Heileif
Redshift has matte object. It makes it so easy to integrate the cg into the plate in the renderview, without doing allot of custom stuff just to get shadow and reflection into the background from the cg objects.

Just a toggle away on the mesh to get it working.

https://help.maxon.net/r3d/houdini/en-us/#html/Matte+Options.html?Highlight=matte [help.maxon.net]

Yeah, it looks like they have a good system, a bit the same as Arnold.

This is an example of what I really want:
Your text to link here... [www.youtube.com]

No one else bothered by this behaviour?

I have an issue which is quite annoying, if I'm modeling on the Geometry level with the Hide Other Objects mode selected, then switch tabs to another tab, then switch back to the Scene view tab, the view has changed and the more I switch tabs, the more it changes.

1 Create a Box at object level.

2 Move the object in the object level, say Translate -2 on the X axis.
* Don't type it in, actually move the Box with a translate handle.

3 Centre the Scene View on the Box.

4 Set the View Mode to Hide Other Objects.

5 Dive down into the Geometry level on the Box.

6 Centre the scene view on the Box.

7 Now go to another tab, such as the Spreadsheet tab and then come back
by selecting the Scene View tab.

You will see that the Scene View has changed it's position, it's not
where it was before you changed tabs and the more times you do this the further the
Scene view moves away from it's original position.

The work around:

If you Freeze the Geometry transformations to a Transform SOP on the
Geometry level.
or
If you create a camera and set the Scene View to use it, then the Scene
view stays constant.
or
If you copy the translate then zero it out, so that the Box is at 0,0,0
again, switch tabs, then in the Objects tab paste it back in (-2 in x),
then the issue goes away.

The problem seems to be, if you move the geometry in Object level by the
transform handles, then dive into Geometry level, and swap tabs, when you
return to the Scene View tab, the view has moved, it only does this in
Hide Other Objects View mode, if you use Ghost Other Objects mode the Scene view
stays put, the way I would have thought it should.

Does this only happen with me? am I missing some hidden switch somewhere?
I would have thought that the Scene view using Hide Other Objects or Ghost Other Objects
would be consistent and the Scene view should stay put.

T.

I'm using Rocky Linux 8.8 and it is incredibly stable.

alexeyvanzhula1984

Tser

alexeyvanzhula1984
Everyone who says that modeling is done in Houdini doesn't mention that it's challenging and time-consuming. Houdini lacks many fundamental tools for proper modeling. Simply moving points is already a pain, and it's the most crucial operation in modeling. I won't even mention proper tools for connecting components or the responsiveness of the tools, etc. These are just a small fraction of the drawbacks that prevent "normal" modeling. Of course, if you revel in nodes and the unhurried structuring of them into containers, you can model in this way. But we need to call things by their names. Modeling in Houdini is slow and uncomfortable, regardless of how you've set up hotkeys and marking menus and even your own workflow.

For me, it's no harder than modeling in Maya, certainly not slow and uncomfortable, I was modeling in Maya for many years, I would still be modeling in Maya if Houdini didn't do the job. I personally love the degree of depth that the Houdini sops give.
T.

I'm not talking about the advantages of Houdini. Houdini's data handling is top-notch. But I will never believe that modeling in Houdini is comparable to Maya. I'll ask the first things that come to mind:

1. Don't you notice viewport glitches?
2. Haven't you noticed that initiating tools makes Houdini's viewport flicker?
3. Do you actually find it convenient to move points as implemented in Maya with manipulators and especially in Tweak mode? I know that Houdini has Tweak mode, but it's so pitiful.
4. How about simply moving the pivot? Is it really as convenient as in Maya with the 'D' key?
5. Do you really enjoy switching from one object to another? In Maya, you just need to click in the viewport.
6. Do you enjoy moving points of two objects simultaneously?
7. Do you like that ugly display of materials in the viewport?
8. Do you truly believe PolySplit and EdgeLoop can replace Multicut?
9. Do you have any analog to the Connect tool, which simply connects selected components?
10. Can TopoBuild be compared to QuadDraw?
11. Are you familiar with viewport crashes?
12. Do you know the feeling of having an open error log window in the bottom right corner?

This is just what came to mind. The main reason there aren't direct modeling videos in Houdini is that it's genuinely inconvenient and time-consuming. And the few videos that do exist are insufficient.

Sorry to here that your system is causing glitches or flickering with Houdini, I don't have these, Houdini is rock solid and stable, if you want to move the pivot just press the "insert" key.
T.

alexeyvanzhula1984
Everyone who says that modeling is done in Houdini doesn't mention that it's challenging and time-consuming. Houdini lacks many fundamental tools for proper modeling. Simply moving points is already a pain, and it's the most crucial operation in modeling. I won't even mention proper tools for connecting components or the responsiveness of the tools, etc. These are just a small fraction of the drawbacks that prevent "normal" modeling. Of course, if you revel in nodes and the unhurried structuring of them into containers, you can model in this way. But we need to call things by their names. Modeling in Houdini is slow and uncomfortable, regardless of how you've set up hotkeys and marking menus and even your own workflow.

For me, it's no harder than modeling in Maya, certainly not slow and uncomfortable, I was modeling in Maya for many years, I would still be modeling in Maya if Houdini didn't do the job. I personally love the degree of depth that the Houdini sops give.
T.

raincole
For people who use Houdini as their main modeling tool, how about sharing your workflow? Or make it a premium tutorial. I think quite a lot people will buy it.

I mean manual box modeling, not a super-procedural one. I mean things like Extrude/LoopCut/Slide and more importantly, when the topology is "generally right but not exactly right" how do you manually fix it? (don't say "if you planned topology beforehand you'll never need to manually fix things". It's like saying if you write code carefully you'll never need a debugger.)

I know Houdini can do all of these. My question is how people exactly do these, like the literal shortcut keyboard strokes they press or the radial menus they use.

I only use Houdini for modeling, I used to use Maya, but now I have adapted to Houdini's work flow, once you know the equivalents you can then use Maya or any other modeling tutorials.
Houdini web site has tutorials on every modeling node there is, I use Polydraw a lot for non procedural modeling, also google will help you out a lot, it did for me.
T.

I have Rocky 8.8, Intel with Asus mobo and RTX, Houdini 19.5 is as stable as a rock.

d3dworld
@Tser @asorgejr
thanks a lot for your replies i appreciete it , i,m 34 now and having a job burnout since years because my job experience only is just Data Entry and nothing else , my whole mind is consumed with the idea that if i start learning will i get a real job ?
or will my learnin efforts be in vain ?
i hate coding but i love artistics only fields

You really don't need to code if you don't want to, as you can play around with that area later on, just use the available nodes and plug them together for an outcome.
T.

d3dworld
Greetings,
Everything said in this thread did a great job of pushing me away and killed what motivations left in me to start learning.

Kind regards.

I have been in the 3D graphic biz for more years that I care to remember, starting off with 3D Studio to 3DStudio Max to Maya to Softimage then back to Maya and along the way using Blender, Lightwave even Bryce along with their multiple plugins and add-ons and now I am with Houdini which I will never leave, it is a truly spectacular software which can do anything the others can do, but better, sure, it may have a sharper learning curve, but once you understand it's logic the learning gets easier. What you have to understand is that the others mainly use pre-configured tools/buttons/scripts/shortcuts to do the things they do, Houdini also has this but more importantly you can create your own as you want, you don't have to learn this, but it is there for you. Learning Houdini to do the same functionality as say Maya doesn't take any more effort that what you put in to learn Maya, it's the deeper functionalities which take more effort, but the rewards are great. Learning and using Houdini is the best decision that I could of made, it's a pity I was seduced by the others first.
T.

Hi Brians, how is the "Out of Core" progressing with Karma XPU, seems like a lot of the other renderers have versions of it, if it was only to take care of the textures and not the meshes then that would be a tremendous help with the memory load.
T.

As I mainly model and use the PolyDraw sop, the integration of the functionality of the Edit sop would be a great help, with it's the ability to select a point on a mesh in the Perspective view and manipulate it with the handles, PolyDraw really needs this, at the moment I have to go into top, right/left or front/back views to move the PolyDraw selected point/s for any accuracy.

citizen

asorgejr

Very good list. When it comes to modeling>symmetry, I wish the mirror SOP would be fixed before adding more features around this.
The way the seam consolidation is handled is absolutely atrocious, no other 3d DCC solves this so poorly. When you finish a re-topologized character for example and mirror it, you have to disable the useless "Consolidate Seam" and manually select the medial borders of each half and fuse them.

I use the Mirror sop all the time when I model, if there was a snap to the points of the mirrored side, that would be nice, but all in all it does what I need.

Is there a way to use NVLink on 2x RTX 3090ti video cards and render with XPU? I need more gpu memory.
I believe Arnold can utilise this to pool the memory.
T.

brians

Tser
It was caused by a material on a group of faces on the parker light, changed out the material and everything works as it should, quite amazing really, the material that was changed out didn't look at all out of the ordinary.

Sounds a lot like a bug
Is it possible for you could create a simple repro scene? (eg sphere on a plane, basic stuff, not any private material you're working on)
If so I'll file a bug report and hopefully look at it soon.

thanks!

Thanks Brians, I created a couple of simple scenes and couldn't reproduce the issue, it only seems to manifest out on complex scenes with many materials, at least I now know what to look for if this issue appears in the future.
T.

RaySop1.png is the result produced in the viewport, RaySop2.png is the result when rendered to either MPlay or to disk, I can't give you a copy of the file for obvious reasons, but it is a straight forward setup. It also works fine in CPU mode.

Edit:
After a lot more testing, it appears that it's not just the Ray sop, it's any object (mesh) which is too close to another, for e.g. if you have a small grid just above a larger grid and the smaller grid has an image and an opacity for a cutout, the image will fade (in XPU mode).
There is an exception, all is well in CPU mode or in XPU mode if one of my GPU's out of the two stops rendering because it's gpu memory is too full.

Edit2:
It seems that it is a GPU memory issue, if you create a basic scene with a few objects, then all is well, but with a complex scene with lots of materials and objects, the problem arises, I am using 2x 2080ti's with 11GB onboard, very odd that Arnold has no problems with rendering the same scene.

Problem solved...
It was caused by a material on a group of faces on the parker light, changed out the material and everything works as it should, quite amazing really, the material that was changed out didn't look at all out of the ordinary.

T.

I'm having issues with the ray sop when rendering with xpu, it works fine most of the time in the viewport, but when rendered out with mplay or to disk at 4k the decal fades away to just a ghosted image on the surface, but sometimes it actually renders properly, very frustrating, I've tried every setting there is and many different images and formats but nothing helps, if I just project a color with an opacity, it always works, or an image without an opacity it works. Arnold works perfectly.
I really need to project decals onto the surface of objects.
Any help with this would be an enormous help,
Thanks,
T.

After a couple of days tinkering around, I created a subnetwork node to create the shadows and AO, it works really well for still shots and works perfectly with Karma XPU.
T.

Nailed it... Thanks for that jsmack, it works really well in XPU, now for some playing around...