@hMonkey I agree with almost all points except for 4 (contribution of animations).

As far as I know, Houdini icons are made as .SVG, but the only place Houdini uses the advantage of vector graphics over bitmaps is in the node editor and the way nodes are drawn (e.g., interactive zooming). For the rest of the UI, bitmaps are used and are generated once (during the first start or when you increase the overall UI scale to, for example, 125%).

Maybe that's why you can't interactively zoom parameter pane elements or change how icons are color-coded on the fly?

As for point 4, I agree that the only place for animation in the UI is in the viewport so it isn't distracting, BUT

I count highlighting as animations too, so you have possibilities for making subtle differences if an icon is clicked with the CTRL key. Let’s say you click a keyframe point with CTRL to change from a bezier curve to linear, and the icon provides feedback with some sort of animation, (it's just a quick example).

I bet there could be more use cases for using vector icons instead of generating bitmaps from SVG vectors like it is now. It just give wider landscape.

I wonder if it’s possible to make the whole Houdini UI not as vector, but as 3D overlaid as a HUD? So you could use Houdini nodes to make it behave like you want and then share it with others? But I guess that could be a slow solution?

Anyway, I’m really surprised at how web-based UIs have evolved. For example, node-based ComfyUI for using generative AI models like Stable Diffusion. All this functionality and flexibility in a browser. Amazing! and then i look at Houdinis 15 year old same UI ....

oldteapot7
afaik Hoidini icons are made as .SVG but only place H. using advantage of vector over bitmaps is in node editor and way nodes are drawn (fe. Interactive zooming) for rest of UI are bitmaps that are once generated (during first start or when you increase overall scale of UI to fe. 125%

I think it's just the way Qt works. Your issue is with scaling and size ratio of elements ie the root cause, afaik houdini's ui uses 85dpi as base, which works out to 25.6" monitor with 1080p resolution, but more likely 27" with 81.6dpi... So I'd rather they fix the base scaling and ratio between elements rather than anything else, as common scale is around 109dpi on a 27" 1440p, larger resolutions vary from 93.2dpi for the 1440p on 32" to 137.7 for the 32" 4k, etc etc... so to have the intended "Normal" houdini scale you would use a value of 1.282, I've settled on 1.173 as my default. I guess they could have made all this irrelevant and automated...

oldteapot7
Maybe thats why you cant interactively zoom parameter pane elements or change how icons are color coded on flie?

Not really sure what you mean.

oldteapot7
as for point 4 i agree that only place for animation in UI is in the vieport so it isn't distracting, BUT

i count highligting as animations too, so you have posibilities for makeing subtle differences if icon is clicked with CTRL key. Lates say you click keyframe point with CTRL to change from bazier curve to linear. and icon feedbacks with some sort of animation wich is differen when clicked without CTRL. (its just quick example)

I meant the actual work that you would do, and not the UI

oldteapot7
i wonder if it is possible to make whole Houdini UI not as vector but as 3D that is overlay as HUD...? so you could use Houdini nodes to make it behave like you want and then share with others?

but i guess that could be slow solution?

It's fine the way it is now, it just needs to be slightly improved here and there... evolve the established and woking solutions... except for the shelf (the most obnoxious and useless thing in houdini), one of the previous posts demonstrates one way of how it could be improved and organised...

The whole point of having a vector-based UI is that it’s resolution-independent, and you can forget about DPI and monitor resolution restrictions because it works seamlessly everywhere with antialiasing enabled.

Google uses it in their UI—for example, in Chrome, all the tabs and icons are made with vectors, not bitmaps. And there’s animation, but animation in UI is like compositing/VFX in movies: when it’s done well, you don’t notice it; you just experience a great feeling.

I don’t know much about programming, but Qt sounds like some low-level framework or something. I wonder if the new Vulkan offers any solutions for creating HUD-like UIs? In other words, what’s holding SideFX back from improving their UI?

Houdini is the best 3D app with the worst UI. And I’m not talking about the modular (node-based) approach, which is the only way 3D should be done.

This whole thread made me wonder if it would be possible to have Houdini Engine for Chrome or any other browser, so I could connect nodes and play with parameters using an iPad or Android tablet while lying on the couch, with all the heavy lifting done on a Linux instance in the cloud. That would be awesome! But also, the new UI should be touch-friendly.

By the way, is it possible to run Vulkan on iPads and Android devices? I’m thinking of buying a new tablet and don’t know which platform I should choose.

Side note: Even for programmers who work with text only (code), UI has evolved. Just look at how VS Code looks and behaves. It’s not scary anymore! ah and this font must be good for working with code and it dosnt look like from MS-DOS era like it is in Houdini

oldteapot7
The whole point of having a vector-based UI is that it’s resolution-independent, and you can forget about DPI and monitor resolution restrictions because it works seamlessly everywhere with antialiasing enabled.

I thought this goes without saying nowadays.

As for what tech — I believe everyone just uses HTML/CSS/JS nowadays for UI. Why use anything else? HTML has it all, but idk, maybe there are good reasons to go with something else in Houdini's case.

ruffemotion
Houdini is the best 3D app with the worst UI

That's not true, it's not the worst. IMO, the worst 3D app UI is Zbrush, PC version. Next worst would probably be 3ds Max.

Zbrush new UI for iPad looks good. Mayby one day there will be Houdini for tablets? would be usefull for look dev on good quality OLED displays. Mayby for animating too?

.

ZBrush on the iPad is looking very nice. Must be a pleasure to work with. Everything they did there exemplifies what I think is the best of current high-end UI design.

ruffemotion
the worst 3D app UI is Zbrush, PC version. Next worst would probably be 3ds Max.

Bingo!

ruffemotion
As for what tech — I believe everyone just uses HTML/CSS/JS nowadays for UI. Why use anything else? HTML has it all, but idk, maybe there are good reasons to go with something else in Houdini's case.

That's taking things too far, and that should only happen if they find current solutions limiting. @animatrix 's comment in the https://vimeo.com/381527406/7b864bde55 [vimeo.com] pretty much sums things up (40s), HModeler also deals with the similar issues.

Let's refocus on what's currently available and how it might be fixed and why and see what @ruffemotion comes up with, so we can constructively critique that This should be about reasonable and short term achievable ui/ux "upgrades" and not a complete overhaul imo.

Going off topic is only going to shut this thread down, keep it Houdini centric.

I might be wrong, but the current solution is very limiting for designers; it looks like stacked bitmap boxes. I watched one tutorial and saw that it's possible to add new parameters to the pane editor just by expressing a line of code or something like that. I must say its cool and give great flexibility even for coders that update nodes and also for users. Mayby thats this trade off? Functionality vs control over how it looks?

In other words, Houdini's UI is just a collection of boxes. Where words in code are drawn as bitmaps. I bet that there is reson why it is how it is but dont know it yet mayby easy of creating parametes panes and easy for coding it without carring how it looks?

I could compare it to text and fonts, but without any control over how the text is formatted—no options for spacing, kerning, line spacing, bold, or italic, size! and so on.

Also, icons are designed as "one fits all." I mean, it doesn't matter if an icon is used on the main screen or in the node editor. They are not designed to look good when living close to each another. (groups of icons)

In my opinion, that's why it looks so old and ugly because of these limitations. Between the code (text) and the graphical UI representation, there's no layer for designers, and there are no tools to use for them.

It reminds me of ASCII art, where artist is very limited by old technology and trying their best to create something cool. The same is with Houdini.

I wonder if that Qt framework doesn’t have some ready-made solutions that can give more control over how the UI looks? It can't be that limiting!

This is how I feel about it, though I’m not sure if that's actually how it works under the hood.


.

Houdini is completely customizable. I posted a theme video and you can create a theme there but I agree that it is not straightforward for Design workflows because Houdini has not been 'for Design'. They only just added Design as a skill for the profiles a few months ago.

If it is easier for you to use HTML etc., use the Python Panel and the QtWebViewer. You can control Houdini from that panel so you can build a HTML/Web based interface. That is completely fine and if you need help learning how to use the Python panel or the Houdini webserver capabilities, we can help with that.

There is one basic lesson from form design that shows Eye movement effects from layout of titles and fields. I have used the computation graph as geo approach to communicate how the eye flows from L-R, L-R, L-R instead of straight down. The graph would be green if it read straight down. Like in COPs where the name is at the top. This can quickly prototype new designs if you use this approach. I can instance the name of the node from the subnetwork adjust the lines etc. So feel free to use this approach to re-imagine how ever you want the expressiveness of the experience.

https://uxdesign.cc/text-fields-forms-design-ui-components-series-2b32b2beebd0 [uxdesign.cc]

I have not had time to see how to implement the text placement shift in SOPs. Does anyone know how?

Is it possible to make icons and their states in rive.app and then (render) as overlay HUDs in Houdinis viewport? With easy connecting icon state (fe. Mouse over) to clicked button.

I would like to have simple template for this workflow. So i could replace current icons one by one xD or make UI for something else.

Or mayby there is other simpler sollution for using vector icons and quicly making them to work in Houdini (linking)

I wonder if it will work fast? and is way to go?

Просто оставлю для вдохновения.

In Qt, there are no issues with properly displaying vector elements in any form. The main problem is that Houdini's interface elements are not native Qt elements but part of an old system that is rendered using Qt. Over time, more Qt elements have been integrated into Houdini, but the majority of the UI still relies on the old system. Unlike Houdini, when Autodesk transitioned Maya to Qt, their goal was to convert every UI element to Qt. In my opinion, this was biggest mistake of SideFX in terms of UI, as since 2014, Houdini still isn’t an application that fully uses the Qt UI. That said, the Qt library is vast, and implementing everything you’re suggesting in this discussion is more than feasible without major issues.

Anyway i see hope it is possible to change UI. Current one is more like bones in rig, it is prototype for programmers where quick changes are required. Missing part is skin (using 3D analogy) wich would be editor for designers where you could add better looking. And final stage is texturing, where user could customize it to his liking.

Its like converting from plain text to html code or something. job more for programmer then designer.

I think that new render status and info is good direction and cant wait when whole UI will be converted.

I know nothing about coding but making small UI with animated buttons in rive and connecting it thru python or something to houdini nodes is my new challege

I forgot to mention that Houdini's node editor fully utilizes Qt, so creating any type of "beautiful" UI in it is not a problem at all.

I remember that Autodesk MAYA and MAX convertion to Qt... it took whole year of working for stuff under the hood accordig to them.

I wonder how long it could take for Houdini? While its node based so i assume its easier, mayby LLM can help in converting that old code?

alexeyvanzhula1984
I forgot to mention that Houdini's node editor fully utilizes Qt, so creating any type of "beautiful" UI is not a problem at all.

https://www.youtube.com/watch?v=3f53T2-vmy4 [www.youtube.com]

oldteapot7
I remember that Autodesk MAYA and MAX convertion to Qt... it took whole year of working for stuff under the hood accordig to them.

I wonder how long it could take for Houdini? While its node based so i assume its easier, mayby LLM can help in converting that old code?

For Maya, it turned out great, but it took a few years. It seems that for 3ds Max, it didn’t work at all, likely due to its heavy reliance on the Windows API.

LOL i have to digest what i just saw snake from nodes and flying and shooting other nodes haha!

So in theory it is possible to overlay those animated nodes on main 3D vulcan vieport, lock zooming and paning and instead of jetfighters xD making cool sliders for morph targets and use it as HUD?

It would be nice to have this kind of parameters for scalar parameters. I found them a lot more versatile than a simple slider next to a parameter.

i like this kind of sliders too, way more compact wich saves sreen space. Mayby its possible to port sliders from Blender?