general/other Comfyui installed with Podman have to install models next to test

[NapoleonWils0n]

Just installed Comfyui using Podman with Cuda support

GitHub - Comfy-Org/ComfyUI: The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface.

The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface. - Comfy-Org/ComfyUI

github.com

Have to installs some models next to try everything out
takes a while to download the models

Also installed the Comfyui manager

GitHub - Comfy-Org/ComfyUI-Manager: ComfyUI-Manager is an extension designed to enhance the usability of ComfyUI. It offers management functions to install, remove, disable, and enable various custom nodes of ComfyUI. Furthermore, this extension prov

ComfyUI-Manager is an extension designed to enhance the usability of ComfyUI. It offers management functions to install, remove, disable, and enable various custom nodes of ComfyUI. Furthermore, th...

github.com

 

[NapoleonWils0n]

Eureka i got Comfyui working with Cuda in a Podman container

You need 2 fixes

1) In the container .bashrc we export COMMANDLINE_ARGS

export COMMANDLINE_ARGS="--xformers --disable-dynamic-vram --medvram"

I use medvram because i only have 4gb of vram

2) we have to comment out some code for comfy_aimdo which sets dynamic vram

vi ComfyUI/.venv/lib/python3.12/site-packages/comfy_aimdo/control.py

we comment out the following code and just add the block of code above it

def init_devices(device_ids):
    return False

#def init_devices(device_ids):
#    global devctxs
#
#    if lib is None:
#        return False
#
#    requested = [int(device_id) for device_id in device_ids]
#    if not requested:
#        return False
#
#    if not lib.plat_init():
#        return False
#
#    device_array = (ctypes.c_int * len(requested))(*requested)
#    if lib.init(device_array, len(requested)):
#        devctxs = [get_devctx(device_id) for device_id in requested]
#        return True
#
#    devctxs = []
#    lib.plat_cleanup()
#    return False

I downloaded this model

aria2c -o sd3_medium_incl_clips_t5xxlfp8.safetensors "https://huggingface.co/adamo1139/stable-diffusion-3-medium-ungated/resolve/main/sd3_medium_incl_clips_t5xxlfp8.safetensors?download=true"

Heres a screen shot of the workflow

Heres the image that it created

which took 9 minutes 17 seconds on my Dell XPS 15 - 2019
with a NVIDIA GeForce GTX 1650 with 4 gig of vram with Cuda 13.2

16 gig of ram using Driver Version: 595.58.03 on Freebsd 15

Its 1024x1024
but you can use real-esrgan-ncnn-vulkan to upscale the image

I converted the image to a jpg for the upload