Welcome to od|forum

Register now to gain access to all of our features. Once registered and logged in, you will be able to contribute to this site by submitting your own content or replying to existing content. You'll be able to customize your profile, receive reputation points as a reward for submitting content, while also communicating with other members via your own private inbox, plus much more! This message will be removed once you have signed in.

Alexey Vanzhula

Houdini 16 Wishlist

558 posts in this topic
1 hour ago, sebkaine said:

Bring back the exemple files in the help.

+1

I would go even as far as there needs to be a help file related to every node. I know that is insane, but I would gladly have an additional download source so they could optionally ship with Houdini. 

2 people like this

Share this post


Link to post
Share on other sites

The possibility to use VEX in channel fields instead of old Hscript channel expressions.

I'd really like this so I wouldn't have to scrub througth two different docs when I dont know a function or stuff like this...

Even in the docs it says Hscript is "Houdini's legacy scripting language", its time to fully integrate VEX with Houdini: I'm learning VEX myself and some of the things I do (most) in VEX are not even possible with standard nodes....

 

Btw this threads are usefull but SideFx by its own is doing a really good job at upgrading Houdini...big uf to them :)

Edited by SteveNi
3 people like this

Share this post


Link to post
Share on other sites
On 2016-10-11 at 6:19 PM, LaidlawFX said:

+1

I would go even as far as there needs to be a help file related to every node. I know that is insane, but I would gladly have an additional download source so they could optionally ship with Houdini. 

That actually is a requirement of adding a new node, but it's not often followed.

1 person likes this

Share this post


Link to post
Share on other sites

It's be great to have a basic volume shader, that has an identical interface to the volume visualization SOP. And have a button on both the SOP and the shader that can easily transfer the settings from one to another. 

With the way things are now, it's possible to set the visualization settings 3 times, (inside the sim, in SOPs, and then in the shader).  Pretty redundant, there should be a lot simpler and faster way to get what you are seeing in the viewport, into the render

Share this post


Link to post
Share on other sites

I love Houdini, 

still a noob---

but coming from Softimage there are a few things that are easier to set up. that i would love to see in H16.

one thing is setting initial pop attributes, have a tab with them as a toggle to turn on to override any incoming attributes. mass, orientation, scale, color, size, and custom attributes. I know that a lot of it can be set up after the pop source. but having it inside an initializing tab. and have the particles start up with let's say random rotations, would have less nodes in your pop tree. 

 

 

Share this post


Link to post
Share on other sites

I think it's not a priority for sesi but  if we can have a (good) dwg / dxf import and also a denoiser for mantra ... then I don't need anymore Max

Share this post


Link to post
Share on other sites

 

6 hours ago, lesath said:

I think it's not a priority for sesi but  if we can have a (good) dwg / dxf import and also a denoiser for mantra ... then I don't need anymore Max

If you are getting noise in your renders post an example file, and some one can show you the parameters to tweak.

Also if there is a issue on dwg/dxf import you should submit a bug/rfe with sesi. We've used them in the past, most often it's related to bad files, as in it wouldn't work in Maya either. You should even submit an example files and explains what controls or data that you are missing from the file. It could actually be there, but just not in an obvious location.

Share this post


Link to post
Share on other sites

 

8 hours ago, LaidlawFX said:

If you are getting noise in your renders post an example file, and some one can show you the parameters to tweak.

 I know there is a great community here to help me if I'm stuck with a render or something else. I'm pretty happy with Mantra but a denoiser is not evil , I still think it could save time in some productions...

 

8 hours ago, LaidlawFX said:

Also if there is a issue on dwg/dxf import you should submit a bug/rfe with sesi. We've used them in the past, most often it's related to bad files, as in it wouldn't work in Maya either

I asked already last year to Sesi for the dxf import and we tried with many files but they haven't updated support since the 1990´s and It was not in their plans to do it. So I was thinking it was the right place to ask for it... 

Edited by lesath

Share this post


Link to post
Share on other sites
21 hours ago, zhyravetsky said:

please add pivot rotation mode and bundle transferring through hip files

+1

Share this post


Link to post
Share on other sites
8 hours ago, lesath said:

 

 I know there is a great community here to help me if I'm stuck with a render or something else. I'm pretty happy with Mantra but a denoiser is not evil , I still think it could save time in some productions...

 

I asked already last year to Sesi for the dxf import and we tried with many files but they haven't updated support since the 1990´s and It was not in their plans to do it. So I was thinking it was the right place to ask for it... 

For the denoising , it's not perfect but neat video is really helpful in certain scenario, the price is great.

https://www.neatvideo.com/

 

For your dxf workflow i would investigate a little with this package 

http://moi3d.com/index.htm

While it looks cheap at first sight it is not. It's a very good poly exporter from CAD mesh.

we use it to convert some NURBS mesh into poly and it give great result maybe it could be the tool you miss in your workflow to ditch Max ...

Let me know in case it works for you 

 

From the site :

MoI is also a fantastic complementary tool for a polygon-based artist since its CAD toolset and advanced boolean functions enable extremely rapid creation of mechanical or man-made type “hard surface” models. The icing on the cake is MoI’s unique polygon mesh export that generates exceptionally clean and crisp N-Gon polygon meshes from CAD NURBS models.

Edited by sebkaine

Share this post


Link to post
Share on other sites

Ability to set and force the bounding box of a VDB , by inserting a reference bbox as reference exactly like the isoffset node.

when using volumesample(0,'surface',@P), you need to be able to control the area you want distance info in.

Share this post


Link to post
Share on other sites
1 hour ago, sebkaine said:

For the denoising , it's not perfect but neat video is really helpful in certain scenario, the price is great.

https://www.neatvideo.com/

 

For your dxf workflow i would investigate a little with this package 

http://moi3d.com/index.htm

While it looks cheap at first sight it is not. It's a very good poly exporter from CAD mesh.

we use it to convert some NURBS mesh into poly and it give great result maybe it could be the tool you miss in your workflow to ditch Max ...

Let me know in case it works for you 

Thank you, I can remember that we tried moi3d at work a long time ago  to convert some rhino files and it was very powerful. Also it seems  great to import dxf (it reads line, polyline arc, curve) so you 're right  it can be the  package I need. So I have to give a try.  But it would be easier to import in native in Houdini and keep a curve as a curve, a circle... And from that  build my project in Houdini.

For the denoiser, it was also my wish to stay in Houdini and not use another tool. Artists  seem  happy with it in renderman so why not in Mantra ?

Share this post


Link to post
Share on other sites

The thing is that in my own XP, we use this for Maya import. Or Maya is supposed to have everything with correct exporter. Truth is that the conversion where not good enough. moi is just working. For certain task like UV layout it is imo better to rely on small external tool , that do work ! 

Have you try to unfold UV in Houdini for a character ... :)

 

For the denoiser i think it would be far better to have better refraction / transparency models first with dispersion and physical transmission (that is the bigest pb with Mantra imo).

Denoiser is ice one the cake. Mantra token are free after all so if you need to scale you can.

Edited by sebkaine

Share this post


Link to post
Share on other sites

As for a denoiser: Did you try "Ray Histogram Fusion"? Quite interesting results. The default is "combine -t 20.0" but smaller values can also work.

+42 for MOI. Awesome software with great support.

As for wishes:
I personally would be most interested in a GPU render mode. For my work Mantra is abysmally slow. AMD is quite keen to get ProRender into more packages, Maxon just announced that it will integrate it in future versions, so maybe SideFX could support it too without having to do all the work themself? It's using OpenCL, so it should work on all platforms.

Otherwise it's mostly small things in my journey so far (If there are solutions already I'd be very happy to hear them):

- Allow to directly click-drag-drop wires from one node port to another without clicking two times. Houdini is the only node based software I know that does this and it's quite annoying as the only option. I do it wrong a dozen times a day and see it happen in every tutorial I watch... It's good if one has to move larger distances, but most of the time I find it annoying.

- Selecting and doubleclicking nodes with a wacom pen is somehow very frustrating. Most of the time I move the node a tiny bit instead of selecting or diving down. I don't have this problem in other software (and use a Wacom for about 20 years) so it seems to be something specific to Houdini. My doubleclick distance in the Wacom Driver doesn't seem to make a difference.

- Diving in and out of subnetworks all day drives me crazy. Why can't I just see my result while I'm inside a solver etc. Makes no sense to me? I hope I'm missing something here...

- Jumping between contexts all the time is really annoying as it is implemented ATM. This could use a major overhaul and rethinking. I found some solutions to ease the pain but they only go so far.

- Why do the built in script editors not show a.) all local variables I can use in a given node/parameter field/wrangler and even more important b.) parameters for functions. 95% of the time I need to go to the helpfile only to find out what parameters a given command has. This feels silly in this day and age.

- Many things in the GUI don't seem to be saved in Desktops or to defaults. For instance the Render View Update Time and Delay that are way too small by default.
[EDIT: got that working now - If I save my own Desktop with this set as I want and use that Desktop as default, it works - not intuitive but hey... ;-) ]
I'm also unable to get the Global Animation Options to stick to Realtime Playback even if I "Save As Default". There are so many ways to customise Houdini so these somehow feel strange...?

- I also don't understand why opening a scene would not select the same camera and view etc. that were active when it was last saved. I normally would want to be exactly where I left.

Even if some of these can be solved by some scripting magic, I think it would make sense to present the newbie with a more intuitive default.

Cheers,

Tom

Edited by Thomas Helzle
Corrected some stuff I found in the meantime...

Share this post


Link to post
Share on other sites
11 hours ago, Thomas Helzle said:

As for wishes:

I personally would be most interested in a GPU render mode. For my work Mantra is abysmally slow. AMD is quite keen to get ProRender into more packages, Maxon just announced that it will integrate it in future versions, so maybe SideFX could support it too without having to do all the work themself? It's using OpenCL, so it should work on all platforms.

 

Mantra going to GPU is a feature lot of people would like to see. The question itself could make a good topic by itself.I am not a rendering expert but at the moment either render engine are :

- working on the CPU

- OR the GPU

you don't have hybrid approach where a CPU engine can have some part of his code executed in a GPU in parallel. (If this already exist i would like to discover who do it).so asking to mantra to go to the GPU is basically rewritting from scratch the engine which is a HUGE / EXTREMELY / DAUNTING complicated task to achieve.

Maxwell Render which is a team focus at 100% on render engine has try to do what you ask convert a CPU engine to the GPU. Maxwell Devs are extremely good and brillant and focus,but they are now just hitting a huge wall with their V4. Has they have underestimate the task a lot. Maxwell 4 is still a Beta GPU engine with lot of CPU feature not working.for the moment you can use the CPU engine OR the GPU you can't have hybrid approach. Vray guys which are also extremely focus also offer a CPU OR GPU approach.In any case i doubt Maxwell GPU or Vray RT will be the futur standard in term of GPU rendering. The same apply for amd engine imo ...

On the other end you have 2 brillant company that fight against GPU complexity with otoy and redshift. both engine have a houdini plugin develop by the same great developper. So i am wondering if monopolizing a lot of sesi dev to convert mantra to the GPU is not a waste of resource. There are so many area where houdini need to progress like rigging / animation / modeling / workflow / real time / viewport.

redshift / octane are at 600$ which is very good price imo , i doubt sesi guys will have the ability to catch with those 2 engine if they try. and the maxwell 4 fiasco and the vray RT limited adoption are imo 2 proof that it's not so easy in the real world to convert a CPU engine to the GPU. Let's wait arnold answer to this question.

If mantra could just focus on being what it is already : a great VFX engine that can scale and that can replace easily VFX cpu engine like arnold / prman / clarisse / 3delight / guerilla , that would be just perfect imo. In this regards denoising would be indeed a great feature to support in futur release ... but better transmission model first would be great ! :)

 

Edited by sebkaine

Share this post


Link to post
Share on other sites

Well, my main renderer is Thea Render which offers extremely efficient simultaneous rendering on GPU and CPU for some years now. If the GPU(s) do(es)n't have enough RAM, it falls back on CPU and the CPU part is also very efficient, so my 6 core i7 contributes substantially to my two GPUs (GTX 660 TI and GTX 980 TI) and my older machines with outdated GPUs can still serve as useful render slaves. Each GPU and CPU can have a separate priority, so you can for instance use your monitor-GPU with low priority for rendering too, while still fluently navigate the OS while rendering. 
Thea also has bucket rendering on GPU which solves many memory issues, especially for large scale rendering. Through it's multiple render cores (biased/unbiased and different levels of complexity) it covers a lot of areas.
It wouldn't be the ideal companion for Houdini yet, since it hasn't got volumes and hair covered so far, but it's rumoured that this is about to change soon. At the same time it offers many things mantra does not cover.
A network render license is 49.- Euro, 30 nodes are 399.- Euro, so it's also very affordable.

Maxwell was the first widespread unbiased renderer but has lost it's initial margin long ago IMO.
I personally am not too fond of Otoy/Octane (I have the Lightwave plugin and demo it for Houdini). It always felt very limited and convoluted in a strange way to me, using nodes but in a clumsy way. It will have to be seen if the 3.1 update makes it a bit more open and versatile with OSL support etc.
I'm not too happy about their subscription roadmap and they seem to go off in all directions ATM. Not my kind of company if I can avoid it.
Redshift is great in principle, but price and license model are not that appealing to me. It also looks like the ideal company to be bought by some larger entity so I'm reluctant... (Old XSI user here ;-) ).
I don't care much for Arnold, even less since Autodesk bought it.

Thea also started as CPU only renderer but integrated GPU in multiple steps, offering more and more complete features with each. I personally think it's the best renderer on the market technically for what it does. The main dev is a genious who always comes up with new ideas and fresh approaches. Not perfect in all areas and as I said, not the ideal Houdini companion (although I often export my scenes to it) but for visualisation and and non-VFX rendering it's ace.

I mentioned AMD ProRender in my post exactly because in that case SideFX would not have to code the renderer itself. The Max integration shows what it's capable off, I currently demo it for Rhino, where it's in an early beta stage. If Mantra itself can be made to use GPUs I don't know, but I no longer question the fact that GPU is the future for rendering (I was sceptic myself for a long time).
The AMD FirePro™ W9100 is available with 32 GB RAM already so the memory limitations will become history soon enough and the bottlenecks between GPU and CPU and the different memory systems are already vanishing.

Overall it would be good to see more GPU support in Houdini.

But sure, my needs may be different from most other users. But that's what a wishlist is for ;-)

Cheers,

Tom

Share this post


Link to post
Share on other sites

Thanks for the very informative answer Thomas. That is a very good news to learn that somewhere one human use BOTH cpu and gpu power in parallel. I have never heard of thearender before, maybe because it's not vfx oriented. But in that case i agree with you having the ability to accelerate the render process IF there is a graphic card on the machine would be great. A scenario where having a GPU is not a necessity but if there is one Mantra use it to speed things up. Exactly like you would with a FLIP or PYRO sims and OpenCL. I love that idea especially if you are not CUDA binded.

For octane while i love the product , i agree that sometime their communication is kinda esoteric ... For redshift don't tempt the devil it might have already ear you !

For the amd renderer if it is available both in H and C4D that wouldn't hurt either. But for VFX work i doubt it will be reliable enough , but like you said might be useful for other non VFX users.

Cheers

E

Edited by sebkaine
1 person likes this

Share this post


Link to post
Share on other sites

By the way Redshift has the same approach as Thea, on that if you do not have enough ram it falls back to the CPU as well. I really doubt that the evil company that we are talking about = ) will be acquiring Redshift, since they just got Arnold. The nice thing about Redshift over octane for example is that you get way more control on the final image, you can use attributes, etc. So you Houdini workflow is maintained to an extend of course. Still you can not do all the stuff that you can in Mantra for example. Also the developers are great they hear their costumers and they have a release pretty much every week, bug fixing, adding new stuff. 

1 person likes this

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!


Register a new account

Sign in

Already have an account? Sign in here.


Sign In Now