Jump to content
Jason

Houdini 10 Wish List

Recommended Posts

Here's three votes from me (these may well have been suggested already - I'm new to this thread).

1) Agents with Fuzzy Logic

I've been spending the last little while playing with Massive and have barely touched Houdini

recently. Massive is very powerful but is built on a relatively lightweight infrastructure which

suggests to me that building an AI context in Houdini is very much doable. By integrating

AI into Houdini you could have the advantages of agents interacting with particle and dynamics

simulations amongst Houdini's many other tricks. I want to see battle scenes with soldier-agents

firing and running away from flamethrowers! Yeah baby! The seat license for Massive is hugely

expensive suggesting that SESI could make some money by building a tool that works standalone

and can be licensed as such or made available as plug-in that integrates with the main Houdini package.

2) Off-the-shelf Shaders

Whilst I'm still yet to put much serious time into Houdini 9 I do really like the friendlier interface.

texturing is still far from straightforward for the introductory user though. It's great to

have the power of vex for us seasoned users but it would be really helpful to have many more

pre-built, off-the-shelf, well-documented shaders for new users.

3) DOPs - 'dynamic spawning'

One thing that really bothers me about DOPs is the requirement for all objects to be defined

up front. So if you know you want an object to break into pieces you have to define the

fragments first - or at least the number of fractured pieces that are going to be needed. I

went along to a SESI presentation of Houdini 9 DOPs in LA last summer and I remember being

told that DOPs forbids having a new DOPs object come into existence during the simulation

but I don't buy that it's impossible to make it work that way. It would be sooooo useful

to be able to 'spawn' a new object procedurally given a trigger event. If we can get that

functionality then we can make objects shatter without having to pre-fracture them which

would be a huge bonus. I'm sure there would be plenty of other uses too.

John Hughes

Share this post


Link to post
Share on other sites

I second the fuzzy logic, start a trend like with l systems ... or maybe it could

even be integrated into CHOPs or POPs

1) Agents with Fuzzy Logic

I've been spending the last little while playing with Massive and have barely touched Houdini

recently. Massive is very powerful but is built on a relatively lightweight infrastructure which

suggests to me that building an AI context in Houdini is very much doable. By integrating

AI into Houdini you could have the advantages of agents interacting with particle and dynamics

simulations amongst Houdini's many other tricks. I want to see battle scenes with soldier-agents

firing and running away from flamethrowers! Yeah baby! The seat license for Massive is hugely

expensive suggesting that SESI could make some money by building a tool that works standalone

and can be licensed as such or made available as plug-in that integrates with the main Houdini package.

Share this post


Link to post
Share on other sites

I can only add that a lot of the things I want have already been mentioned. More than anything finish and document all those half implemented things like hairs lack of styling tool set !.

R

Share this post


Link to post
Share on other sites

I am pretty new to Houdini...so it's a bit hard for me to really nail down something that needs improvement besides myself.

I think that the only thing I find annoying is the license. Every so often, I get a pop-up message saying that it can't find a license or something. It always works after a few tries, but I guess it can use some improvement.

~Ilan

Share this post


Link to post
Share on other sites

1. Mac Version!

2. Better Documentation and A Lot More Tutorials for New, Intermediate, and Advanced Users. Progression of learning as you progress.

3. Better Python Integration

4. MAC Version (It is just that important to me!)

5. Speed improvements over the entire package

6. Better support of multi-threading over the entire package

Just my two cents

mayadesigner

Share this post


Link to post
Share on other sites

Just a minor rfe spurred by the email...

Speed up the weight painting brush workflow a bit. If using ZBrush or Silo etc to sculpt with, and then having to come to Houdini afterwards to paint weights, it's really painful (it just feels clumsy and slow). So just borrow a little from what they do :).

- add an assignable sticky key. I'd see the user having the smooth operation assigned to it most often to allow quick toggling between painting and smoothing, but what ever they fancy really. My preference would be to have this sticky key assigned to the shift key (which is inline with ZBrush / Silo - not sure what other apps do though if there is a standard??).

- lmb / rmb are for painting etc

- ctrl+rmb is for the operation options popup

- mmb drag horizontal to adjust the brush size.

- mmb drag vertical to adjust the opacity.

- alt+lmb or alt+rmb acts as a sticky key which does the opposite of lmb / rmb. eg: If lmb is set to paint, then alt+lmb acts as reduce and so on... (or perhaps this should be assignable too, but maybe then there'll be too many options??)

- also, when painting it'd be convenient to have an option to automatically make the last painted part of the geo act as the pivot for the next viewport navigation tumble action without needing to always press space+z.

- finally, some method of quickly dragging the mouse in the viewport to scrub the playbar without actually having to move the mouse to the playbar location.

Add some or all of these and painting weights could actually be fun. :)

Edited by Moose

Share this post


Link to post
Share on other sites

Hmmm, so much to choose from.

I've deliberately either not read, or forgotten, the rest of this thread to avoid undue influence on myself :) Also, a lot of things fall into the same categories, so rather than order by importance, I'll order by group importance and just list what's needed (IMO) in each category.

-UI customization.

Why are there 3 Snap icons next to the viewport when so many more important things are buried in submenus? I don't care, as long as I can customize.

Equivalent to OPmenu but for the RMB Channel menu on parameters. Countless opportunities to simplify and expose power, like Stamping.

All menu customization. Submenus just are bad UI design. Again, don't care as long as I can move stuff around.

-Lighting. It's awesome that Mantra has PBR and all this newly exposed customization stuff, multi segment blur etc.

We need IPR, Hal Bertram speeds. Without this, don't bother shipping 10. IMO.

Proper in-viewport Light Linking.

Take Spreadsheet (can anyone use Takes seriously without this?)

VOP library for production shaders.

There needs to be, at minimum, a standard system of VOPs that output deep rasters for each light's diffuse and specular, etc etc

All VEX functions with all options be exposed in VOPs.

Native, supported and complete Texture Baking (-u).

-Documentation. Lightyears better than Houdini 7 and 8 but of course, there's always more...

Some love for the HDK. I just started coding in the Maya API and wow, having some docs means I can actually produce something useful in a day!

Lots more examples, especially for Python stuff.

Lots more examples generally :)

-Character tools. Wow, Autorig is awesome, but ...

Unify Muscles and Capture Regions.

Make all the capture weight editing tools work with Muscles. Bone (Muscle?) Handles, spreadsheet etc.

Cloth. Real, nCloth competition. Not trivial I know, but essential.

Complete Fur system. The components are there, now bring it together!

-Misc

Gzipped backup files at least.

The Fracture SOP mentioned for 9.1 but never quite finished in time...

Native SubD primitives for modeling speed. Or at least optimize the SubD SOP so it only re-SubDs deltas.

More threading of slow-ish operators. POPs threading would be nice.

-COPs - should I even bother mentioning it? So much potential...

Sleepy, must sleep. More later as I'm annoyed by things while working :)

Cheers,

Peter B

Share this post


Link to post
Share on other sites

I originally investigated Houdini for use in a real-time 3D rendering engine. I've never used it, but it's definitely got potential.

I really admire that it's entirely procedural, and that it has python scripting.

The only thing it needs right now is a real-time renderer that can be used for games.

I'm still interested in this, but right now I'm busy with Python-Ogre, which is a great project, if a little difficult to build.

Perhaps a good interface between any Houdini libraries available, Collada, and Python-Ogre would be warranted for this sort of functionality.

Procedural assets are nice because of their importance in any sort of open-source repositories that may exist for Houdini assets. You can modify every aspect as the original designer would, as if you had a PSD rather than a JPG, to use raster graphics terms. Also, the ability to program shaders would be so nice for game development. Python is also great for folks who are new to programming, and also for people who know they're going to need a lot of meaningful code for their game project.

Also, sorry if this is a little off discussion; I'm not a regular here, and came only because of the bulk-email. If anyone here is already doing this in some form or other, I wouldn't mind hearing about that. :)

Share this post


Link to post
Share on other sites

Sorry if these are suggested elsewhere:

1) Enhance the modeling tools a bit to support a bit more of the non-procedural workflows. This is more of a concession of my ignorance (I'm no stu! ^-^), but when I model something, save, then reopen the file the next day and somehow the point numbers have changed; my model is garbage because my edits don't stick. Bridge polys, or edges should perhaps work in a fashion more like XSI or Modo, and so on. Houdini would be that much better if I could mix modeling styles without using other packages.

2) Also coming back to Modo: 3d texture painting/projecting! :) Just think of the power if we could use vops to create textures or even brush textures, etc.. It makes me drool.

3) Spline control for my VOPs stuff. Or is this in 9.1 by now? hehe

I don't know enough about the entire package to really say more. I'm lacking a lot of knowledge of the character tools, but I'd love a great way to mix animation clips... wait a sec... CHOPs... never mind. ;) Is there a groovy sequence editor though?

Share this post


Link to post
Share on other sites

1- the ability to create custom widgets. QT implementation would be great or an API based on Houdini widgets. We need to be able to

create menus, interfaces etc. I think the reason why Maya is so popular is because it's so easy to do that programatically. Not that it's

not already possible to do it externally, but I'd prefer a direct binding

2- clean up DOPs. It's very incosistant and rather annoying to get used to. I like the modularity of having diffrent otls but there isn't

enough documentation and the workflow is always sloppy. I think the DOP commands are very well though of but we need to be able to

query one object without having to write extremely long expressions. And having to open a details view to find the object and its

parameters is annoying.

3- Spend more time for more in depth documentation on HOM and DOPs.

4- don't remove HScript but add more functionality to HOM and please more documentation I can't stress that enough.

5- Again on the documentation side, we need better docs ond a wiki for the SDK.

6- It's been said before but here it goes again. Add full 64 Bit support and multi-threading. I hear more applications are hyperthreaded

now such as XSI and ZBrush. I think it could be a good selling point.

Basically performance is key.

SESI team probably cursing us right now as they're getting a bunch of RFEs :P

Share this post


Link to post
Share on other sites

not had the chance to play with houdini as much as i'd like but here goes!

1) Nurbs, Polygon/Subd based muscle objects with identical muscle capture features and properties as current metaballs muscles. ( i need to be able to shape my muscles exactly as i want them and also drive their shapes with blendshapes or displasement maps if need be. Metaballs don't allow for such control because of their everchanging topology and point count

Basically we should be able to use our choice of custom-shaped poly or nurbs surfaces as our muscle objects for everything including capture regions and painting sliding modifiers

2) Autocad DWG import/export (ideally with layers recognizes as groups in the file Sop) Autodesk 'OWNS' anything cad and design as long as maya and max are the only apps that trully support .dwg.s

3) simple straightfoward shaders like every other package has...

4) A non-procedural super modeling sop intuitive and artist friendly, similar to the legendary model sop that is supposed to have existed in houdini back in the day.

Just for modeling puposes with no intention of maintaining proceduralism in model. Heck pretty much a stand-alone modeling app like silo, clay or nendo enclosed i a sop

Share this post


Link to post
Share on other sites
3) DOPs - 'dynamic spawning'

One thing that really bothers me about DOPs is the requirement for all objects to be defined

up front. So if you know you want an object to break into pieces you have to define the

fragments first - or at least the number of fractured pieces that are going to be needed. I

went along to a SESI presentation of Houdini 9 DOPs in LA last summer and I remember being

told that DOPs forbids having a new DOPs object come into existence during the simulation

but I don't buy that it's impossible to make it work that way. It would be sooooo useful

to be able to 'spawn' a new object procedurally given a trigger event. If we can get that

functionality then we can make objects shatter without having to pre-fracture them which

would be a huge bonus. I'm sure there would be plenty of other uses too.

John Hughes

This has always been possible, though not necessarily as elegantly as I might hope:

http://www.sidefx.com/index.php?option=com...opic&t=8037

Mark

Share this post


Link to post
Share on other sites
...

it would be really helpful to have many more pre-built, off-the-shelf, well-documented shaders for new users.

Instead of yet more specific single-purpose shaders I'd love to see a proper ubershader with a heap of bells and whistles (that can be turned on or off).

A bit like the default all-purpose materials in many other packages. Most materials are not burlap or eyeballs or leaves :)

When in a hurry it would lower the initial setup time cost for quick projects.

eetu.

Share this post


Link to post
Share on other sites

Hi all

1. There was a time when artist from other application tried to switch to houdini n said the GUI is very complicated and so the sidefx guys have reworked the GUI

so it can be comfortable for everybody wich is very appreciated..But i guess they have made it very heavy and it takes a lot of time to response.

I really love the new look of houdini but they should do something to make it work more faster or some option for switching back to older GUI...

2. Integrating Python will be very helpfull

3. open up ur arms... Try and built better integration to other application..I've read people asking to integrate 2D n 3D tracking or improving modelling to modo or zbrush level..but this is not wise. Instead they should come up with solutions that helps us use different packages in pipeline and not to worry how we are going to in corporate these stuffs in houdini..

4. I loved the new Fluid System but its damn slow..even volumetric rendering takes ages to render...

5. mental Ray is one of the most used and appreciated renderer...do somehin abt it..

6. a long time ago Alise|Wavefront had worked out with ILM to integrate their Ocean System in maya as maya's Ocean Shader for perfect storm and then was added to maya4.5..I'm sure sidefx could do somethin similler with other studios(ie Digital Domain) or convince them to release their tools as pluggins for Houdini..

Share this post


Link to post
Share on other sites

There's only one thing for me...

Better documentation.

Here's 10 things I'd LOVE to see:

1) Stability. It just needs to work! There have been numerous occasions where it's hung or caused a crash.

2) The search is REALLY lacking. For example, do a search for "lsystem". You'd think it'd return the page for the LSystem SOP wouldn't you? Nope! Try it if you don't believe me. Just imagine what else you must be missing if it can't even find that.

3) There's no index. If the search isn't reliable, then an index is essential.

4) Missing files. I've had numerous occasions where there's a missing page and it can't find a file it's looking for.

5) The Load / Launch buttons sometimes don't work or cause a crash. I've had missing files here too.

6) Make it standalone. There are times where it'd be great to be able to view Houdini Help without having to have Houdini open.

7) The auto layout feature drives me nuts! Having the contents/search panel move around when I'm trying to resize the window to get a better look at the docs can get quite frustrating.

8) More extensive documentation regarding the "hidden" features, such as using "op:" to prefix a node to use its path as a filename. These things are CRUCIAL to know about when learning Houdini. There should be a page with all these tips and tricks on it. Again, just try searching for "op:" in the help - it won't accept the colon, or even just "op". For such a syntax heavy application as Houdini, it should definitely support special characters in the search field.

9) When I do a search, I'd like to be able to sort the results in terms of relevance, topic area, or just alphabetically. I can't do that at the moment. If I get 500 results when searching for something, I generally only want to see the results in, say, SOPs for example.

10) This is the big request! Please ditch the whole help webserver architecture and put it all together as a single CHM file. I've been told by people that CHM is windows only format. WRONG! Take a look at XSI's documentation. They use the same CHM file on both windows and linux. They have a program called "hh.exe" which displays CHM files identically to that in windows. It's fast, the search works, it allows sorting search results (althought admittedly not by relevance), it's stable, and you don't need Houdini to be running at the same time. Over half of the requests I've mentioned would be solved just by using this!

Documentation is key to people adopting a package and using it in an efficient manner during production. If the help file only serves to annoy and frustrate, that's a huge block you're placing in front of people. New users won't care about flashy new features if they never get past the basics.

Pretty please ;-)

Andy

Share this post


Link to post
Share on other sites

Texture based OPs (paint on texture, project an attribute on texture, texture on texture projections etc.) would be great in addition to vertex based ops especially in the cases where you need a high resolution data for sampling from but don't have a possibility to increase for example a density of mesh or so. Not sure how to fit better these textures into existing Houdini pipeline.

Share this post


Link to post
Share on other sites

my wishlist,

1. A fracture system like Blastcode would be really cool.

2. Faster Rendering, PBR is nice, just improve the speed. Especially, Faster GI.

3. Faster Dynamics, like maya Ncloth, an interactive cloth system.

4. More modeling tools, like Bridge tool, (really need that one)

5. Render time options, Like in scatter SOP there could be a viewport count and a seperate rendertime count. You could also have that in the subdivide SOP.

6. One massive shader, with a lot of options, Like the ARCHITECTURAL MATERIAL in Mental Ray. That would be really cool, could make life really easy.

7. Also ability to take procedural maps, like noise map or cellular map in the shaders and not having to do that in VOPS. Like you have readymade 3d maps in Maya or XSI or MAX I would really like a Falloff Map( setting up colors according to Incidence angle, Light angle or other options), a stencil map/Mask map and a Layer Map also a Ramp/Gradient Map like XSI, with ability to distort using other Procedurals.

8. Also faster rendering of subdivision surfaces. Right now it is quite slow.

9. In the documentation, bring back the old tuts that were there in Houdini 8, They are all gone. If some one knows where they are let me know

thats all I can think of right now.

with regards

Rohan Dalvi

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×