Hi everyone,
First off, I really love where Houdini is going and what artist can do with it. I have the deepest respect for for all people working at Side Effects and everyone that uses Houdini. I may not know everything about all topics that I am talking about, but that is exactly my point.

That said, I am frustrated with Houdini's approach to everything concerned rendering and I sincerely hope to rattle some cages, so that we may get some serious improvements in the future.

Rendering is quite possibly the most essential step of any 3D process, but every part of it in Houdini is antiquated or requires incredible effort to learn.
I am coming from a background in Maya and 3DS Max and you might say, that I am just spoiled, but I know it is possible to offer premade modules that are noob-friendly, fast to use, efficient and give beautiful results, while still offering granular precise control.
1. Why not make one ueber-shader builder of each type of Material (atmosphere, procedural, surface, etc) and give them presets. - That way you can quickly, switch between presets to see what changes. - Of course, these presets should attempt to use the same inputs, so that when I have a texture in the diffuse slot, it should be in all the diffuse slots of all the presets, when I switch around. Furthermore you should be able to dive into every preset to see how it works and to modify it if necessary. This would greatly help the user to learn.
2.A: VEX context should have the option to give a (flexible) preview above each node. Particulary whenever textures are involved it is helpful to just see what effect you are causing with each node. This is modern industry standard and greatly helps to organize and to work efficiently.
2.B:Not to mention that diversifying the look of all (VEX) nodes, or at least increasing the size of their icons (or even keeping the size of icons constan until you zoom out too far and they overlap), would greatly improve readability of networks.
3. Furthermore there should be a better library of better premade materials. I know this has been called for before. Standard Materials should include:
Paper, Concrete, Clay, Car paint, Gold, Water, Glass, Diamond, Mirror, Emissive, Carbon, Silver, Iron, Rusty Iron, Brushed Metal, Chrome, Wall Paint (fully procedural with paint splashes and all), Brick (with fully procedural binds bump maps etc), Rubber, Plastic, Polished Wood, and more. You can call this my lazyness, but there are two great reason to include this sort of thing:
- Teach by example.
- Every company has deadlines and time is money.
4. There should be standard Materials that immitate modern gaming engine standards. Particulary Materials that are the same as the Unity 5 Standard Material and the Unreal 4 Standard Material and work with the same maps. This should be even more obvious with the possibility for application in Houdini Engine.
5. Takes, bundles, flipbook, visibility nodes, layers? Even if all these things were necessary, the user should have a simple centralized control and a simple overview over the functions and purposes of these things. One menu to rule them all. One menu to bind them… One menu to control everything that controls what gets rendered where, how and why.
5.B. Implement preview-options into the menus so we get a rough idea of what we work on.
5.C Baking anything (to textures) is entirely unnecessarily complicated. Micromesh rendering can be used to render to textures - How? Why? Why do I need to have a specifically named camera in the scene when I don't even use it? - Why can I not have a menu to select, what to geometry to bake from, which geometry to bake to, and what and how to bake? Standard maps should include Diffuse, Specular, Normal, Height, Ambient Occlusion, Cavity, etc. and an option to bake custom designed properties baked in a custom designed way. There should also be an option to use a standard cage or a custom mesh. And of course, It should be possible to bake animations.

I hope this thread will result in some lively discussion of my ideas. I also hope, that I will not be shot down by people who have invested a great deal of time and pain into learning these things for themselves. Nobody wants to diminish your achievements. I strongly believe that it would be better for every Houdini expert, if the tool became a more viable choice for more companies, by simply being more user friendly.

what kind of production are you doing?…

Well, I suppose I am looking at it from a gamedeveloper/visualizer point of view.

http://cargocollective.com/dimitrisCG/Procedural-Tank-Houdini-FX [cargocollective.com]

The needs of my ideas aside, I think strong rendering and baking tools are necessary for every modern 3D package.

I thought that some real life examples of rendering projects in Mantra from forum members on ‘everyday’ projects would be really informative.

For example even if not your plain vanilla everyday project I would love some stats on the rendering part of the ‘Monstro’ project.

Most of the post-mortem and customer stories have been on feature film projects, which by their nature have much more resources on their disposal than average TV or game projects.

I really can't see how (speedwise) could Mantra PBR compete with for example Vray/Arnold/Redshift3D/MentalRay/Cycles on such projects and I would love to see examples if any. Mantra in Micropoly or limited raytracing mode is pretty fast but then there is the extended setup time if any form of GI is to be used.

+1 for baking.

My current workflow at work for creating procedural hi-poly models for baking is to export it to FBX to open in Modo and bake it there.

But it involves some issues where I sofar only have managed to export the “Cd” attribute as a vertexmap that Modo understands, and then in Modo have to convert that to a weightmap instead of a colormap, before I can use it as a mask in the shader setup I did in Modo. Would have been so much easier just to bake it Houdini than to handle these 200mb FBX files which for some reason almost kills Modo.

Reading the list I was hoping for something new to have a big discussion about but if you scratch the forums, it all the normal requests.

Good ideas, though, they are suited to some, but not all, workflows.

I was aware that some of the stuff has been requested and discussed before. Never the less, I feel that it is important to colllect all problems and look for a systematic common solution.

Many updates/upgrades to big software packages fail rather miserably, because tools are being added and patched in, while the whole segment needs an overhaul.
For example, in Maya they rather recently added new modelling tools. They added a whole tab of modelling tools. And the funny part is, that the tools are significantly slower than before and incomplete. You still have to go through old modelling menus to get to the old tools you frequently use. 3DS Max is even worse for the same reason.

If we want real improvement, we have to talk about all problems of the same family at the same time. Only then will we be able construct satisfying solutions.
For this reason, I invite everyone to discuss related topics, here.

MartybNz, please elaborate what workflows could be hindered. I think that would be an important addition to the topic and exactly a thing that would be worth discussing.

DASD
MartybNz, please elaborate what workflows could be hindered. I think that would be an important addition to the topic and exactly a thing that would be worth discussing.

Nothing will be hindered by the additions you are talking about, but not all workflows need those things. i.e. it works currently.

It's like the modelling improvements thread in the SI forums- we all know they are great ideas, and will improve the software. I just wanted to state the ‘big discussion’ has been had by many already and I didn't see that it was mentioned in the OP, but perhaps I'm wrong and there will be a big discussion

MartybNz

DASD
MartybNz, please elaborate what workflows could be hindered. I think that would be an important addition to the topic and exactly a thing that would be worth discussing.

Nothing will be hindered by the additions you are talking about, but not all workflows need those things. i.e. it works currently.

Ah yes, my bad for misunderstanding.^^

- I was hoping people would also come out with old grievances and old ideas. Or current problems they didn't want to talk about.
And then maybe people will start talking more and more and start designing better solutions… Like in the modelling improvements thread.

Another point:
We know, that after a certain point in shading, textures are far more important than the shading.

Substance Designer is a fully procedural texture generator. It is faster and more powerfull than any Houdini workflow for procedural texture generation (COPs anyone?). All of its presets result in better materials than most of the material presets from Side FX. And it has tons of them. In combination with substance Painter (a step that should be procedural in Houdini) the tool gives brilliant results in a comperatively short amount of time.
https://www.youtube.com/watch?v=0ETs6v9HlVk [youtube.com]

Look at Quixel NDO and DDO (the whole Quixel suite, actually). It is a poorly implemented bunch of Photoshop filters, but the workflow is fully procedural and the results can be great.
https://www.youtube.com/watch?v=YyJJAp17K-Y [youtube.com]

I see no reason why Houdini would not be able to do better. And I see no reason why anyone would use an inferior tool, if a better one is available.

*rattle* *rattle* ^^

Ultimately the push should come to make these excellent tools seemlessly interoperable.

There will always be a place for a newer-shiny tool that does the next cool thing better, so as long as you can take that output into Houdini as painlessly as possible, then that is only second to having it in native to Houdini itself.

DASD, why don't you just put RFE? Why do you post it here? There are tons of such a newbie topics like this. This is drives me crazy. Why peoples want tens of pre-build materials, instead of learn shading and build whatever you want?

Takes, bundles, flipbook, visibility nodes, layers? Even if all these things were necessary, the user should have a simple centralized control and a simple overview over the functions and purposes of these things. One menu to rule them all. One menu to bind them… One menu to control everything that controls what gets rendered where, how and why.

Yeah, one button to make it look cool! Why use brain, use a button!

Stalkerx777
DASD, why don't you just put RFE? Why do you post it here? There are tons of such a newbie topics like this. This is drives me crazy. Why peoples want tens of pre-build materials, instead of learn shading and build whatever you want?

Takes, bundles, flipbook, visibility nodes, layers? Even if all these things were necessary, the user should have a simple centralized control and a simple overview over the functions and purposes of these things. One menu to rule them all. One menu to bind them… One menu to control everything that controls what gets rendered where, how and why.

Yeah, one button to make it look cool! Why use brain, use a button!

I would guess mostly because of the time constraints? Not every one has the time/resources to write vex code or build shaders from scratch. Building your own library could take months of work, which most of the smaller studios/freelancers don't have. Even the shaders provided right now need a bit of work to be used in productions (see tangent map threads), while most of the competition works out of the box.

Stalkerx777
DASD, why don't you just put RFE?

I had to look up what RFE means. ^^ I don't think Requests get much attention without proper community support. I also explained in quite a lot of detail why I believe it is important to discuss all problems of the same type at once.
What is wrong with improved controls? What is wrong with improved workflows? Every serious and big enough 3D company that I know hires people to make better tools, so that all artists can do their work faster and easier. It is an investment that pays off for them.
Does a painter need to make his own paint to create great works of art? Does a photographer need to be able to assemble a camera from the tinyest parts to make great photographs? - Whoever can make their materials, still retains an advantage. But everyone, including the biggest experts, can work faster with better tools. We would be able to create more and at better quailty. At Google they write tools in python, not assembly. And those who want to learn about the details, can still learn better from great examples than from pure theory or a very limited amount of mediocre examples.
Why fight for the right to work in a more cumbersome way? Why resist progress that is already here and will steam roll you if you try to stand against it? ^^

@MartybNz I kind of agree. I would like to be able to control/create substances from within Houdini. But I suppose it would also be possible to just immitate the functionality and workflow inside of Houdini, without the need to develop some sort of plug-in. This sort of tool seems essential to me. And it does not seem very wise to to make essential tools third-party plug-ins. On the other hand, you are right and the next better tool is around the next corner. So probably a plug-in system would be the wiser choice after all… I don't know.^^ Would be nice to hear more opinions on this.^^

I don't mean to derail this thread but something to keep in mind is that SESI has a huge list of feature requests on top of the list of things they plan to do internally - plus bug fixes - they have to prioritize these in a way that makes sense for them…I know for a fact that the developers are aware of most of these issues but it all comes down to resources…and something fundamental to SESI's development (as I see it) is that they give their users the tools they need to build whatever they want - rather than huge black box systems…
on the one hand this is great - you can build your own solvers, deformers, etc etc
but on the other hand you don't get a gallery of pre-built, production ready shaders…
I think this is something that the Orbolt site was intended to help…

just wanted to place these issues in the right context…

arctor
and something fundamental to SESI's development (as I see it) is that they give their users the tools they need to build whatever they want - rather than huge black box systems…
on the one hand this is great - you can build your own solvers, deformers, etc etc
but on the other hand you don't get a gallery of pre-built, production ready shaders…
I think this is something that the Orbolt site was intended to help…

Yes, I do think that is the point of the Orbolt store. To slowly, but surely create a library of tools and materials to use and learn from. I guess that is a good way of approaching it. Maybe it would make sense to talk about which shaders would be great to discover and use. Maybe it would make sense to encite competition amongst senior users to provide great free shaders.

arctor
I know for a fact that the developers are aware of most of these issues but it all comes down to resources…

I think that we could help the developers (resource-wise) by discussing details and pros and cons of possible solutions.


I also think that some issues cannot be really fixed by (normal) users. Specifically everything that comes down to basic interface principles and usability. - Stuff like preview pictures on (VEX) nodes. - Such things should be discussed and analyzed thoroughly, before anybody even considers implementation.
I think Houdini does many things much better than the competition and the concept that users make their own tools is very strong, but I think parts of Houdini are a bit dated.

Ah ! The biggest irony of it all is that houdini has probably one of the most powerful frameworks for baking maps via the gather loop vex/vop function. We have made all sorts of transfer, such as occlusion, bent normals, lighting removal, etc. And at a much faster speed than most apps that do baking out there . However , I would not say that the gather loop is the most discoverable tool. I do hope that sesi wraps the gather loop into a surface transfer node which would pale all others (and put it in orbolt! )

Substance is MUCH easier to use than cops, most likely because how much of it was molded to mimic and surpass the very familiar photoshop. Howevee, substance does miss the flexibility of VEX cops, which I have used to do VERY custom operators (i.e. a color range filter on steroids) I would be the first one though to jump in joy once we can load substances in houdini

Fwiw I do think sesi listens to its customers and would not be surprised to see some of these improvements make it in in the near future

gather loop vex/vop function - I would love to see some examples of that implemented and or tutorials on the topic.

grayOlorin
Fwiw I do think sesi listens to its customers and would not be surprised to see some of these improvements make it in in the near future

Yes, I think so, too. Otherwise I wouldn't bother.^^

take a look at this thread… I was in your same boat 4 years ago or so

https://www.sidefx.com/index.php?option=com_forum&Itemid=172&page=viewtopic&t=22062&highlight=transfer+textures [sidefx.com]

This pretty much sent me in the right path. There is also documentation of the gather loop Vex function (under vex documentation) as well as if you drop a gather loop VOP, there is a help card

hope this helps!

Thank you! I will dig into it the moment I find the time.^^