Hi jumaku.

Just a small question for you,

Why would you think ‘tags’ would be a quicker/faster way for Animation and Modeling?

Houdini just isn't really build for ‘conventional modeling’. So why compare
And for Animation there is a wonderfull world called Chops. Its basically the most advanced/flexible/overlooked animation tool in the world.


I'm sorry to say this but ‘tags’ like in Cinema just doesn't make sense in Houdini, at all.
The problem I think (no disrespect) is that you spent way to much time in the tree view, like if it is the same as in Cinema/Maya etc. But it is really not. So close the treeview and open your nodegraph, and watch yourself fall in love.

And please quit with the UI for ‘old’ users thingie. It's quite offensive.
Houdini is ITS interface. People get upset if someone would touch that.

The only thing I want Sesi to focus onfor H13 is SPEED, Multitread the sht out of Houdini.
Lightcach would be awesome too, maybe even some GPU accelerated stuff in PBR.

just my 2Cnts

I haven't read everything here, but I agree some areas of Houdini could really use some improvements. For me personally that's UVs, but I'm sure there are also other areas. The main issue with UVs is that a lot of the projections are static instead of procedural. Pretty much the only projections that can be used procedurally are the UVUnwrap and UVPelt tools, for all the others the projection is static and you need to wrestle your geometry into place to get a good projection…
Also things like UV packing can only be done with UVUnwrap, and to do it you pretty much need to wrestle your geometry again… and then you still only have a mediocre/unoptimized packing procedure.

Now I don't mind wrestling around with my geometry, but I can't help but feel these things take too much effort (compared to other software) and that it's far from fast, efficient and often still gives only mediocre results compared to what other software can do with the click of a button.

I'd also like to see more options for group selections, in other software it's quite easy to select things like edge loops for example while Houdini has no procedural tools to do such a thing (at least not afaik).

The only things Houdini needs are: 1) Rendering on GPU, 2) Keep improving simulations on GPU.

This software is mostly for creating pipeline. I thing that it is the best software on which studio's pipeline can be created. It is very good for VFX, procedural modeling, animation, and rendering. It is not texturing and sculpting software. It is very funny to see people, which compare Houdini with modo, zbrush, and other tools.

Rendering on GPU is the reality for these days. Fluids are already computes on gpu, so the next and the most important step is RENDERING ON GPU!

GPU rendering has alot of limitations.
especially GPU ram, its just too little

+1 for more if unwrapping options. Would love to see lsmc unwrapping algorithm as an unwrapping method. Also a few improvements to attribTransfer and polyreduce to better handle property discintinuities

Did I forget to mention? SPEED!!!!!!!

hy all!
my wishlist is:
- ‘opencl’ toggle to gas calculate
- a faster gas field vop 8)
- improved code editor ( auto-indent and auto-completion )
- group attributes if possible
:wink:

skogen
hy all!

- improved code editor ( auto-indent and auto-completion )

:wink:

A python debugger in houdini would be great too !!

**-x NPR x-**

qbick
GPU rendering has alot of limitations.
especially GPU ram, its just too little

Aye. It's also a lot slower in practice, including for most simulations, since they've only got a tiny amount of fast memory (a few MB at best). The big chunk of memory they have is extremely slow, there's no prefetching unless they've changed their architecture, and the clock speed is much slower than on CPUs. The way GPU companies claim “ZOMG 2000000x SPEEDUP!!!” is by writing crap CPU code to compare against their optimized GPU code.

I found this out the hard way where I used to work; we spent months trying to get a GPU implementation as fast as a multi-threaded, vectorized implementation for CPU that only took a week to write, and the best we managed was 2x slower than a Core i7 on the best GPU available at the time.

HOUDINI 13 = scuplting and 3d texturing abbilities + planar/3d tracker + HOT + material library + motion capture library

1. speed.
The goal should be real-time & no lag, for everything from resizing a pane to rendering. Speed reduces costs and increases the number of refinements work can go through per day… its so important.

2. spare us the death of a thousand cuts.
Fix all the little things that are dysfunctional or never passed the prototype stage. Like the painful COPs performance, laggy and inaccurate COPs viewport handles/guides, laggy UI, the default near-degenerate triangulation on new primitives, dysfunctional polyextrude inset on concave polys, missing group output options for nodes that create new geometry, mossing local mode for transform handles, tiny function field in the channel editor, mplay not inheriting variables, etc, etc… All the small annoying things, I'm sure there's a huge list of them in the support database by now.

3. keep it clean, keep it portable.
keep the $HIP workflow, I know its hugely unpopular these-days but its one of Houdini's great advantages; you get a zero-maintenance pipeline for free. Also allow Houdini/mantra to be run from its own directory on windows.

cpb
1. speed.
The goal should be real-time & no lag, for everything from resizing a pane to rendering. Speed reduces costs and increases the number of refinements work can go through per day… its so important.

2. spare us the death of a thousand cuts.
Fix all the little things that are dysfunctional or never passed the prototype stage. Like the painful COPs performance, laggy and inaccurate COPs viewport handles/guides, laggy UI, the default near-degenerate triangulation on new primitives, dysfunctional polyextrude inset on concave polys, missing group output options for nodes that create new geometry, mossing local mode for transform handles, tiny function field in the channel editor, mplay not inheriting variables, etc, etc… All the small annoying things, I'm sure there's a huge list of them in the support database by now.

3. keep it clean, keep it portable.
keep the $HIP workflow, I know its hugely unpopular these-days but its one of Houdini's great advantages; you get a zero-maintenance pipeline for free. Also allow Houdini/mantra to be run from its own directory on windows.

I've never heard of a real time production renderer.

Great feedback! By all means, please feel free to continue with this discussion of features you'd like to see in Houdini, but if you'd like to actually have your BUG or RFE recorded and tracked in our database, please take the time to log it here.

http://www.sidefx.com/index.php?option=com_forum&Itemid=172&page=viewtopic&t=25347 [sidefx.com]

Thanks, and please keep those wishes coming!

Fantastic you guys are supporting this.

chrism
…but if you'd like to actually have your BUG or RFE recorded and tracked in our database, please take the time to log it here.

I always spam support long before I post on forums…

Delight0092
Fantastic you guys are supporting this.

agree . that sounds pretty great indeed !

–

here is another tiny dot :

- wd be great to have an Image Manager .
.. which manages all the images ( and sequences ) loaded within a scene .

ive been using LightWave for a decade and this little but useful feature is one that i miss ( esp when the project has a lot of references and textures to jump from and to ) .

i hope the attached pic could look pretty self-explanatory about its usefulness .

oc , other additional / alternated options wd make it fit perfectly into houdini's enviro .. IMHO .



.cheers

cpb
2. spare us the death of a thousand cuts.
Fix all the little things that are dysfunctional or never passed the prototype stage. Like the painful COPs performance, laggy and inaccurate COPs viewport handles/guides, laggy UI, the default near-degenerate triangulation on new primitives, dysfunctional polyextrude inset on concave polys, missing group output options for nodes that create new geometry, mossing local mode for transform handles, tiny function field in the channel editor, mplay not inheriting variables, etc, etc… All the small annoying things, I'm sure there's a huge list of them in the support database by now.

Chips all in on this.
There are areas suffering from some serious neglect. I think I would be okay with zero new features in H13 if instead the existing stuff got patched up. Things that have always worked fine are starting to exhibit buggy behavior. Basic things like hooking up nodes. Modeling and viewport interaction in particular are kind of painful right now. And yep I've submitted the bug reports and RFEs :]

threse days i load massive simed files in houdini to render use PBR,it,s very fast, 5000 agents use about 16 miniutes. but most of the time are coseted on copy sop.
so i wish the copy sop multithreaded.

PBR is fast untill you would like to render an interior scene