Tree / plants generation tools.

   542   8   3
User Avatar
Member
9 posts
Joined: Jan. 2015
Offline
Hello there,

It would be really awesome if a couple of tools could be added to handle tree generation and modeling.

I know, everything needed is there, but this kind of work require math skills and expertise I don't have, and despite the large size of our studio, there is no resource to create that kind of tools in-house. (We are only a bunch of Houdini artist here @Gameloft).

This is one of the only reason I have to start 3dsmax these days, only for GrowFX (https://exlevel.com/features/) … it's a really great product, but it will be a such a better fit in Houdini, … relying on a stack modifier system for this kind of modeling is a real PITA. Having to use Max to create procedural content hurt me every time

Beside, TheGrove for blender (https://www.thegrove3d.com/releases/the-grove-release-6/)has a very nice approach too, it's more on the life simulation side (self organization) than the simple recursive distribution. It's however deadly slow to generate half dense model … and again, this kind of procedure should probably flight once done properly in VEX … well I suppose.

I don't know what would be the better approach, or the easier to implement … but a little bit like heightfields was a really amazing addition (and easy to use), maybe a couple of HDA might help for this kind of tasks. Anastasia work on this looks just wonderful ! : https://vimeo.com/229252148 [vimeo.com]

Thanks.
Edited by PaQ WaK - Sept. 11, 2018 19:03:15
Gameloft
User Avatar
Member
27 posts
Joined: Dec. 2016
Offline
I opened an RFE recommending adding plant/vegetation growth/simulation and was told to just use L-systems. ¯\_(ツ)_/¯

On a more serious note, there is this https://gumroad.com/l/zWFNX [gumroad.com]

It would be nice to see something native to Houdini though, with all the support and documentation that goes along with it, something that will be maintained as new versions of Houdini come out.
Edited by RobotHeadArt - Sept. 11, 2018 19:55:45
User Avatar
Member
17 posts
Joined: April 2016
Offline
In Anastasia Opara's excellent talk:
https://vimeo.com/228391688 [vimeo.com]

She demonstrated a very advanced procedural tree generation system without L-system. To achieve this advanced knowledge on vex and biology(On how a tree grows) is definitely required.
User Avatar
Member
27 posts
Joined: Dec. 2016
Offline
Kai Xu
In Anastasia Opara's excellent talk:
https://vimeo.com/228391688 [vimeo.com]

She demonstrated a very advanced procedural tree generation system without L-system. To achieve this advanced knowledge on vex and biology(On how a tree grows) is definitely required.

Exactly! Imagine that as native tools in Houdini that could also interact with the Terrain tools. You could have ecosystems that grow according to elevation/erosion/waterflow. SideFX could really give e-On a run for their money competing with Vue+Plantfactory.
User Avatar
Member
70 posts
Joined: June 2013
Offline
RobotHeadArt
Kai Xu
In Anastasia Opara's excellent talk:
https://vimeo.com/228391688 [vimeo.com]

She demonstrated a very advanced procedural tree generation system without L-system. To achieve this advanced knowledge on vex and biology(On how a tree grows) is definitely required.

Exactly! Imagine that as native tools in Houdini that could also interact with the Terrain tools. You could have ecosystems that grow according to elevation/erosion/waterflow. SideFX could really give e-On a run for their money competing with Vue+Plantfactory.


Except that it is the role of an artist to study and learn and experiment and develop in their area of interest in order to make excellent art that addresses their needs. Houdini provides a huge workshop equipped with excellent tools to let them do that, not a widget factory with a showroom up front where you can order the thing you want and have it drop out the other side.

Really do not think SideFX is concerned with giving e-On a run for their money when their software is lightyears beyond what Vue is capable of creating. Thankfully the engineers are focused on the core elements of Houdini not studying botany. And no I am not some crusty Houdini veteran wizard in a high tower, I am a crusty Maya veteran in the dungeons who sees there is a better way and is looking forward mastering it so I can make whatever I want not just pick a nice tree from a list of nice trees… (Orbolt is for that.)

Cheers!
User Avatar
Member
177 posts
Joined: Jan. 2018
Offline
Rebus, I'm really thankful that I don't see any indication that the SESI team shares your point of view.

It wouldn't surprise me one bit if H17 has exactly the type of tree generation tools that RobotHeadArt and PaQ WaK are talking about.
>>Kays
User Avatar
Member
460 posts
Joined: Oct. 2016
Offline
In my experience, the “plant-topic” is suffering mostly from lack of (financially backed) interest. I have spoken to a couple of developers involved in (in-house) projects for realistic plant-creation-tools and while developing up to a working proof-of-concept stage is more or less trivial, making the output LOOK GOOD is what takes time, experimentation and research (in a context that in ancient time was called “nature” and today is considered a service provided by Facebook).

I have used Grove a couple of times and really liked its approach, however, its pythonesque single-threadedness makes it almost unusable, as you have to experiment with settings to get the looks just right. Yet, I am pretty sure that some of its ideas could be converted into multi-threading and while this wouldn't make it FAST it could improve on the user experience quite a bit. The question is: Who is willing to pay for the development needed? I don't see anywhere close enough to sufficient sales being made from a multi-threading-version of Grove, so all development would have to be paid in advance.

Those devs I spoke to said that *in* *theory* they would be willing to create a library (that could then be turned into Houdini nodes) for multi-threading and GPU-based plant-creation that is art-directable (not exactly realtime, but something like 2-5 fps with two million points being placed) if there was “real demand” (as in “income” to be made). Looking at sparetime-projects released for Houdini that have been discussed here on the forum, my reaction was to just shrug and say “OK, I see your point”. With libraries of ready-to-use plants out there it would be almost impossible to break into the Arch-Vis market for example (why use a tool that requires setup when you can use ready-made models?), and that, according to those devs, is the best paying market within this context.

The quintessence, as far as I can see, is: There isn't enough demand for a GOOD plant-creation-tool to create one “for free” within Houdini (that is how I, were I employed at SESI, would see it). There isn't enough demand for a PAID tool, because most customers, who'd be willing to invest money would rather buy EXACTLY the model they need instead of wasting time on trying to create it.
What's left, is: Everyone who needs it builds it for themselves. And since that will always be pipeline-oriented, it probably isn't useful to anyone outside.

Personal sidenote: It would be cool if there was a creative, constructive Houdini community where such a project could be initiated, a small group of able devs would meet or frequently connect and build tools like this out of their personal interest, selling it once its done (but not for the money). In my experience, the able devs are employed and working overtime, the ones who “want” such a tool cannot help in building it and those with the money to support the development have different interests.

Sorry for the negative view, feel free to downvote me

Marc

Software Sewers Service (aka Pipeline Problems Solver)
https://vimeo.com/marcalbrecht [vimeo.com] - https://www.marc-albrecht.de [www.marc-albrecht.de]
User Avatar
Member
1469 posts
Joined: March 2014
Offline
Guys, you know you can email Anastasia and show your interest and willingness to pay for such a plant generation tool. The more users are requesting, the higher the chances that she'll consider monetizing the time and effort she's put into that system so far. I'm a little presumptuous, of course - maybe she has no interest whatsoever in making that tool public. But other might be. Launch a kick-starter or something and see how many are actually interested in this, because ultimately it's about resources regardless of where are those going to come from.

I'm not, in principle, against SideFX creating high-level tools and systems. But as others have said, there are currently many low-level things that need immediate addressing, which means that only SideFX can implement them properly i.e. best integration possible.
https://www.artstation.com/artist/mcnistor [www.artstation.com]
metaphysical wildcard
User Avatar
Member
447 posts
Joined: June 2006
Offline
so much talk for a link
https://forums.odforce.net/topic/20263-houdini-procedural-tree-generator/ [forums.odforce.net]

file: https://forums.odforce.net/topic/20263-houdini-procedural-tree-generator/?do=findComment&comment=128707 [forums.odforce.net]

somebody has made a setup 4 years ago that is quiet good.
otherwise you can use speedtree.
  • Quick Links