I can confirm that I get a lot of python error messages in Houdini which I never got before with fairly simple modeling and no hda or plugins loaded. Most of the time I can ignore the error messages.

Houdini rarely officially supports Apple graphics cards that currently come with laptops and desktops. I learned this with purchasing a Mac Pro. If I were to do it over again, I wouldn't purchase Apple products for serious 3d work. Apple's phones and laptops are pretty reliable but I have had to replace almost everything for defects in my recent Mac Pro. At least it was under warranty.

The 2023 version is available on Steam

You don't get updates or upgrade pricing but it has no limitations on features.

Substance painter is $150 on steam for a permanent license. It is much better than photoshop for organic models. There are other free alternatives but I've found them buggy.

For hard surface models, photoshop alone might work. You need to create uv's in Houdini, generally with a uv flatten node and seams. You might want to add a uvlayout or uvedit node to fix rotations to make painting easier. Use the layout in photoshop as a template (generally setting it to multiply mode on a separate layer). After you paint the texture, turn off or delete the template layer and save the photoshop file as jpg, tiff, or whatever. Use this image as a texture in a principled shader in Houdini and assign it to your model. It is old school but often works.

Fairly commonly I import a model from other programs into Houdini that has already been textured. Typically, there are the usual albedo maps but frequently a specular map, normal map, and displacement map. To reassign these accurately in Mantra shaders, I've done a lot of guesswork for typical settings. I often need to reduce the normal scale to around 0.1 and displacement scale to 0.001. I've assigned specular texture maps to the reflectance texture but also frequently inverted the specular texture in Photoshop and assigned it to the roughness channel as well. For the normal scale, in Mantra, one can pick normal range of 0 to 1 (for 8b) or -1 to 1 for floating. Which is more common from substance painter, zbrush exports, or textures from 3dscanstore or texturingxyz? There is also settings in Mantra to invert the x or y. From which programs will one need to do this? I would like to do this logically rather than just play around with the settings until it looks good. What are your suggestions?

Thanks. I'm not sure why there is a difference in display (apart from the advantages with dual quaternion),but here is the display differences. Note, that if I use bonecapturebiharmonic, the display WILL work properly if I set the visualization to single bone instead of multiple bone:

Initially I had problems with setting up a rigged hand model using KineFX as the bonecapturebiharmonic did not display capture regions even when the visualization was set to color>capture regions. It turned out that this was just a display issue and I was able to paint weights and rig without problems.

Recently techsupport suggested an alternative setup using the jointcapturebiharmonic node. This avoids having to create a bonecaptureline and tetembed or tetconform node. But I discovered that more importantly, I got better deformations, particularly if I set the jointdeform to dual quaternion. Not only was there less volume loss but I could often omit a deltamush node to correct creasing.

Is the bonecapturebiharmonic method, shown on most SideFX tutorials, an obsolete method? Is there any disadvantage to using jointcapturebiharmonic instead?

Thank you very much! It appears there are two separate issues. The inability to capture was user error due to my capitalization. But there is a separate display issue. If I set visualization to bone capture, it only works if I have single bone visualization selected. The biharmonic capture node itself is supposed to show the visualization. I think this is the bug that SideFx has logged.

Fortunately I was able to do weight painting, add a delta mush and set up a rig for the hand model, so all is good.

The issue is the instability of capture. Here is the fur dude and if you set the visualization to color: capture regions or capture single bone, it shows failure to capture.

Thank you. It was purely a display issue of weighting. That still seems a bug, but a minor one. I added the rest pose with all the skeletal parameters and a bone deformed and it is working. Tech support logged this as a bug after receiving the file.

I've rigged in Houdini with bones, softimage, Modo, and C4D without problems. Either I am missing something with KineFX or it is buggy.

This is not a parenting, controller issue, hda, weighting, promoting parameters, etc. it is a failure of capture.

Edit: it looks like this is a visualization issue. I set the visualizer to single bone capture and the capture seems to have worked.

Thank you. I think something is broken in Houdini 19.5 with KineFX. The biharmonic weighting seems to be the issue. I have sent the file to support. The fur dude is a bit unstable but sometimes captures. Obviously without capture, I can't proceed with controls and creating an hda.

I am struggling with KineFx. The old bone system was quite reliable. Here is a hand model with skeleton and bone capture biharmonic doesn't assign any weights. I have double checked by adding a visualizer for weights and it is all black. Here are the steps that work fine until the bone capture biharmonic is added. I've tried tetconform as well and played with the settings of all of the nodes without success. Any suggestions?

Unfortunately, this is exactly the same setup as in the pdf and in the vimeo tutorial, so it is how SideFX suggests setting up the rig. I tried tetconform and got the same problem and changed the settings. I think the downloaded bgeo.sc file must be bad. I downloaded again and tried the daily build and got the same problem. The rig is present (see attachment)

I have no trouble rigging with the old Houdini bone system. However, following the tutorial on KineFx with fur_dude, I get an error with capture: "invalid hierarchy". How can this be resolved?

The problem seems to be solved to saving the file to a different directory, so I assume there was some minor corruption. I'm not sure why save or save as to the same directory didn't fix it, but the problem disappeared with saving to a different directory.

Recently, with the production and most recent daily build of Houdini, I am seeing an annoying display issue in geometry mode. For certain polygons, it appears as if they are disconnected, which makes selecting edges and points more difficult. I'm not sure if this is a Houdini issue, some setting that got changed, or an OSX graphics driver issue, but here is how it looks. It occurs if there is a moderate angle between polygon faces and one is zoomed in. Any suggestions?

There is no real separation of polygons and adding a fuse node does nothing. There is no difference if I set preferences to openGL or openCL for misc graphics.

Specs: MacPro, W5700X graphics card, high spec.

Very elegant, Konstantin! I have always had to do the selection manually. Can you explain the VEX code:

Soft Boolean is definitely worth it. Nothing is perfect, but compared with Modo mesh fusion and various other systems out there, it does an excellent job. As was mentioned, you can use quad remember to soften this up if you wish (second attachment).