project "Houdini, a great modeler"

   223508   609   9
User Avatar
Member
8 posts
Joined: April 2010
Offline
McNistor
Jholen
SideFX, make the right choice, consulting or hire Vitaly Bulgarov for improving Modelling inside Houndini.

What's wrong with me? Now I'm hurt.

Oh no, not my intention! You doing a great job

I think the best is take some famous name and work directly with him for improving but also publicizing the tools.

I started my softimage experience watching and studying the great tutorial done for gnomon by Vitaly.

So, is my opinion you (SideFx) need to improve modelling workflow and tools set, but also publicizing it very well (maybe with some spectacular tutorial like done by Vitaly with softimage). Houdini is know for FX and procedural work flow in general, but also is known to be a poor modeler. You must improve but also break this bad perception.
User Avatar
Member
210 posts
Joined: Jan. 2014
Offline
McNistor
I'm proposing ways to make Houdini a better tool for modelers too, the way I know. It's inevitable that most of this is going to be based on past experience which in my case is Softimage (mainly but not only), but that's irrelevant.
What's relevant is this: “will this improve the workflow or not?”

A dev. team member endorsed my idea of creating a list with things for a better modeling and viewport interaction which is exactly what I'm doing. In the end, if people like it with the will of the dev. team, things will get implemented. Or not.

Most of your points are workarounds. I'm proposing improvements. If I wanted workarounds I'd stick with Adsk, Maya or god forbid, 3dsMax.

There are plenty bad tools out there and people are used to them. Not long ago, people were used to the idea of a flat Earth.
Ultimately, I'm not going to fight “get used to it” arguments as this is not a rational approach by which progress is made.

Thank you for your input.

I havn't read the entire thread so far that's why I'm not bringing up my suggestions at this point but this one thing I wanna point out:

Thank you McNistor for this thread!
I had these discussions as well here in the forum and I just wanna say, I know what it feels like. :wink:
As soon as I have read the thread completly I will add my suggestions here if they were not mentioned yet.

Edit:
Since there is a lot of discussion about loop and ring select, is everybody here aware of Cinema 4Ds loop and ring selection. They are, as far as I know, the best solution 3D world has ever seen. They are easy/ intuitiv to use because of their great visual feedback.
You can skip the first 3 minutes if you like:
http://www.youtube.com/watch?v=MSuBpTNRJYE [youtube.com]
User Avatar
Member
210 posts
Joined: Jan. 2014
Offline
pezetko
Double click method are not that userfriendly with tablet as input device. I prefer the loop selection with ALT key as it's now in XSI, it's faster and works with tablet without problem.

Pie menus (marking menus) that are context sensitive like in Maya are huge timesaver also. I really like them. It takes a little time to learn them and the speed boost is incredible. And they work great with tablet or mouse.
E.g. I could switch coordinate space for objects (local/global etc) much faster in Maya then I'm able in XSI.

Quadmenus from 3dsmax or popup menus/right click menus are nowhere near the speed of using marking menus.

Marking menus +1
User Avatar
Member
385 posts
Joined: Nov. 2008
Offline
Korny Klown2
Edit:
Since there is a lot of discussion about loop and ring select, is everybody here aware of Cinema 4Ds loop and ring selection. They are, as far as I know, the best solution 3D world has ever seen. They are easy/ intuitiv to use because of their great visual feedback.
You can skip the first 3 minutes if you like:
http://www.youtube.com/watch?v=MSuBpTNRJYE [youtube.com]

Not from my point of view.

Visual feedback looks nice but I cannot see any benefit in it. How do you do partial loop selection in Cinema4d? How do you combine ring and loop selections? Do you have to switch the tools for this or it's all under one selection tool? From the video it looks like it needs a lot of switching between different tools for selection loops, rings etc. and that looks pretty slow compared e.g. to XSI or Maya selection modes (with ctrl, alt and double clicks for fast loop/ring selection, partial loop/ring selections).

From other videos it looks like it has separate tool for (only) whole loop selection, another separate tool for whole ring selection and so on. So I cannot agree that Cinema 4d provides anything best in this area.

Where the pre-visualisation is important is before creating new edges, new edge loops and so on. So you can see what effect it will have before you confirm the operation so you can decide if it would be better to move it slightly to the side (instead of blindly creating new edge loop, then selecting it and modifying it)
User Avatar
Member
210 posts
Joined: Jan. 2014
Offline
pezetko
Korny Klown2
Edit:
Since there is a lot of discussion about loop and ring select, is everybody here aware of Cinema 4Ds loop and ring selection. They are, as far as I know, the best solution 3D world has ever seen. They are easy/ intuitiv to use because of their great visual feedback.
You can skip the first 3 minutes if you like:
http://www.youtube.com/watch?v=MSuBpTNRJYE [youtube.com]

Not from my point of view.

Visual feedback looks nice but I cannot see any benefit in it. How do you do partial loop selection in Cinema4d? How do you combine ring and loop selections? Do you have to switch the tools for this or it's all under one selection tool? From the video it looks like it needs a lot of switching between different tools for selection loops, rings etc. and that looks pretty slow compared e.g. to XSI or Maya selection modes (with ctrl, alt and double clicks for fast loop/ring selection, partial loop/ring selections).

It's many years ago that I've been working with Cinema 4D but I remember something like selecting the “loop selection tool”, you get a visual feedback of what edgeloop you are about to select and then I think there was something like shift+drag to just select a portion of the loop, starting at the edge you started dragging from.
Yes loop and ring selection are two different tools but
1. that isn't much of a problem, as long as they are easy to access via a marking menu for example
2. I don't think it is impossible to combine them, so you get the best of both worlds.
User Avatar
Member
210 posts
Joined: Jan. 2014
Offline
So, now I have read the thread almost completly. I didn't had the time to read each post but most of them, so I'm now ready to participate on this thread.
So, if any of my suggestions was already mentioned here, I apologize in advance.


- Snapping to one axis
I mentioned this in another thread a few weeks ago.
Holding down “v” to access vertex snapping, grabbing, for example, the x-axis of the object/face/edge/vertex, so it only moves and snaps on this one axis. It neither moves nor snaps in any other axis.

- Make the interactive creation process optional in the preferences
This is one of the most annoying things when modeling. I just want to get a boolean node and hook up the inputs myself.
I call the boolean node from whatever place in Houdini and then it starts:
“…please select this to continue….now select that to continue…when you are finished press enter to confirm….” something like that. Maybe for some of you this is helpfull or you just got used to it but for me it is a really exhausting and time consuming process.

- Add “Access all nodes in TAB menu everywhere” option to the preferences.
What I mean by this is, when I'm modeling (on two screens) it is just time consuming to go back and forth between the viewport and the network pane, because some node you only get from there and other nodes you get from there. I don't mean with that, that the TAB menu always should display a list of all nodes, the displayed list can stay context sensitive but when searching for a node by typing it's name, then I want all nodes to be considered in that search. I hope thats clear.

- Implement an easy way to add a selction to a group. (Not a group to selection!)
So far the only ways to get my selected components into a group is to either select the components first and create the node then or to create a node and then type the components manually into the group slot. But when I already have a node with a group, let's say a polyextrude and later on I decide, that I want to have some more primitives/faces/polys to extrude, so I want to add them to the group, I have to do that manually. Why is there no “add selection to group”/“remove selection from group” button?

- Could someone please fix the bend deformer?
I know that many of you come up with their custum build tools for this but call me old fashioned when I pay 4.500 USD for a software I want to assume that the tools, it comes with, are functioning.

- I would like to have a real edgeloop SOP, not a polysplit in edgeloop mode.

I would have some more issues I would like to address but McNistor wants this thread to be explicitly for modeling so I will abstain from these points.
User Avatar
Member
1743 posts
Joined: March 2012
Offline
Korny Klown2
- Make the interactive creation process optional in the preferences
This is one of the most annoying things when modeling. I just want to get a boolean node and hook up the inputs myself.
I call the boolean node from whatever place in Houdini and then it starts:
“…please select this to continue….now select that to continue…when you are finished press enter to confirm….” something like that. Maybe for some of you this is helpfull or you just got used to it but for me it is a really exhausting and time consuming process.
I definitely agree. This is why I do everything node-based from the network editor. The inconsistency between the network editor, which actually lets you manage the nodes, and the viewport, which gives a rough semblance of being able to manage nodes, but doesn't quite, is confusing at best.

- Add “Access all nodes in TAB menu everywhere” option to the preferences.
What I mean by this is, when I'm modeling (on two screens) it is just time consuming to go back and forth between the viewport and the network pane, because some node you only get from there and other nodes you get from there. I don't mean with that, that the TAB menu always should display a list of all nodes, the displayed list can stay context sensitive but when searching for a node by typing it's name, then I want all nodes to be considered in that search. I hope thats clear.
Another excellent point. All *real* nodes are accessible from the network editor, but adding to the confusion, those real nodes have different names in the viewport TAB menu for some reason. Please submit an RFE for this, since you're probably far from the only one who's gotten confused by this.

- Implement an easy way to add a selction to a group. (Not a group to selection!)
So far the only ways to get my selected components into a group is to either select the components first and create the node then or to create a node and then type the components manually into the group slot. But when I already have a node with a group, let's say a polyextrude and later on I decide, that I want to have some more primitives/faces/polys to extrude, so I want to add them to the group, I have to do that manually. Why is there no “add selection to group”/“remove selection from group” button?
I really want this, because it's so annoying to have to reselect everything, (by hitting backtick or something esoteric like that, and it only works when the mouse is over the viewport and the node is selected in the network editor and the moon is waxing gibbous). When I asked some other people, they said it might just be me who runs into this all the time, so please submit an RFE for this.

…I want to assume that the tools, it comes with, are functioning.
<sarcasm>I'd recommend not trying the Cookie SOP too much.</sarcasm> (Sorry, that one's a pet peeve of mine.) Of course, in the viewport, it's called “Boolean” instead of “Cookie”. There was apparently once an actual Boolean SOP, but it was removed. If you can submit a bug report with a specific test case where the bend deformer produces incorrect results and give a description of the correct results (or give correct results themselves, generated elsewhere), that'd be great!

- I would like to have a real edgeloop SOP, not a polysplit in edgeloop mode.
Does it work to wrap the PolySplit SOP up as an asset named EdgeLoop with just the parameters you want exposed, or is it missing behaviour?
Writing code for fun and profit since... 2005? Wow, I'm getting old.
https://www.youtube.com/channel/UC_HFmdvpe9U2G3OMNViKMEQ [www.youtube.com]
User Avatar
Member
4189 posts
Joined: June 2012
Offline
Jholen
I started my softimage experience watching and studying the great tutorial done for gnomon by Vitaly.

So, is my opinion you (SideFx) need to improve modelling workflow and tools set, but also publicizing it very well (maybe with some spectacular tutorial like done by Vitaly with softimage). Houdini is know for FX and procedural work flow in general, but also is known to be a poor modeler. You must improve but also break this bad perception.

A super-star modeller using Houdini FTW; just as there are lots of people exulting the FX, shading, lighting & rendering currently.

Jholen
Oh no, not my intention! You doing a great job Very Happy

Agree, McNistor is a great leader for this Modelling Revolution. Let the Long March begin!
Edited by - March 18, 2014 17:56:59
User Avatar
Member
4189 posts
Joined: June 2012
Offline
ndickson
Another excellent point. All *real* nodes are accessible from the network editor, but adding to the confusion, those real nodes have different names in the viewport TAB menu for some reason.


Is that correct, which nodes are named differently? the viewport does add the shelf tools though.
User Avatar
Member
210 posts
Joined: Jan. 2014
Offline
One example I know is boolean (viewport) / cookie (network pane)
Probably there are a few more.
User Avatar
Member
4189 posts
Joined: June 2012
Offline
Korny Klown2
One example I know is boolean (viewport) / cookie (network pane)
Probably there are a few more.

Not too sure about that as Cookie is available in both the viewport and network view.

Please see the screen capture and explanation of shelf tools vs network tools in your other thread.

http://www.sidefx.com/index.php?option=com_forum&Itemid=172&page=viewtopic&t=31151 [sidefx.com]
User Avatar
Member
1767 posts
Joined: Oct. 2010
Offline
Does it work to wrap the PolySplit SOP up as an asset named EdgeLoop with just the parameters you want exposed, or is it missing behaviour?

I think Korny perhaps refers to the ability to procedurally create an edge loop group (i.e. based on an edge selection or an edge group). There is an example of this (I believe) in orbolt, as well as I sent an example of one that I developed for my work to sidefx (which I thought was a pretty good implementation ). PM me if you would like more detail

The note about the polysplit sop did make me think that sometimes, certain sops have some functionality embedded which may be useful as a separate sop altogether. For example, I love the divide's new triangulation options, but so many times I have found myself wanting to use that algorithm to fix existing triangulation (it currently only works on primitives that are not triangulated). The carve SOP is cool for generating curves out of wireframes, but I feel with a bit more work, you could also have a generate curve SOP which uses an edge group as an input and outputs a curve

In regards to poly split, being able to feed a curve as the split guide in the second input (instead of creating that crazy string!) would be VERY nice
-G
User Avatar
Member
210 posts
Joined: Jan. 2014
Offline
Not too sure about that as Cookie is available in both the viewport and network view.

Please see the screen capture and explanation of shelf tools vs network tools in your other thread.

http://www.sidefx.com/index.php?option=com_forum&Itemid=172&page=viewtopic&t=31151 [sidefx.com]

Ok, that explains why it behaves like that with shelf tools and nodes tools but, I think, even though it explains this behavior it doesn't justify it. This behavior can be quite confusing and meanwhile I know, that I'm not the only one who feels this way and since this behavior is technically not necessary it should be reworked at least in a way, that the user can decide in the prefs if he wants that behavior or not.

The carve SOP is cool for generating curves out of wireframes, but I feel with a bit more work, you could also have a generate curve SOP which uses an edge group as an input and outputs a curve

I want that generate curve sop
That, by the way, is a point I want to express explicitly. Naming in Houdini is a big issue and although most Houdiniusers will stone me for saying this I would like to see that addressed.
Why is the cookie node called cookie? It doesn't cook anything! It is a boolean operation.
There are so many things in Houdini called “primitives”and so many things called “group”.
Yes, I understand that primitive stands for a lot of things when your are going for the exact meaning of the word. But Houdini is neither an english class nor a dictionary where the exact meaning of a word matters. We just want a software that is powerful and easy to use so who really cares at the end of the day what “primitiv” really means, just call the one thing primitive and the other face. This is confusing and not necessary, I repeat that: It's not technically necessary. My mom called this “tidy your room” and this is one big step to make Houdini a great, intuitiv modeler.


…oh and since it has been mentioned by someone already
viewtool is redundant +1
User Avatar
Member
53 posts
Joined: March 2014
Offline
Sorry if this has been mentioned previously, I've just been skimming the posts, but… one thing I've always wished Softimage would have picked up for modeling is Modo's falloff's.

If you haven't seen them in action, check them out.

I love how they visually represent them in the layout. In Softimage's M-key you had an fCurve to modify the falloff, but you couldn't choose different shape falloffs like you can in Modo and you're limited to seeing points or edges light up based on the falloff.

-Paul
User Avatar
Member
210 posts
Joined: Jan. 2014
Offline
Here are a few more things I discovered.

- on geometry level there is no way to select the entire primitive object
1 is selecting the geonode itself on object level
2 is for points
3 is for edges
4 is for primitives/faces
If I have a geo node with two primitive objects in it (box and sphere) and I wanna grab the entire sphere and move it with the move tool “T”, I can't. I can either move it numerical in the parameters tab, select all components and move them with an edit sop or use transform node but it is not possible to move just the primitve object on it's own with the move/translation tool.

- I don't know if this was already pointed out the viewport is not very modeling friendly.
1. grey geometry on grey background (I know there is a dark/black version but that one is no better since the smooth shading fades into pure black and objects are hard to see from some angles.)
2. the initial shading (smooth shading) looks horrible, espacially the pure black
3. the lines of templated objects are often hard to see since they have the same thickness and color like the ortho grids. This is especially true for boxes. This might be a personal psychological thing but since the template flag is pink, my eye expects to see templated objects as a pink wireframe.

- maybe this is already possible nad if so I apologize. I sometimes feel the need to have the display flag follow the current selected node
User Avatar
Member
210 posts
Joined: Jan. 2014
Offline
Paul Griswold
…Modo's falloff's.

If you haven't seen them in action, check them out.

1. I did check them out
2. they look preety awesome
3. now I want them too

It's like a mask input in Nuke.
This would be a nice tool in Houdini.

Apart from the falloffs I just found this one on youtube.
http://www.youtube.com/watch?v=i1_OI5U9avE [youtube.com]
Is something similar to modos “constrain to background” feature available in Houdini?
User Avatar
Member
1751 posts
Joined: Dec. 2006
Offline
not answering your question but http://www.youtube.com/watch?v=DxaWGSD_znc [youtube.com]
daniel bukovec | senior fx td | weta digital
qLib -- http://qlab.github.io/qLib/ [qlab.github.io]
User Avatar
Member
1743 posts
Joined: March 2012
Offline
Korny Klown2
Why is the cookie node called cookie? It doesn't cook anything! It is a boolean operation.
There's a long argument that I don't really care much about, but to present part of the side that I don't agree with, Cookie doesn't actually do a boolean operation when you have open or non-manifold surfaces. If you Cookie an arm against a grid that's cutting it off at the shoulder, it tries to do a flood fill to keep (or remove) the whole arm, even if it crosses back over the plane of the grid away from the actual grid. (I'd retort that it doesn't even handle a lot of closed, manifold surfaces properly.) I agree with you that it should probably be named Boolean, so that it can be found more easily in the network editor, instead of only being aliased as Boolean in the viewport.

There are so many things in Houdini called “primitives”
This one, I don't quite follow. In Houdini, a detail (I certainly agree that's a weird name for a piece of geometry) contains primitives, vertices, and points. Primitives own vertices, and vertices reference points. That's really all that a primitive is; an owner of its vertices. That stays consistent regardless of the primitive type, even with a bunch of the more esoteric primitive types that nobody uses or knows about anymore.

It's often advantageous to have multiple different primitive types in a single detail, and I feel that it might be more confusing if polygons weren't considered primitives, because they behave just like every other primitive from a generic point of view. Yes, there are operations that are specific to particular primitive types, but there's a lot that is in common between all primitive types.

What sort of use case were you thinking of to benefit from a distinction between polygon primitives and other primitives? If it's for making selections for polygon-specific tools, we could probably change the text so that it says something like “select polygons” instead of “select primitives” once the tool's been specified/selected to make it clearer that it only supports polygons.

and so many things called “group”.
I think you'll have to clarify this one, because groups just specify groups of things, so at that level, it doesn't really seem like jargon to me. Yes, there are primitive groups (groups of primitives), point groups (groups of points), edge groups (groups of pairs of points), vertex groups (don't tell anyone I told you they exist), and breakpoint groups (I don't really know what they're for), but they're all just groups of things. Could you elaborate a bit?
Writing code for fun and profit since... 2005? Wow, I'm getting old.
https://www.youtube.com/channel/UC_HFmdvpe9U2G3OMNViKMEQ [www.youtube.com]
User Avatar
Member
210 posts
Joined: Jan. 2014
Offline
buki
not answering your question but http://www.youtube.com/watch?v=DxaWGSD_znc [youtube.com] Very Happy
I'd say this answers my question quite perfect. thanks. Although I think, once again, that the name “ray” is a bit confusing. When I hear of a raynode I expect it to do something with volumetric light instead of constraing geo to the volume of another geo.

ndickson
and so many things called “group”.
I think you'll have to clarify this one, because groups just specify groups of things, so at that level, it doesn't really seem like jargon to me. Could you elaborate a bit?

What is the first thing you think of, as a normal human being not as a Houdini user, when I say the group?
A group of people, a group of trees…..maybe a group of nodes?
What is the first thing you think of when I say the word selection?
I think you see what my point is.

Many nodes a have group slot which tells the node which components to operate on. This in my opinion is more accurately named as something like “component selection” or “component mask” because it masks out on which components to operate on which not.
Then there is the group node which basically just stores a selection, of whatever component type, for later use. So this node should be named more accurately something like “selection store” or so.


Edit:
I know I'm bothering but there is one thing I just found out which would be pretty helpful. In the Python doc it's hard find what you are searching for. What I was desperately searching for is a library where you can find all Python commands and parameter expressions, something like that:
http://download.autodesk.com/global/docs/maya2013/en_us/index.html?url=files/GUID-D9C648CB-2C71-4CE2-99EC-719445BEEDC0.htm,topicNumber=d30e87720 [download.autodesk.com]
I remember that I once found a list of commands somewhere in the help but I can't find it anymore and it wasn't complete I think. So I would be very happy about a complete Python and, for the HScript users, HScript command reference that is easily findable in the help.
Edited by - March 19, 2014 14:08:38
User Avatar
Member
1743 posts
Joined: March 2012
Offline
Korny Klown2
A group of people, a group of trees…..maybe a group of nodes?
What is the first thing you think of when I say the word selection?
I think you see what my point is.
Actually, I really don't, sorry. How is a group of points any less of a group than a group of trees? It's still a group of things, albeit that points are intangible, being inside a computer. Maybe I'm missing some distinction.

If I asked a random person not in graphics what “a selection” is, they might say something like “the act of choosing a single thing from among many”, not “the group of things to which I want to do something”. If I asked them what “a group” is, they might say something like “multiple specific things”, which is pretty much what a group is in Houdini.

Many nodes a have group slot which tells the node which components to operate on. This in my opinion is more accurately named as something like “component selection” or “component mask” because it masks out on which components to operate on which not.
Then there is the group node which basically just stores a selection, of whatever component type, for later use. So this node should be named more accurately something like “selection store” or so.
Groups don't have to be created from selections, though. Groups are quite often procedurally generated with SOPs instead of being selected in the viewport, in which case, they were never selections.
Writing code for fun and profit since... 2005? Wow, I'm getting old.
https://www.youtube.com/channel/UC_HFmdvpe9U2G3OMNViKMEQ [www.youtube.com]
  • Quick Links