Rumors and thoughts on Houdini 20?

I spent my Sunday morning sending 7 RFEs/bug reports just for the Geometry Spreadsheets ALONE, and I'm not even done about it.
Then I realized that one of them is actually a duplicate of a RFE I had already sent almost 4 years ago.

The problem with SideFX not perceiving that UI/UX improvements are as well important (and thus not really committing to it) might really well be in the lack of RFEs submitted and their volume compared to other areas.
So wouldn't be fair here to blame SideFX, and I'm far from doing it.

But here's the catch about this possible scarcity of RFEs and bug reports about UI/UX:

1. the user has a time budget that can allocate for writing to the Support and I can imagine he would spend said time for stuff that directly impact the work he's currently doing. I too give priority to ask for help for my very short term issues in a project.
   
2. UI/UX issues don't impact you in the short term indeed. They are not perceived as an immediate danger. But in the long term they affect every workflow and they add up easily.
   
3. UI/UX issues are often hidden or not perceived (and thus not reported to the Support) because sometimes they don't really break anything. They might just slow you down but you don't know it.
   
4. UI/UX issues are harder to communicate and often you give up in reporting them. This affect even more a big portion of the user-base that is not native english speaker.
   Indeed it's easier to report that a specific node or parameter is broken, because you can easily build an example file to showcase the problem.
   It's harder to explain how your workflow is affected and how could it be made easier and faster by adopting some UI/UX solution: you are speaking about more intangible stuff like wasted mouse clicks, wasted mouse movements, readability, etc.
   

The above considerations lead me to 2 conclusions:
1) we as users should really put more effort in filling RFEs and bug reports for UI/UX. This is understandable harder and more time consuming than writing about more technical issues.
2) SideFX is very welcome to be more pro-active about UI and UX improvements, without wrongfully believing that there is no interest from us because of the lack of RFE and bug reports.

Gaalets
Houdini's interface is frankly bad in my opinion. It seems to me that Houdini will remain in a narrow niche of complex simulations. Because simple simulations will be done in familiar, convenient programs like Blender, which will be overgrown with them. There are already simply amazing sets of nodes for modeling and animation. And new people just stay in a very convenient blender. And Houdini's audience is collapsing more and more. And this is a big mistake from the point of view of development, imho. The niche is narrow, the price will have to be raised, which will lead to an even greater collapse of the audience.

Blender rules the world!ohyeah

Gaalets
Houdini's interface is frankly bad in my opinion.

Edit: I did type a response to this, but then I remembered the Golden Rule of the Internet: Don't Feed the Trolls.

I hate to interrupt the existing blender versus everything conversation but here is my H20 expectation list:

SOP Flip presets for bubbles, ripples, waterfalls, melting, splashes, sinking, drops
Updated whitewater and new foam presets
Explosions pyro presets for volcano eruption, magic FX, serial explosions, cannon fire
Flame pyro presets for realistic and magic flames, flamethrower, fire sparks, and pyro spreading
Smoke presets for smoke trails, frost, sand storm, mist
MTLX shaders for water and pyro
SOP POP rebuild
Weather FX: Snow, Rain
Vellum wet grains
Vellum Muscles system release
KineFX presets for rig building, constraints, weights and motion capturing
KineFX fully functional auto rig for human and quadruped characters
Updated motion clip pose library
Spring physics quick setup from spiral/helix
KineFX vehicle rig tools
More KineFX plus physics simulation stuff
Crowd terrain footsteps system
Library for LOP lighting presets
LOP lens flares shader and self-illuminating lighting improvements
More LOP tools for world building and scattering
LOP and viewport atmosphere tool
City building toolkit
Interactive tree building toolkit
Quadremesh SOP
Feather tools
XPU release
Karma texture baking
New Vulkan viewport and Vulkan HDK

Apart from the marquee features, I'd like to see a lot of polish. In no particular order:

• General:
  A "Create Object Merge from Selected Node" command
  
• Display Options toggles (point marker display, etc.) added to the Hotkey Manager (I know how to do this with Python, I don't want to use Python for this though).
  
• Top-level FOV hotkey control for views and cameras, without messing with HUD sliders. Focal Length is critical when framing a shot, and the controls for it should be as accessible as dollying and panning are.
  
• Separate view undo/redo command with its own undo queue.
  
  SOPs:
  
• Easy mask growth/spread controls
  
• Easy mask edge noise controls
  
  
  Animation:
  
• A non-black Animation Editor background. It's a very harsh look. A usable dark grey would be nice.
  
• A simple Time Offsetting system for animated copies.
  
• Revamped, modern Dopesheet. That thing is not fun to use in its current state.
  
• Mute keyframes in the Graph View. For some reason this command is only available in the Dopesheet.
  
  Rendering/Solaris:
  
• XPU Cryptomattes (hopefully added in a daily build before H20)
  
• Material thumbnail previews in all material editors
  
• A brain-dead Batch Render Queue system. All existing workflows are far too complicated when you just want to queue up a few frame ranges/ROPs. The system should be "Add to Render Queue" on the ROP, and then a Render Queue pane with a "Batch Render Queue" button on it. Not everything needs to be wedged.
  
• Built-in camera switcher/mixer/manager for Solaris. There's a nice Light Mixer to manage all your lights, but nothing for cameras.
  
• A suite of scripts to automate object transition from /obj to Solaris, with material conversions to MaterialX and prim assignments.
  

There's likely a lot more, but I'll leave it at this for now.

my H20 expectation list:
tutorials, tutorials,and tutorials and courses, and especially recent ones, to learn Houdini correctly.
Notably to learn how to use karma...
and maybe an evolution of modeling tools, so that people use houdini for modeling and not just VFX.

1. Viewport fixes, like not automatically switching to local coordinates when using "Hide Other Objects" in SOP context and messing up viewport cam orientation if the object is rotated on the object level. There is a separate function for that which should not be activated only because I want to hide objects.

2. Background image sequences loaded into RAM instead of Vram

3. An HQueue submitter for Karma without having to fiddle with yet another command syntax

4. UI lockable so it would be harder to inadvertently minimize a toolbar

5. Better autosave system - one that can stay on with a limited number of saves(cyclic)and versioning for normal saves

6. Better project management: Remember the last used project, do not remove the current project path from the recent project list when opening a scene from that project without selecting the project in the list first. That's just silly.

7. Solaris Camera functions in the normal object context (i.e. Lookat mode).

8. True animate-able extrude along curve - not deform extrusion by curve post-extrusion.

9. Better Scene management: Multi object renaming, Grouping, finding/selecting by material, creating hierarchies, sorting, structuring. Basically a better scene tree view/editor (preferably like the on XSI had )

(A year ago this list would have been much, much longer even though this one is probably not complete. There was a lot of progress in quite a few areas.)

I really hope for Animation-focused release above else:

- physics-based character animation, Cascadeur [cascadeur.com]-alike.
- better xform gizmos.
- kinefx autorig/presets.

Other:

- rewritten COPs, if possible more Substance Designer alike + Karma baking.

Lots of good ideas. Here's my take: I'd be happy if XPU were to be hardware-agnostic, and not locked in to a singular certain manufacturer.

osong
Lots of good ideas. Here's my take: I'd be happy if XPU were to be hardware-agnostic, and not locked in to a singular certain manufacturer.

From the Houdini Manual:

Supported Hardware
Karma XPU currently only supports CPU and NVidia Optix devices, with Optix being the most optimized code path in XPU. Future versions are expected to have improved CPU performance and support more GPU hardware.

NVidia users will require a minimum driver version of 495.89, and a GPU of Compute Capability 5.0 (Maxwell) or higher

https://www.sidefx.com/docs/houdini/solaris/karma_xpu.html [www.sidefx.com]

osong
Lots of good ideas. Here's my take: I'd be happy if XPU were to be hardware-agnostic, and not locked in to a singular certain manufacturer.

If Renderman XPU development, which is developed intentend to be hardware agnostic, is an indication of anything, we may not see Karma XPU on AMD GPUs for maybe 5 more years? Personally I would like their resources devoted to optimize, implement QOL features and improve performance better than hardware agnosticism and that's coming from somewhere who after decades on Mac decided to build his own PC in order to get the best out of 3D apps. But I get your point and Mac users may be pissed off.

i mean, you do realize that making hardware-agnostic XPU would immediately be the definition of QOL and performance improvement for everyone who happens to not own an Optix GPU?

Developing software is always bound by constraints. Do you delay getting a renderer to a 1.0 status for x years so that it can run on all hardware? or do you get to 1.0 status earlier and target 95% of users? Look at the market, everyone has gone for the latter.

Octane came out in 2015 and still doesn't support AMD, Redshift came out in 2014 and only got early AMD support this year. It's clearly not trivial to get cross platform GPU support.

i didn't say it was, i was writing my wishful thinking and "nice-to-haves"; it is H20 rumors thread after all

Maybe replace wishful with some more thought into why it's so difficult to support everything, and as pointed out, limited Dev
resources Vs target audience. You are talking in the realm of 1% or less of users in your GPU example, QOL benefit to whom?

MTLX shaders for water and pyro

I cant stress this enough, but making some kind of MTLX node for those will be extremely awesome for other renderers as well, imagine redshift with the ocean spectrum or the whiteWater shader.

lewis_T
Maybe replace wishful with some more thought into...

how about no

my initial post verbatim: "I'd be happy if XPU were to be hardware-agnostic"
i did not say "okay everyone drop whatever you do and start working on it NOW"

I just hope our long standing RFE's finds it's way into Houdini.

Two major requests would be.

1. Animation Editor:
KineFX is amazing, but helps very little if animators refuse to work in Houdini because of the messy Animation Editor.
Give us better animation workflows. Look at other industry standards and make it easier to use the Animation Editor.

2. New Sculpt Node for editing. I'm not talking about millions of SubD's here. Just brushes to edit existing meshes. This will help the tweak/blendshape process.
Build on the Attribute Paint to create sculpt brushes. The current Sculpt node is shockingly bad. We need tools that use industry standard keys.
Brushes under one Node that can...
1. LMB Sculpt out
2. Ctrl + LMB sclupt in
3. Shift + LMB to smooth
4. Grab or Move brush to move points in screen space.
5. Ctrl + Grab brush to peak points on their Normals.
6. Pinch brush
7. Ctrl + pinch to expand.
8. Integrated surface or radius falloff filter.

More features about camera and camera animation in solaris are needed,like camera composition guide,camera dof focus distance linked to object,camera morph,camera follow spline,render region.