LostSideDead Year 2026

Created with ACMX2

Created a podman container hosted on ghcr.io for the classic Acid Cam apps. You can find the repo here: acidcam-dev and a step by step Guide.

Worked on tightening up the Linux launch workflow for ACMX2 / acidcam-gpu using a Podman container. This makes installation simpler and avoids dependency problems on the host, while still allowing full access to the GPU, camera, audio, and desktop display.

ACMX2 is distributed as a container and requires an NVIDIA GPU with the proprietary drivers installed on the host. This build uses CUDA, so it will not run on AMD or Intel GPUs.

System requirements (host):

Linux (x86_64)

NVIDIA GPU + NVIDIA proprietary drivers

Podman

NVIDIA Container Toolkit (Podman support)

X11 or XWayland

Webcam device (example: /dev/video0)

Audio input device (microphone)

Quick start (container):

podman pull ghcr.io/lostjared/acmx2:latest
cd podman
chmod +x run-acmx2.sh
./run-acmx2.sh

The launch script handles the annoying parts automatically: NVIDIA GPU passthrough, camera and audio device access, and opening the ACMX2 interface window on the desktop via X11/XWayland.

Shared files workflow:

I am using a shared folder on the host for quick transfers (videos, models, shader packs, output files). Files placed in ~/container_share on the host appear inside the container at /root/share.

Troubleshooting (driver check):

nvidia-smi

NVIDIA license notice:

This container includes NVIDIA CUDA libraries and is subject to the NVIDIA Deep Learning Container License. By pulling or running the container, you agree to NVIDIA’s license terms.

First post of 2026, took a short break. I have been working on a new software project target for Linux with CUDA. Here is a demo of the app running. Stay tuned for further updates.