild my 3d sketch to a procedural network.
Or I can define some constraints in the modell to define parametrical parts - like dimension of parts, line to be perpendicular to othher etc … (something you can do in CAD Software like Fusion 360).
Here some video links to zbrush zmodeler and c4d polygon pen as idea for enhanced polydraw:
http://pixologic.com/zclassroom/lesson/poly-qmesh [pixologic.com]
Both concept have their pros and cons!
What I would find very usefull is to find some kind of bridge between the two methodologies.
Lets say I block out some basic shape of a 3d Model with the destructive toolset - some kind of enhanced version of the presented poly draw - more like Zbruh Zmodeller. And have some kind of analytics tools which rebuild my 3d sketch to a procedural network.
Or I can define some constraints in the modell to define parametrical parts - like dimension of parts, line to be perpendicular to othher etc … (something you can do in CAD Software like Fusion 360).
Here some video links to zbrush zmodeler and c4d polygon pen as idea for enhanced polydraw:
http://pixologic.com/zclassroom/lesson/poly-qmesh [pixologic.com]
Edited by nicoM - 2018年8月24日 06:49:13
- raistlinf
- Member
- 19 posts
- Joined: 11月 2014
- Offline
in my opinion POLY DRAW should be such as https://www.youtube.com/watch?v=PSD_nISLolY [www.youtube.com] or https://www.youtube.com/watch?v=wyqukJQPFU8 [www.youtube.com]
- DASD
- Member
- 453 posts
- Joined: 2月 2013
- Offline
@vux:
Agreed.
@nicoM:
I agree with the idea that there should be a button to turn the destructive process into fully procedural nodes.
The problem with the destructive workflow is simply just that. It is not procedural and therefore cannot be used safely for anything that should be procedural or iterative.
If we can turn a destructively created mesh into a procedural mesh than we can have the best of both worlds.
@raistlinf:
Agreed. The draw lines to create polys stuff can be very intuitive to use.
Agreed.
@nicoM:
I agree with the idea that there should be a button to turn the destructive process into fully procedural nodes.
The problem with the destructive workflow is simply just that. It is not procedural and therefore cannot be used safely for anything that should be procedural or iterative.
If we can turn a destructively created mesh into a procedural mesh than we can have the best of both worlds.
@raistlinf:
Agreed. The draw lines to create polys stuff can be very intuitive to use.
- hsolomon
- Member
- 15 posts
- Joined: 7月 2006
- Offline
Houdini 17 - Modeling Feedback
The boolean based Workflow by Fiana is awesome. This is Houdini's key strength and please keep improving the complex model building by boolean Ops and think of every possible proceduralism to improve this workflow. Good
Now… Thanks for the sneakpeek of Polydraw 0.1 not a bad start.
I have been modeling for 25 years started with 3Dstudio 1.0 then to Alias and then Maya… then tried Modo for a while then gone back to Maya for my Box(Non-organic) Modeling workflows. But recently I stumbled with Zmodeler and blown away by its simplicity and power. Hands down… Zmodeler is the next generation Box(Non-organic) Modeling tool. I was amazed how some one like Mr.Alexey Vanzhula's could create some thing like the Zmodeler kind of UI in Houdini.
Don't copy Maya or other DCCS in this box modeling approach but please come with a typical “SideFX -Worflow”
Retopology workflow:
Please… Please…Please stop copying what Maya or other DCCs does for Polygon strip creation where we need to seed points after points by mouse click after click say on a sharpedge/groove in order to just lay an edge flow. Its the most slowest and dumbest way in today's world when you have a wonderful curve drawing tool where we can sketch curves directly on the model and then simply weave quadstrip along those curves. Houdini already has this Curve drawing functionality BUT its not implemented in the way it can be made useful for the modeling functionality. Please can you guys dust it off and give a fresh outlook to sketching based modeling which would be a real innovation.
Here is my dream retopo tool.
Either I bring in a hirez sculpt say from Zbrush or create a complex polygon model using Boolean and VDB workflow and then run edge, plane, curvature (Gaussian curvature, thereshold, fillets, radiuses rtc) detection on the model. In the end I expect curves procedurally laid out on the model along the creases, grooves etc which basically tells the edge feature flows on the model which in turn can be used to create Isoparms that would create the Quadmeshes. Further we, the users should have the option to draw our own user defined “feature” curves on the mesh that should be integrated in to the Quadmesh building as well. (Infact I was in the product development of a retopo software called Paraform - Stanford incubator project started around 1997 which did exactly this for which we were awarded Technical achievement award by the Academy in 2002. Paraform went into so many acquisitios and finally its with Dassault now - just a side note).
Retopolgy should be like Sculpting where drawing a stroke on the model should give the quaded mesh. When I see seeding point after point and create a Quad and then strip….. come on….. am I in 90's timeline?. This is 2018 Come on guys.. SideFx can really do better.
Also Man… I don't have time to figure out how I have to cut thru the N-gons and take the edge flows across the model to make a full Quad Mesh. This Quading - meaning breaking N-gons into Quads with Isoparms -edge flows solved to create Grided mesh MUST be automated. I would rather take my Mesh to Zbrush and do Zremesher to get the quaded mesh.
Quad Remeshing needs to be made even more smarter. Can I bing in a quad model and with a click of a button can I make it a quadmesh? I don't think so. CAD models has nice edges and boundaries… the only stupid thing with them is this sliver long skinny polygons. In Zbrush we can do polugrouping of the surfaces based on Orientation and use the edge boundaries for the Quad mesh boundaries and build Quaded Meshes nicely.
Here is my suggestion for Houdini's Modeling revamp:
For box modeling: - Go with ZModeler/Alexey Vanzhula' kind of approach. Tool behaviour and Target mode paradigm. Simple and straightforward. If you can do more.. its even better.
Houdini's current boolean based workflow is awesome: - Keep it up. Iterate and improve if you get a chance as still there are some bugs and failure cases thats annoying especially since I am in Procedural paradigm. This same failure cases dosn't bother us much say if we are in Maya because I can always delete it and move on by building fresh.
Retopolgy: - As I said it should be one click Quad-ReMeshing based on the features thats already in the mesh. Period. If I don't like the edgeflow done by the automation, then I sketch my intended “feature” curves and Houdini should figure out the way to integrate my intended edgeflow weaved into the Gridded/Quad mesh.
I am sure you guys can do it. I revere you guys.
God bless.
Thanks for your time.
Srinivasan Samuel.
The boolean based Workflow by Fiana is awesome. This is Houdini's key strength and please keep improving the complex model building by boolean Ops and think of every possible proceduralism to improve this workflow. Good
Now… Thanks for the sneakpeek of Polydraw 0.1 not a bad start.
I have been modeling for 25 years started with 3Dstudio 1.0 then to Alias and then Maya… then tried Modo for a while then gone back to Maya for my Box(Non-organic) Modeling workflows. But recently I stumbled with Zmodeler and blown away by its simplicity and power. Hands down… Zmodeler is the next generation Box(Non-organic) Modeling tool. I was amazed how some one like Mr.Alexey Vanzhula's could create some thing like the Zmodeler kind of UI in Houdini.
Don't copy Maya or other DCCS in this box modeling approach but please come with a typical “SideFX -Worflow”
Retopology workflow:
Please… Please…Please stop copying what Maya or other DCCs does for Polygon strip creation where we need to seed points after points by mouse click after click say on a sharpedge/groove in order to just lay an edge flow. Its the most slowest and dumbest way in today's world when you have a wonderful curve drawing tool where we can sketch curves directly on the model and then simply weave quadstrip along those curves. Houdini already has this Curve drawing functionality BUT its not implemented in the way it can be made useful for the modeling functionality. Please can you guys dust it off and give a fresh outlook to sketching based modeling which would be a real innovation.
Here is my dream retopo tool.
Either I bring in a hirez sculpt say from Zbrush or create a complex polygon model using Boolean and VDB workflow and then run edge, plane, curvature (Gaussian curvature, thereshold, fillets, radiuses rtc) detection on the model. In the end I expect curves procedurally laid out on the model along the creases, grooves etc which basically tells the edge feature flows on the model which in turn can be used to create Isoparms that would create the Quadmeshes. Further we, the users should have the option to draw our own user defined “feature” curves on the mesh that should be integrated in to the Quadmesh building as well. (Infact I was in the product development of a retopo software called Paraform - Stanford incubator project started around 1997 which did exactly this for which we were awarded Technical achievement award by the Academy in 2002. Paraform went into so many acquisitios and finally its with Dassault now - just a side note).
Retopolgy should be like Sculpting where drawing a stroke on the model should give the quaded mesh. When I see seeding point after point and create a Quad and then strip….. come on….. am I in 90's timeline?. This is 2018 Come on guys.. SideFx can really do better.
Also Man… I don't have time to figure out how I have to cut thru the N-gons and take the edge flows across the model to make a full Quad Mesh. This Quading - meaning breaking N-gons into Quads with Isoparms -edge flows solved to create Grided mesh MUST be automated. I would rather take my Mesh to Zbrush and do Zremesher to get the quaded mesh.
Quad Remeshing needs to be made even more smarter. Can I bing in a quad model and with a click of a button can I make it a quadmesh? I don't think so. CAD models has nice edges and boundaries… the only stupid thing with them is this sliver long skinny polygons. In Zbrush we can do polugrouping of the surfaces based on Orientation and use the edge boundaries for the Quad mesh boundaries and build Quaded Meshes nicely.
Here is my suggestion for Houdini's Modeling revamp:
For box modeling: - Go with ZModeler/Alexey Vanzhula' kind of approach. Tool behaviour and Target mode paradigm. Simple and straightforward. If you can do more.. its even better.
Houdini's current boolean based workflow is awesome: - Keep it up. Iterate and improve if you get a chance as still there are some bugs and failure cases thats annoying especially since I am in Procedural paradigm. This same failure cases dosn't bother us much say if we are in Maya because I can always delete it and move on by building fresh.
Retopolgy: - As I said it should be one click Quad-ReMeshing based on the features thats already in the mesh. Period. If I don't like the edgeflow done by the automation, then I sketch my intended “feature” curves and Houdini should figure out the way to integrate my intended edgeflow weaved into the Gridded/Quad mesh.
I am sure you guys can do it. I revere you guys.
God bless.
Thanks for your time.
Srinivasan Samuel.
Edited by hsolomon - 2018年8月24日 12:44:06
- BabaJ
- Member
- 2129 posts
- Joined: 9月 2015
- Offline
Form what I saw in the polydraw intro video is that it could benefit greatly as is with no further work by just improving the existing construction plane ‘tools’.
* Easily create contruction planes on the fly with the ability to save like as a library with specific name/numbering.
* Objects that can be placed on the construction plane with easy orientation to the construction plane, e.g. make them planar or perpendicular. And abilitiy to translate on the construction plane.
* Able to set views planar/perpendicular to the construction plane and subsets of those viewing orientations.
* Able to set polydraw points to constrained measurement values to objects seen in planar view of the construction plane as if those objects were projected on the construction plane. (Ok sorry, that's a bit much - can't help my SolidWorks background coming in here).
The existing construction plane ‘tools’ already have some functionality towards these points like ‘set orientation picking’.
This allows to do things like set up construction plane relative to polygon faces or whatever other geometry points that can be clicked on.
The next step for that tool would be able to save those construction planes and allow the objects placed on the plane or ploydraw points drawn, curves placed, etc. to ‘inherent’ to orientation of that construction plane.
In the hip I've attached I created a rough embedded asset that creates orientable grids from which construction planes can be made via the ‘set orientation picking’.
But it is very limited because of the lack of orientation to the plane.
I believe that with the above implementation that allows quick creation and orientation to the construction plane many objects can be created quickly with the boolean tools.
When I used to work in SolidWorks I was able to create many organic shapes easily and quickly because of the innate working of the construction planes.
In the hip file the objects created are terrible for an example, but I think it's enough to give an idea.
There's more ‘artist’ types coming to Houdini who just want to use existing tools and make something and not dive into making their own tools, which by the way is completely possible to implement on ones own for my above example.
One ‘niche’ market I can think of that would help with the above is 3D printing.
The construction plane tools and features I mentioned above along with boolean operations helps immensely especially if there is a need for designing parts and pieces, even if organicly shaped in nature, to have specific measurement tolerances, sizes, etc.
Being able to create in Houdini as such for the final product also allows corresponding base scene geometry to created animations, images for promotion material.
Like, say design some LED flashlight prototype, but at the same time illustrate that with a scene showing someone using it camping.
* Easily create contruction planes on the fly with the ability to save like as a library with specific name/numbering.
* Objects that can be placed on the construction plane with easy orientation to the construction plane, e.g. make them planar or perpendicular. And abilitiy to translate on the construction plane.
* Able to set views planar/perpendicular to the construction plane and subsets of those viewing orientations.
* Able to set polydraw points to constrained measurement values to objects seen in planar view of the construction plane as if those objects were projected on the construction plane. (Ok sorry, that's a bit much - can't help my SolidWorks background coming in here).
The existing construction plane ‘tools’ already have some functionality towards these points like ‘set orientation picking’.
This allows to do things like set up construction plane relative to polygon faces or whatever other geometry points that can be clicked on.
The next step for that tool would be able to save those construction planes and allow the objects placed on the plane or ploydraw points drawn, curves placed, etc. to ‘inherent’ to orientation of that construction plane.
In the hip I've attached I created a rough embedded asset that creates orientable grids from which construction planes can be made via the ‘set orientation picking’.
But it is very limited because of the lack of orientation to the plane.
I believe that with the above implementation that allows quick creation and orientation to the construction plane many objects can be created quickly with the boolean tools.
When I used to work in SolidWorks I was able to create many organic shapes easily and quickly because of the innate working of the construction planes.
In the hip file the objects created are terrible for an example, but I think it's enough to give an idea.
There's more ‘artist’ types coming to Houdini who just want to use existing tools and make something and not dive into making their own tools, which by the way is completely possible to implement on ones own for my above example.
One ‘niche’ market I can think of that would help with the above is 3D printing.
The construction plane tools and features I mentioned above along with boolean operations helps immensely especially if there is a need for designing parts and pieces, even if organicly shaped in nature, to have specific measurement tolerances, sizes, etc.
Being able to create in Houdini as such for the final product also allows corresponding base scene geometry to created animations, images for promotion material.
Like, say design some LED flashlight prototype, but at the same time illustrate that with a scene showing someone using it camping.
Edited by BabaJ - 2018年8月26日 10:49:10
-
- Quick Links