Houdini 10 Wish List

[andrewlowell]

I tried that with a Wave CHOP and it seems to work.

Offhand, I don't see a sample rate limitation. The data depth is somewhat problematic since there's no architecture for allowing the user to specify audio save options like it is for images.

EDIT: Oops, looks like audiere is the limitation

ah, ha .. the wav reads in at full bit-rate in 9.1! writes that out correctly in 9.1 also,

But import and export is still limited to 16bit on win at least, but o well big improvement. I think last time I tested import was for 48k video audio on Houdini 8.2 or whatever.

- import 32bit wave types

- .wav export bit depth / sample rate dialog

Edited January 23, 2008 by andrewlowell

[yoann]

I'm really looking for good hair stylling tool and a good way to make NPR rendering.

Maybe something like the really cool freestyle npr renderer.

http://freestyle.sourceforge.net/index.php

souce http://prdownloads.sourceforge.net/freesty...ar.bz2?download

Edited January 23, 2008 by yoann

[edward]

Maybe something like the really cool freestyle npr renderer.

There's always SOHO ...

[hoknamahn]

Don't get me wrong, I would LOVE glsl/cg shaders hardware rendering for mostly for previs purposes; but I wonder which development would take us further: hlsl vops or VEX rendering of texture swatches? I'd think the latter since its the most applicable for all other phases after the previs stage.

Arhhh remember you said about GLSL? Actually my dream is a VEX->GLSL compiler not only GLSL authoring in VOPs.

I would not hesitate in believing that if you show commit an animated feature to be rendered in Mantra, SESI would step up to the plate and do something like this. IMHO this is such an uncommon use of Mantra that I'd rather have more globally useful tools. Perhaps reusing IPR caches for left/right eyes can be a good basis for stereoscopic optimization?

Sure this is not the feature No 1 and some optimizations could be done in currently available versions. But stereoscopic movies become more popular and this feature could be useful in the nearest future.

From what I hear, Intel are looking at 80-core CPUs in near-so-distant future - possibly treading on the GPU/Cell territory.

I agree this is needed.. (not for me,though).. In fact I was certain SESI insinuated that one of these solutions was coming for 9.1; but apparently not.

They employ the people who have a good understanding of a theory of raytracing and experience in development of raytracers. Possibly they do R&D of realtime raytracing for next gen hardware.

[Jason]

Alright; so.. anyone have any off-the-wall ideas? All our ideas so far are so ... practical.

[andrewlowell]

Alright; so.. anyone have any off-the-wall ideas? All our ideas so far are so ... practical.

Well, I'm full of these ideas ...

a. LADSPA CHOP .. takes in and processes incoming data with the specified LADSPA plugin (or maybe VSTi for windows). Supports automation parameters with named channels in the second input.

b. Stereoscopic preview mode in viewport

c. Integrated render-farm management and/or asset management similar to autodesk backburner and vault.

d. Point attribute detection on sound occlusion objects to adjust various things like hardness, filter amount, etc

e. particle-parenting and ik pops, to form hirarchies and IK relationships with particle transformations

[michael]

c. Integrated render-farm management and/or asset management similar to autodesk backburner and vault.

9.1 will poop out stuff to Alfred and [something else I can't remember]...

and asset management is far too studio specific...SESI would end up building something that no one would use...

python can do all this for you right now...

[3__]

Alright; so.. anyone have any off-the-wall ideas? All our ideas so far are so ... practical.

yep replace bloody python with higher level assembly (HLA).

then work on rewriting the rest of the app in assembly.

[digitallysane]

I'll stay in the practical realm and add: faster raytracing (not only PBR).

With H9, mantra's raytracing became actually usable (as opposed to H8). Now make it fast and competitive.

Dragos

[sibarrick]

How about a VOP like environment for making HDK/Python nodes.

or

Zbrush style modelling in mplay that can be exported back into sops via a sculpt node and displacement maps auto generated into cops.

or

An AI context for doing massive style sims...

or

Some simple nodes for linking into the GPU to make accelerated sops... (no idea how that might even work, I'm looking at using opengl to "render" weights for an upgrade to my Warp sops and that at present is best done as a standalone app, I can see the potential is there to tie GPU stuff directly into houdini for more general access for such things)

Edited January 25, 2008 by sibarrick

[Jason]

*) Visualization of DOPs Execution Order: DOPS confuses me when it comes down to understanding (a deep and arcane lore) which order these solvers and microsolvers get executed in,... which seems to be important information if you are trying to perform some more advanced things.. such as advection custom fields along velocity fields etc. So I was thinking of, say, a view of a continuously updated dry-run of the DOP network, similar to what the Performance Monitor outputs.

[hoknamahn]

What really confusing me is a bunch of things like:

1. Duplicate(Utility) / Duplicate(Manipulate)

2. Delete(Modify) / Delete(Attribute)

3. ...

Guys we have almost catch up Maya in illogicality (Maya has the same problems like Paint Attribute / Paint nCloth Attribute / Paint Whatever Else Attribute instead of one generic paint tool like Paint SOP). Is it _really_ necessary? I like the old workflow. Delete is delete and Blast is blast. And do you think that it is good to have for example Cloth object available in tab menu when you are in SOP context and Tool Options flag is set to Create In Context. I believe it would be better to hide and forget about all options which are outside of a context.

Edited January 26, 2008 by hoknamahn

[MADjestic]

- I'd really love to see more CHOPs integration into other contexts of Houdini,in H10.

We've got CHOPs and Channel Editor\View. Why not making those a single entity? E.g. you open a Channel editor and can use CHOPs directly in there - no CHOP network\Channel SOP voodoo magic any more! =) Or something like POPs\VOP SOPs do - place a (VOP)CHOP SOP - dig in - do whatever you want with channels - get out - and see your data already affected. Right now it's so much easier to use expressions\custom vops to process data\channels. But there's so much power in CHOPs that should be used more!

What really confusing me is a bunch of things like:

1. Duplicate(Utility) / Duplicate(Manipulate)

2. Delete(Modify) / Delete(Attribute)

3. ...

Guys we have almost catch up Maya in illogicality (Maya has the same problems like Paint Attribute / Paint nCloth Attribute / Paint Whatever Else Attribute instead of one generic paint tool like Paint SOP). Is it _really_ necessary? I like the old workflow. Delete is delete and Blast is blast. And do you think that it is good to have for example Cloth object available in tab menu when you are in SOP context and Tool Options flag is set to Create In Context. I believe it would be better to hide and forget about all options which are outside of a context.

Personally I so rarely use viewport for the aformentioned OPs that it was a surprise... but yeah, now when looking at it - it looks strange. Is there a special reason for those 'pairs' ?

--------------------------

I also like the sibarrick's suggestion about AI context.

Keep the awesome work, SESI!

P.S. Oh, yeah - I know this is another 'blue dream' - maybe more appropriate for H>11? But what about more OpenGL harware shading\rendering support? Modern hardware shaders are starting to look very sexy - why not using those?

[andrewlowell]

- I'd really love to see more CHOPs integration into other contexts of Houdini,in H10.

We've got CHOPs and Channel Editor\View. Why not making those a single entity? E.g. you open a Channel editor and can use CHOPs directly in there - no CHOP network\Channel SOP voodoo magic any more! =) Or something like POPs\VOP SOPs do - place a (VOP)CHOP SOP - dig in - do whatever you want with channels - get out - and see your data already affected. Right now it's so much easier to use expressions\custom vops to process data\channels. But there's so much power in CHOPs that should be used more!

I think if this was done correctly it could certainly help Houdini out in the procedural animation department, especially if stuff like block-style editing was incorporated. However, they'd really have to be careful not to mess up the proceduralism inherent in a true CHOPs network, compared to the editable yet non-procedural curve/keyframe approach.

[3__]

sticky attribs for sop vops.

testing a condition with compare, then setting an attrib with if

will work, but if the test fails the attribute will revert back to

its input value. it would be very handy to have an override that

leaves the attribute as it was from the previous round (frame,

sample or whatever).

or put a frame input on import attribute.

-cpb

[digitallysane]

Some improvements and additions to existing editors or panes could tremendously improve the animation workflow:

1. Long overdue: let the user change the color of channels in the Curve Editor; also let the user specify a single color for all the channels in a channel group; give the option that those colors overly the whole timeline in the Spreadsheet view and the Dopesheet view

2. Allow the Parameter Spreadsheet to show groups of parameters chosen by the user; allow it to show channel groups; allow the user to color the rows (and even the columns). (there was some discussion on SESI's forums about this, http://www.sidefx.com/index.php?option=com...ht=spreadsheet), but I'm more for a customizable spreadsheet.

Both 1. and 2. are actually intended to improve spreadsheets as animation interface. In principle spreadsheets can be a wonderful animation tool but that depends heavily on the visual layout, because without good layout and coloring, they can be very distracting and hard to read. I recently watched this interview http://www.navone.org/blogger/2006/06/infa...-interview.html and liked the spreadsheet UI quite a lot (of course I'm not sure i understood how it works). Houdini's Channel Spreadsheet could be very useful if channel groups could be colored (along the whole timeline), now it's too hard to read. Also, the Parameter Spreadsheet is a wonderful tool for having a global view at the parameters you're interested in at the current frame, but again, it needs the capability to put params of choice on multiple rows, colored in a meaningful way. If I put a character asset in the spreadsheet now, I'll get a mile long scrolling line of parameters, and this is not useful.

Dragos

[malexander]

Drop the Cell thingie, is a waste of time and resources, and concentrate instead to where the industry is going.

Even if you never own a Cell, there's still a lot of good things coming out of this project that'll benefit everyone. Even some things that have been suggested here

[shikung444]

I request:

1. Multiple diffuse bounces for global illumination in mantra micropolygon renderer (without having to use PBR). This is a feature I often use in final gather in mental ray. I'd like to be able to use that in mantra without the added overhead of PBR setup and rendertime.

2. Soft Groups. I request, (in 3ds max terms) a proper soft selection that can be passed into any deformer (or anything else for that matter). In Houdini terms, this would be like a combination of the function of groups and attributes, meaning you could make a "soft group". Right now this function can be achieved with metaballs and attribute transfer, but the workflow is clumsy and longer than it has to be. In 3ds max, you can use any other mesh object as a selection (which can be done with Houdini group sop as well), but when soft selection is turned on, there is a smooth transition as the objects come close together. In Houdini, the soft selection suddenly happens once the first point is in the group. Just think, this would be like having procedural soft peak functionality in any sop!

3. Cusp normals without changing point or vertex count, and without actually separating the faces. In 3ds max there's a simple autosmooth button that will smooth or show a hard edge based on the degree of the polygon faces. Maya has a similar function. Other major software can do it, why can't we?

4. Implement something like Frantic Films' Krakatoa in Houdini.

5. Update the render region function. Something about it just feels sluggish and a bit unreliable to use. Just compare it with XSI's render region. (Note: I've only used Houdini on Windows 32 and 64, so if the Linux version is stable and snappy, then just fix the windows version.)

6. Update the shader ball for vop shader development to a real time customizable global illumination environment with optimizations for the shader ball geometry so that it is real time. Also make it somewhere that we can preview a displacement shader on a model and get a visual result whose scale resembles what we might get on our scene objects.

7. A full blown AI node context well integrated with chops.

8. Real time Open GL based Volumetric previewing and rendering.

9. A surfacing operator which will fit and weld a series of beautiful quad meshes onto arbitrary geometry.

10. Create, not only a full bodypaint/photoshop style 3d painting system which will paint /clone on existing meshes, but something that can paint quad polygon strips into existence that autoweld themselves intelligently within a given threshold complete with updating uv information and paint strokes or (anything else) automatically applied, directly in 3d space. I'm talking about something that greatly outshines maya's paint fx. Something like corel painter meets polyboost, fusion, and massive

Thinking of Number 10 just made me think of Number 11. I hope you all don't mind.

11. Houdini users could create this themselves if there was a node based tool or function creation mode, that would control the program's response to input devices. We should be able to do this without scripting. For example, there would be a pen pressure node, a brush shape node, a viewport transformation matrix node, a keyboard node, a "current selection" node, a "Create op" node, a window node, a custom ui object node, that, when linked together, control low level functions of the software's interface behavior. This way users could create custom viewport modes, such as "rotate around selected points", or even custom brush behaviors, such as a "create n-sided polygon, weld together in this intelligent way, create and paint on bitmap texture, and set uv coordinate, while rotating the viewport based on a 3dconnexion spacemouse" brush. They could also create the much needed ramp vop they've asked about for so long.

In other words, open up the interface. Big effects houses strip down maya and even build their own interface out of mel. Would be cool to be able to do in Houdini with nodes though.

Alright, I think I'm done

[malexander]

cops

clean viewport refresh.

no blinking, flashing, rescaling, transforming, buckets, etc.all i

Buckets off:

Edit->Preferences->Compositing, Cooking page, "Cooking Feedback" -> None (or Progress Bar)

Resolution reduction off:

Edit->Preferences->Compositing, Cooking page, "Use Fast Interactive Cooking" -> off

You can also change this as required in a less drastic way using the Fast Interactive Cooking button (racecar icon, hotkey "f").

[hoknamahn]

Hey guys what about offscreen flipbooks?