Stable Diffusion on Jetson AGX Orin and Xavier

lekto · October 4, 2022, 7:25pm

Hi all,

I came up with a Dockerfile to run Stable Diffusion inside a contianer on Jetson.

Jetson AGX Orin and Jetson AGX Xavier with big VRAM space make it pretty straight forward to run Stable Diffusion, so have a try if you happen to have the hardware.

“Robots marching down a street in Japan”
grid-0001

dusty_nv · October 4, 2022, 8:26pm

Really cool, thanks for sharing @lekto!

lphung · October 6, 2022, 1:33pm

Hi @lekto . I tried this on my Jetson Nano but got the error bellow with “./docker/run.sh”. I am working on this now but it would be nice if you can point out what is wrong here.

lekto · October 7, 2022, 9:59pm

With only 4GB for RAM, I think it is almost impossible to run Stable Diffusion on Jetson Nano.

lphung · October 9, 2022, 4:10am

Yes, thank you for your information @lekto . I just want to run some tests with it.

seflands · October 21, 2022, 11:34am

Hi @lekto and @dusty_nv Would it be possible to use the xformers to speedup your image generation on Orin in a Docker container? Xformers · AUTOMATIC1111/stable-diffusion-webui Wiki · GitHub

dusty_nv · October 21, 2022, 4:29pm

Hi @lphung, if you check @lekto’s Dockerfile, that container image was built against JetPack 5.0:

github.com

chitoku/stable-diffusion/blob/de9538934367d991b63a8d7e769681c5871b6d4e/Dockerfile#L1


      
          FROM nvcr.io/nvidia/l4t-pytorch:r35.1.0-pth1.13-py3
          
          RUN apt-get update && \
              apt-get install -y python3 python3-pip
          
          RUN pip3 install transformers==4.19.2 diffusers invisible-watermark
          RUN pip3 install "numpy>1.19.3" "opencv-python-headless<4.3"  omegaconf
          RUN pip3 install einops 
          RUN pip3 install "pytorch-lightning==1.5.10"
          RUN pip3 install taming-transformers
          RUN wget https://raw.githubusercontent.com/CompVis/taming-transformers/master/taming/modules/vqvae/quantize.py -O /usr/local/lib/python3.8/dist-packages/taming/modules/vqvae/quantize.py

So you would probably need to rebuild it against the version of JetPack you are on with Nano (JetPack 4.x)

Hi @seflands, I’m not entirely sure if this is applicable or not, but here’s also this optimized version of DALL-E Mini for PyTorch:

sergei.sleptsov · December 27, 2022, 12:34pm

what about Jetson TX2? Does anybody tried to run Stable Diffusion there?

abhigoku10 · March 1, 2024, 5:54pm

@dusty_nv how to rebuild the container for jetpack 5.1 , i am trying to run stable diffusion on xavier nx development kit

@lekto where u able to run the stable diffusion on xavier nx if so what are the packages and its version used can you please mention it here

dusty_nv · March 2, 2024, 2:10am

@abhigoku10 you can find instructions for building containers from jetson-containers here:

github.com

dusty-nv/jetson-containers/blob/master/docs/build.md

# Building Containers

[`build.sh`](/build.sh) is a proxy launcher script for [`jetson_containers/build.py`](/jetson_containers/build.py).  It can be run from any working directory (in the examples below, that's assumed to be your jetson-containers repo).  Make sure you do the [system setup](/docs/setup.md) first.

To list the packages available to build for your version of JetPack/L4T, you can use `--list-packages` and `--show-packages`:

```bash
$ ./build.sh --list-packages       # list all packages
$ ./build.sh --show-packages       # show all package metadata
$ ./build.sh --show-packages ros*  # show all the ros packages
```

To build a container that includes one or more packages:

```bash
$ ./build.sh pytorch               # build a container with just PyTorch
$ ./build.sh pytorch jupyterlab    # build container with PyTorch and JupyterLab
```

The builder will chain together the Dockerfiles of each of packages specified, and use the result of one build stage as the base image of the next.  The initial base image defaults to `l4t-base`/`l4t-jetpack` (for JetPack 4 and JetPack 5, respectively) but can be changed by specifying the `--base-image` argument.

This file has been truncated. show original

The container images on DockerHub for stable-diffusion-webui are a month old: dustynv/stable-diffusion-webui:r35.4.1

abhigoku10 · March 4, 2024, 3:20am

@dusty_nv
Q1. thanks for sharing the link , i was able to setup the packages as mentioned by @lekto and was getting the following error

Is it related the the packages not downloading properly or something else ?

Q2. did u try running the GitHub - kuprel/min-dalle: min(DALL·E) is a fast, minimal port of DALL·E Mini to PyTorch at ur end was it successful ??
Q3 I am trying to any LLM and Stable diffusion models in Jetson Xavier , do u have any suggestions

Topic		Replies	Views
Does Stable Diffusion run on Jetson Orin Nano? Jetson Orin Nano generative_ai	15	8116	July 6, 2023
Total noob and i suck at this - guidance needed Jetson Orin Nano generative_ai	5	176	September 10, 2025
Stable Diffusion "run.sh" error on Jetson AGX Orin Jetson AGX Orin generative_ai	8	627	March 26, 2024
FLUX AI Image generation jetson-container Jetson AGX Orin cuda , python	5	1880	August 12, 2024
Stable Diffusion Tutorial failure Jetson Orin Nano generative_ai	12	1481	May 18, 2025
Installing stable-diffusion-webui - jetson-ai-lab.io Jetson Orin Nano generative_ai	3	478	September 10, 2025
Stable Diffusion container fully crashes my Jetson Jetson AGX Xavier containers , stable-diffusion-xl	2	86	February 12, 2025
Unable to build stable-diffusion-webui Jetson AGX Orin generative_ai	25	1059	February 2, 2025
Issue Trying to Install Stable Diffusion Jetson Orin Nano jetson	13	617	September 24, 2025
How to run Stable Diffusion v1.5 on Jetson AGX Orin Jetson AGX Orin	2	148	April 1, 2025

Stable Diffusion on Jetson AGX Orin and Xavier

Related topics