Houdini 16 Wishlist

[Skybar]

Not sure if it's been mentioned before, but the thing from Fusion where you can hold shift and drag a node off connections. We can already drop a node wherever on a wire, but to get it free again we have to shake it. Simply holding shift (or another button) would be awesome. This isn't already in some setting anywhere is it?

[rich_lord]

I need a really quick way to go to a different frame, but not re-evaluate everything. This is pretty important when doing alot of key framing, as I often need values to match at different frames. Maya does this with the middle click on the time-line. Super fast. I'd love to see this brought into Houdini.

I know I can turn off updating with the little widget at the bottom right. But its too slow for an operation I do many times every minute.

[cojoMan]

5 minutes ago, rich_lord said:

I need a really quick way to go to a different frame, but not re-evaluate everything. This is pretty important when doing alot of key framing, as I often need values to match at different frames. Maya does this with the middle click on the time-line. Super fast. I'd love to see this brought into Houdini.

I know I can turn off updating with the little widget at the bottom right. But its too slow for an operation I do many times every minute.

so set a shortcut for it .. I have "auto update" on alt+a, and "manual" to alt+shift+a - and no complaints here.

[rich_lord]

Thanks for the tip, but I haven't found a short cut that i'm comfortable with. I have to switch modes constantly when I'm animating so I'd have to remember the state of this toggle. I'm probably a bit dense, but this is difficult to track when I'm animating very fast. Particularly for character animation where I have many channels in play at once. It's not as intuitive as the way I'm used to from maya, which is explicit and mode-less. 

 

[pezetko]

4 hours ago, rich_lord said:

I need a really quick way to go to a different frame, but not re-evaluate everything. This is pretty important when doing alot of key framing, as I often need values to match at different frames. Maya does this with the middle click on the time-line. Super fast. I'd love to see this brought into Houdini.

I know I can turn off updating with the little widget at the bottom right. But its too slow for an operation I do many times every minute.

 

Middle click drag works without any update for objects.

Tricky part is that you have to have object you don't want to update selected before you do middle mouse drag in the timeline. So if you don't want to update whole scene, select all objects then MMB drag in timeline, you should see orange line from old frame to new frame. I think it was designed for keying same pose of selected object on different frames.

For DOPs you can disable just simulations (dop networks) in menu or LMB click on that little brain icon in the right bottom corner.

 

[rich_lord]

Holy crap! You are right pezetko!!!! 

I totally missed that nuance. Only took 2 years to figure it out! This is exactly what I want. 

Thank you very much.

[pxBomber]

4 hours ago, rich_lord said:

I need a really quick way to go to a different frame, but not re-evaluate everything. This is pretty important when doing alot of key framing, as I often need values to match at different frames. Maya does this with the middle click on the time-line. Super fast. I'd love to see this brought into Houdini.

I know I can turn off updating with the little widget at the bottom right. But its too slow for an operation I do many times every minute.

+1 for this

2) Also support for multi-gpu workstations will be very beneficial.

3) And just more options where to use that OpenCL brutal power.

4) And I found one bug/feature super irritating: when holding spacebar/alt and RMB in the viewport there is still a good chance for the popup menu "VIEW(Space to overlay)" to show up.... It really bothers me a lot - even worst is the fact that items in this menu reacting to RMB Release - so I accidentally "Homing all, Homing selected, etc" quite often.

- I just prefer some assurance when I'm using a tool for  cam.dolly  it doesn't depend on the length of stroke I did with my mouse. 

[Guest tar]

19 minutes ago, pxBomber said:

3) And just more options where to use that OpenCL brutal power.

IIRC sidefx want to hear where OpenCL is being slowed down in real world situations. i.e. collisions, then they can concentrate resources to optimise that part.  Even when OpenCL is run on the CPU it can be way more efficient than C++ code.

[sebkaine]

Font SOP > Horizontal Alignement add the Justify Text Option.

[Guest tar]

Be able to navigate all UI panels, i.e. shelf, parameters, Geometry Spreadsheet etc, similar to the Network view. Makes working with a tablet much better.

[pxBomber]

 

On 5/17/2016 at 1:07 AM, marty said:

IIRC sidefx want to hear where OpenCL is being slowed down in real world situations. i.e. collisions, then they can concentrate resources to optimise that part.  Even when OpenCL is run on the CPU it can be way more efficient than C++ code.

Hi,

yes on cpu it can be a little bit faster but on 4xTitanX or upcoming titanX80 it should be complete diferent story. Real world scenarios: FLIP sim even without collisions - GPU is slower in smaller scales (up to 3mil particles) and when it starting to payoff is already to slow to sit and watch it cooking - you basically have to cook it over night or on farm witch is CPU based in most cases anyway. The only way how to take advantage of GPU in FLIP is to adjust voxelSize/partCount ratio away from 1/8 witch is optimal in most scenarios. There is probably to many memCopying in the process so you don't see any real advantage until it's so slow that you don't care anymore.

 

edit: Sorry I mean not completely without collisions but with just a basic coll.container 

Edited May 20, 2016 by pxBomber

[paxsonsa]

My Wishlist:

Instancing:

Instancing with CVEX and Mantra is great but when I have a rig that is animated and I need to control the animation through CHOPs the instancing still needs a file to source in order to randomize frames but if I need to speed up or slow down the animation I am fixed to whatever geo files are on disk.

  Possible Solution:

  - Copy SOP has a compute at render button or better template computing for faster viewport display with instance numbers past 2 mil.

  - Allow CVEX to alter the geometry post mantra instance at render time to allow me to blend my animation based on the instance points attributes. Right now i can only alter the transform matrix but it would be great to have that option even though I imagine that would break the efficiency

 

VEX Wrangles and Editors:

- variable highlighting ( having a cursor in a variable highlights it uses in the rest of the code)

 

Edited May 20, 2016 by paxsonsa

[edward]

Sounds like agent primitives would help there?

[abvfx]

Forgive my ignorance but has there been anyone mention or suggestion of VOP/VEX textures to show in the viewport. 

I remember some years ago it was said to be difficult and when it was removed as a shading mode pre v9 is was unstable. I have not heard much on the topic since then. Seems like viewport feedback would be amazing.

Edited May 21, 2016 by abvfx

[bunker]

I have a simple one:

Mplay should have a timeline with the real frame numbers, not just starting at "1"
eg: sequence 1001-1200, mplay timeline is 1-200, that's wrong.
it's even more of a pain if you start having sequences like this: 1280-1590. How do I get to the frame 1344 directly?
 

Edited January 24, 2017 by bunker

[anim]

On 5/20/2016 at 0:59 PM, paxsonsa said:

...Instancing with CVEX and Mantra is great but when I have a rig that is animated and I need to control the animation through CHOPs the instancing still needs a file to source in order to randomize frames but if I need to speed up or slow down the animation I am fixed to whatever geo files are on disk....

currently you can use Packed Alembic Primitives with varying abcframe intrinsic attribute to control exactly which time from abc file each instance uses, since mantra will directly sample abc file at that time it as well interpolates between frames, example here: 

 

[sebkaine]

Be able to render images with the ROP OpenGL without a Houdini Indie Watermark like In Houdini 14.

Edited May 23, 2016 by sebkaine
foo

[gpapaioa]

Small scale FLIP shelve with various tools (surface tension, crowns, elasticity, etc)

[paxsonsa]

On 5/23/2016 at 10:51 PM, anim said:

currently you can use Packed Alembic Primitives with varying abcframe intrinsic attribute to control exactly which time from abc file each instance uses, since mantra will directly sample abc file at that time it as well interpolates between frames, example here: 

Very cool! Though I still think have geo manipulation at instance time would be helpful, though....it would prob ruin the efficiency anyways

 

[anim]

Deffinitely, CVEX control over delayed load procedural geometry, and being able to chain/nest procedurals and cvex shaders at render time is really needed, but in meantime we have to manage without