Blender Arch Pack

[Domenec Reynolds]

Blendersupports a multitude of hardware offloading options for rendering. After preparing your system according to your hardware (see following paragraphs), you can set these your graphics card as a compute device under Edit > Preferences... > System.

For NVIDIA, Blender supports two GPU rendering backends: CUDA and OptiX. Most modern NVIDIA GPUs are supported. In order to make use of these, you need to install cuda. Afterwards, the rendering backends should be available in the systems options.

Blender is becoming increasingly well known in the professional industry. As such, there are now alternative rendering methods to EEVEE and Cycles, in the form of plugins. This should serve as a list of the major professional rendering plugins that are released or upcoming for Linux.

A BlendNet render farm consists of one Manager instance, which distributes tasks to multiple Agent machines.The blender Addon connects to the Manager and the Agents to schedule renders.See the BlendNet wiki for more information on BlendNet architecture.

LuxCoreRender is an open source rendering method that can also make use of OpenCL to render. To make use of it, simply install the (blender-plugin-luxcorerenderAUR) package and then enable the LuxCoreRender addon in the User Preferences box in Blender.

Pro-Render is an open source Blender rendering plugin from AMD that will allow any machine using an OpenCL 1.2 compatible AMD GPU the ability to create realistic GPU renders, allowing for faster work compared to the CPU.

Blend4Web is an open source framework for creating and displaying interactive 3D graphics in web browsers. It contains a Blender add-on to create and export 3D scenes directly into the web. A Blend4Web-specific profile can be activated in the add-on settings. When switching to this profile, the Blender interface changes so that it only reveals settings relevant to Blend4Web. See the documentation on how to install Blend4Web SDK.

Official blender (from the Arch repos) reports missing libraries (the first is libusd_ms.so which I can only find in blender downloads, though it should be installed from the usd, or at least usd-extra package, but no sign of it).

I can fix that by copying the library to /usr/lib. Then it complains of missing python libraries, and if I fix that, it complains of more and more missing libraries. I gave up, and removed all my hack handed fixes.

Oddly, I then went for blender-git (from AUR), which builds and runs. I had to install optix separately and rebuild the git version to get optix (which is out of date in AUR, but trivial to get the latest version, though I'm pretty sure I've never had to install optix discreetly before).

Herbert and Damien are right, the estructor topology is the cleanest topology for this, use that and not the boolean or my bastardized circle/square topology. Any panel in the blender gui can be changed. The properties looks like this.

Basically with boolean you create two objects and move them so they overlap in some way you want. then select one of the objects and add a boolean modifier to it. in the modifier you select the other object and the type (difference, union or intersection) after you apply it, on a difference for example, you need to hide the 2nd object in order too see the effect. It can result in some weird topology though and is somewhat fussy and unpredictable.

Thanks! I managed to make my arch today, and made a number of margin notes in the book so I can locate the properties/modifier buttons (they were off-screen). I then applied the same basic principle to make the curves in the wall and cut little tubes in the arch for the lights.

I am also having the same issue upon installing arch on a new system 3 days ago and trying to install blender. I was able to install and run it yesterday, but it seems the most recent update to blender has broken something.

Alternatively one could enable the testing repos and see if it is fixed in the newer version (17:4.1.0-3 has already marked outdated). I currently can't report bugs anymore as I did not migrate to gitlab with arch Perhaps someone could check.

Running into the same issue. I was able to use blender without issue yesterday, but today it keeps giving the exact same error. Seems to be related to materialx getting updated 12 hours ago (as of writing). The version on blender's website seems to work, even with my AMD RDNA3 card which previously was borked.

Has anyone reported this issue to the package maintainers on the Arch gitlab instance? Has anyone tried rebuilding the package to see if that resolves the issue? I tried rebuilding 17:4.1.0-3 in a clean chroot without testing enabled and it failed during the final link:

When not using precompiled libraries, the preferred way to installdependencies under Linux is to use the install_linux_packages.pyscript in the Blender sources. It will try to install system packagesfor all known dependencies, required or optional.

This is not recommended for most users, as this is mostly used by Blenderplatform maintainers and not well tested on other systems. You may need todig into the CMake code yourself to solve build problems.

The make deps systemwill build libraries from source and put them in ../lib/linux_x86_64_glibc_228.The Blender build system automatically picks up libraries from this locationover precompiled libraries.

To install Blender into the system directories, a system installationcan be used instead. Disable WITH_INSTALL_PORTABLE to install intoCMAKE_INSTALL_PREFIX which uses a typical UnixFHSlayout: bin/, share/blender/, man/ etc.

Remove the files created by make install since they will becomestale. However, Blender still needs to be able to find its scripts oryou will have a very minimalist looking interface. This can be solved bylinking the source directory to your Blender binary path so copyingisn't needed and scripts are always up to date.

after rsyncing all /opt in the other fs removing the local one and adding a symlink again, it now can be launched, but crash, there are already bug reports about it, so I can consider this specific problem is solved.

I still wonder why for, some AUR package installation override the symbolic link, when there is /opt/... in the path. Is this related to a specific right to the directory or an option to add to makepkg.conf, or at install time?

I think the issue is being caused by issues with Intel's OneAPI base toolkit, I fixed the problem by removing blender, then removing intel-oneapi-basekit-2024.0.0.49564-2 (the OneAPI developer toolkit thing), and then installing blender again which installed:

Download Blender directly from their website; that version works. The Blender in Arch's repository also doesn't support Embree on GPU (hardware raytracing), which the website version supports. The only plus side of using the repository version is that I don't have to manually upgrade it, but if this kind of problems keep happening, I would just switch to the website version and won't use the repository version.

I'm on Arch Linux 6.0.12, I installed the ROCm developers tools from the rocm-arch package in the arch4edu repo, and my GPU (Radeon RX 5700 XT) and my CPU (Ryzen 5 3600) are fully detected by Blender. I'd like to get better rendering so I'd be thankful to anyone who helps me to solve this problem.

When compiling this package, I got a prepare() error and warnings about non allowed 'file' transport and non-ability to clone source files into cache folders... What can I do?

Also, I noticed there's an "out-of-date" mark on the blender-rocm AUR package since the 2022-12-08.

The opencl-amd package didn't work with me, I got both my GPU and CPU recognized by Blender but it wouldn't detect the hipcc compiler. Kinda forced to try with blender-rocm.

Do anyone know how to revert the

The log snippet doesn't show why the build stopped, but I'm guessing you built against opencl-amd ?

That build fails with "sh: line 1: /opt/rocm-5.4.1/llvm/bin/clang: No such file or directory" .

I noticed there are now rocm packages in community-testing repos, so it seems someone is trying to migrate rocm from aur to repos.

Maybe it's smartest to wait a bit until that migration has finished,

I checked Blender several times in a week with the blind hope to get it working with magic, so yes, I would say for sure I restarted my computer. I even remember "updating" the PATH variable by myself.

i updated amdgpu-pro-libgl recently and didn't rebuild blender. i need blender to run with GPU support, so i can edit .obj files produced from a ML pose-recognition network. i was planning on using these to make some gifts for family using a CNC (windows only of course).

what's very frustrating: the amdgpu-pro-libgl package is missing. version 3.5 doesn't build for the reasons stated above and apparently introduces an openmp dependency -- and if you don't know, that means to expect quite a bit of disruption in builds. think pipewire, but more short-lived because it only affects users who care about CUDA and run AMD GPU's with pro drivers. i would expect quite a few builds to be breaking over the holidays while people are using spare time to work on projects in arch (which is NOT in any way supported by AMD afaik).

all in all, the OpenMP stuff is a huge move forward -- if you have a datacenter or a $50,000 homelab. it will also help quite a bit when combining GPU's from multiple vendors on a single machine. but why can't i still install amdgpu-pro-libgl?

Rendering on rx6500xt does not work for me either. I installed the blender-rocm package as recommended by the archwiki. While rendering via hip on integrated intel GPU works flawlessly, on discrete GPU I get the same error. I thought that support for hip and rocm is already built into blender from the standard repository, but nowhere is it mentioned exactly which packages need to be built to support rocm on free amd drivers. I mean which are sufficient for his work. If you share information on this issue, I will be grateful, since English is not my native language and it is problematic for me to deal with the stack of these technologies.

3a8082e126