Houdini 17 Wishlist

[Guest tar]

1 hour ago, McNistor said:

@Alexey VanzhulaWhat do you meant at 11? You can already do that in Houdini if I'm not misunderstanding your request.

It's appears to be totally undiscoverable, weirdly. 

[sub]

GPU Mantra

UV-Tools 

Modelling Tools

 

[ikoon]

"String / Number" input type for Wrangle Parameter Interface. I mean chf/chs hybrid:

 - you could write small expression "@P.z + @emission + 15" and have it included in the compiled vex
 - you could enter just a number, it behaves like "chf", with a slider (also "includes" the number in the compiled vex)

I am hacking this with chs and backticks `` now.

[pxBomber]

cop/img network improvements:

-I would like to have the same flawless experience like in SOPs - when the node stream has a plane it should be a rock-solid truth across all the frames, vopcops etc.   

-automatic mem managnment (maximum from pc), no need to "clear compositing cache" all the time

-much faster .exr/.pic read

-wrangle 

-possibility to see whats cooked and cached in timeline

basically read times, caching and related bugs is the main difference between nuke and h (at least for me).

If I will be sure I always can do my render layers compositions in H without stress it's like a 4000eur price reduction on HFx (basic N lic) and that's something.

 

 

 

 

Edited June 21, 2017 by pxBomber

[animatrix]

2 hours ago, pxBomber said:

cop/img network improvements:

-wrangle

 

 

 

 

I made these long ago but unfortunately they are not CVEX.

http://orbolt.com/asset/animatrix::inlineVexGenerator::1.00

http://orbolt.com/asset/animatrix::inlineVexFilter::1.00

I would also like to see OpenCL COP, multi-threaded cooking/generating thumbnails in the COP network like Nuke.

Any new feature added should IMO be mirrored in every other context where it makes sense. It's a lot of work, that's why unifying all contexts would help with that.

 

[LaidlawFX]

Change the line color as you would change the node color, or be able to change the wire color universal as a start.

If blocks in sops, so programmatic oriented people can read the node chain down. This would be a hybrid of the for each loop where a the ui has a bounding box, and the switch would appear to be at the top of the chain.

[LaidlawFX]

Wrap around text for nodes names that are too long. With a variable cap based in the ui preference.

[iFight4theUser]

Diff Function for the Node Network
This would highlight nodes that have changes, similar to how the performance monitor colors the network.
- This would allow you to compare 2 subnets or tree selections to find what single variable or node has changed.
- This would also be a great feature that could be enabled automatically inside unlocked OTLs, so as you unlock and start making minor changes you can easily find them later. It's always scary opening up another artists hip file and wondering what they have changed inside a very complicated unlocked OTL. 

Drag and Drop Swap
- In Nuke you can Ctrl+Shift drop a node on another and it swaps them out. I keep forgetting you can't do that in Houdini.

[Guest tar]

Colour the bone nodes to the same colour as their viewport bones. Much easier to identify.

[willow wafflebeard]

- have multple output from object subnet. or atleast in object level HDA.
- python support for handles, or maybe a better way of setting up handles.
- a way to extend or install a custom component in the new autorig.
(my concern mainly is to save small portions of my rigs, automating the whole thing doesn't seem to be a houdini way. but i understand that this is the trend)
- seems also that HDA now saves the userdata in its definition. that seems an odd behavior. is it intended? or bug?
- the path+ikspline combo supports 360+ twist. but it seems like the "deform by bones" sop will flip anyways going beyond 360
- "save content as lock" doesn't work if the asset was once had it turned off. ( can't lock it anymore ).
- in the "change operator type", it would be nice to see the "version" namespace.
- im using houdini 16.0.504, linux mint 3.2.7, forgive me if most bugs i said was squashed.

Edited July 9, 2017 by willow wafflebeard

[Alexey Vanzhula]

SOP nodes thumbnails

[michael]

On 09/07/2017 at 2:28 AM, willow wafflebeard said:

- have multple output from object subnet. or atleast in object level HDA.
- python support for handles, or maybe a better way of setting up handles.
- a way to extend or install a custom component in the new autorig.
(my concern mainly is to save small portions of my rigs, automating the whole thing doesn't seem to be a houdini way. but i understand that this is the trend)
- seems also that HDA now saves the userdata in its definition. that seems an odd behavior. is it intended? or bug?
- the path+ikspline combo supports 360+ twist. but it seems like the "deform by bones" sop will flip anyways going beyond 360
- "save content as lock" doesn't work if the asset was once had it turned off. ( can't lock it anymore ).
- in the "change operator type", it would be nice to see the "version" namespace.
- im using houdini 16.0.504, linux mint 3.2.7, forgive me if most bugs i said was squashed.

please submit these as bugs/RFEs
 

[LaidlawFX]

For the end or the beginning of node chains i.e. if they have no parent or no child, have a Network Pane Display option where you can highlight the bottom or top half of the node. Either with a gradient or a circle. This way you can detect dead ends, or the start of chains. Theoretically for instance in a "common style" where OUT is at the end, or you have a file or a primitive node these would be highlighted. This would help say you miss wire something or were debugging and forgot to reconnect a wore on how to connect them again. The alternatives are running a performance monitor test to check for dead leads(assuming all switches are working), pain painstakingly check all connections, or OCD reorganizing so the wiring is clear.

[ikoon]

"Edit multiple Transform SOPs at once"
      - just a "handles feature", doesnt have to be procedural
      - also can be applied to GEOs

Imagine a table, chairs, dishes on the table. This probably wouldn't be procedural, because at first you want to select all chairs (but one), then you select some dishes, then just some spoons.

With two checkboxes (and ctrl/shift hotkeys)
      - apply to all individuals (elements have their own pivots)
      - apply to the "matrix of pivots" (rotate them all, but keep individual R)

So you can:
      - scale just positions (without scaling individuals)
      - scale just individuals (retain their positions)
      - scale all
      - rotate "positions" (without rotating individuals)
      - rotate individuals, keep their positions
      - etc combinations
      - translate all selected with single handle (when inside Geo, then this is not possible)

Pivot swap:
      - pivot in the center of last selected
      - pivot in the center of all

Also align and distribute buttons?

Soft transform with radius from the pivot?

[LaidlawFX]

5 hours ago, ikoon said:

Soft transform with radius from the pivot?

There is a soft transform sop, if that is what you are looking for?

Also I'm not sure if you are looking for a viewport "handle" control for multiple node transforming, but in the network pane you can select several nodes of the same type and edit their parameters together.

[ikoon]

27 minutes ago, LaidlawFX said:

There is a soft transform sop, if that is what you are looking for?

Also I'm not sure if you are looking for a viewport "handle" control for multiple node transforming, but in the network pane you can select several nodes of the same type and edit their parameters together.

Hello Ben, thank you. Sorry for bad explanation, I meant the following scenario, multiple Transform SOPs selected and one Handle to edit them all. Now transform just the cubes proportionally. Then move the sphere and the middle cube. I understand, that it is not that needed for VFX pipeline. I know I can merge them in another Transform. But for motion graphic, setting a scene with handles like this will be useful. I know I am thinking non-procedurally here, but those handled numbers are non-procedural anyways.

 

I believe it could be done with some temporary python/merge/wire/handle/wire-back. Btw when Restart Selecting (hotkey `) is called, for example on PolyBevel, the scene wiring is also temporarily "hacked", unwired, shows errors and after selecting prims and hitting ENTER, the wiring and display tag goes back to PolyBevel and the right numbers are filled in. This is some internal magic of Houdini, which I dont understand, but it is the same type of solution:

[michael]

On 12/07/2017 at 11:05 AM, LaidlawFX said:

For the end or the beginning of node chains i.e. if they have no parent or no child, have a Network Pane Display option where you can highlight the bottom or top half of the node. Either with a gradient or a circle. This way you can detect dead ends, or the start of chains. Theoretically for instance in a "common style" where OUT is at the end, or you have a file or a primitive node these would be highlighted. This would help say you miss wire something or were debugging and forgot to reconnect a wore on how to connect them again. The alternatives are running a performance monitor test to check for dead leads(assuming all switches are working), pain painstakingly check all connections, or OCD reorganizing so the wiring is clear.

this is a nice RFE - please submit it

[Alexey Vanzhula]

Cumulative modeling SOP, like Edit SOP, but for all the main modeling operations: transform, split, loop, bevel, extrude, etc.

[LaidlawFX]

Well looks like my want for HIK compatibility with Autodesk is no more... Funny how things work...

https://www.autodesk.com/content/autodesk-game-middleware

 

[Guest tar]

3 hours ago, LaidlawFX said:

Well looks like my want for HIK compatibility with Autodesk is no more... Funny how things work...

https://www.autodesk.com/content/autodesk-game-middleware

 

AD's incumbent reputation intact!