ANNPLOT
Module 0 · Full ComfyUI to SVG/HPGL Runbook
Full instruction set

Run ComfyUI on the Acer and turn its output into SVG and HPGL.

This module is a complete command-line runbook. It avoids hidden assumptions, shows what to run, what you should expect before and after each step, and ends with a working local workflow: prompt → image → SVG → HPGL.

Start at preflight Use the included scripts
End result

ComfyUI server + export path

By the end of this module, the Acer should be able to run ComfyUI locally, generate a text-to-image result, and export the latest image to both SVG and HPGL.

HostAcer Ubuntu machine
GPURTX 3060 12 GB
AI UIComfyUI
Plot pathSVG → HPGL
Overview

What this module includes

This is not just a note page. It is meant to be a stable ANNPLOT reference page you can return to later or share as a free course module.

What you will do

  1. Check the Acer and the NVIDIA path
  2. Install or update ComfyUI properly
  3. Download SDXL-Turbo as the first text-to-image checkpoint
  4. Start the ComfyUI server
  5. Generate a first image
  6. Export the latest image to SVG and HPGL

What this module does not hide

  • Dependency installation is explicit
  • Fresh install and update paths are separate
  • The model download is a separate step
  • The raster-to-vector export path is separate from ComfyUI itself
Preflight

Do not install blind

Run the checks first. If these fail, fix them before moving on.

Command

./scripts/00_preflight_checks.sh

What you should expect before running it

  • You are on the Acer
  • You have terminal access
  • The scripts folder from this module exists locally

What you should expect after running it

  • You should see Ubuntu release information
  • You should either see nvidia-smi output or learn it is missing
  • You should see Python, pip, git, and disk space status
Install

Base packages and ComfyUI

Use the fresh install path if ComfyUI is not already installed. Use the update path if it is already there.

Fresh install

1) Install base Ubuntu packages

./scripts/01_install_base_deps.sh

Before: You may not yet have git, ImageMagick, or potrace.

After: You should have python3, python3-venv, python3-pip, git, imagemagick, and potrace.

2) Install ComfyUI manually

./scripts/02_install_comfyui_manual.sh

Before: Either no ComfyUI exists yet, or you want to build a clean venv under ~/ai/ComfyUI/.venv.

After: You should have a ComfyUI repo, a Python venv, PyTorch installed inside that venv, ComfyUI requirements installed, and a torch/CUDA verification printout.

Existing install

Update the existing ComfyUI install

./scripts/03_update_existing_comfyui.sh

Before: ~/ai/ComfyUI and ~/ai/ComfyUI/.venv already exist.

After: Your ComfyUI repo is pulled forward and Python requirements are refreshed in the correct environment.

Model

Download the first text-to-image checkpoint

This module uses SDXL-Turbo as the first model because it is a strong, fast first checkpoint for getting a local text-to-image path working.

./scripts/04_download_sdxl_turbo.sh

Before: ComfyUI must already exist and its venv must already be created.

After: The file sd_xl_turbo_1.0_fp16.safetensors should exist in ~/ai/ComfyUI/models/checkpoints.

If the download fails because of Hugging Face access or license acceptance, do that first and then run the same script again.

Run

Start the ComfyUI server and generate a first image

This uses the default text-to-image workflow path from ComfyUI, then the generated file is processed outside ComfyUI for SVG/HPGL export.

Start the server

./scripts/05_run_comfyui.sh

Before: ComfyUI and the model must already be installed.

After: A ComfyUI server should be listening on port 8188.

Inside ComfyUI

  1. Open the browser UI
  2. Load the default Image Generation workflow
  3. Select the SDXL-Turbo checkpoint in the checkpoint loader
  4. Use a short black-and-white prompt
  5. Queue one generation
black and white sacred geometry made of precise ink lines, radial symmetry, clean white background
Export

Convert the latest ComfyUI output into SVG and HPGL

This is the full post-process path: latest PNG/JPG from ComfyUI → thresholded bitmap → Potrace SVG → vpype HPGL with a custom 7470A-style config.

Install the vector pipeline

./scripts/06_install_vector_pipeline.sh

Before: Base Ubuntu packages should already be installed.

After: A separate vpype venv exists and a custom ~/.vpype_7470a.toml config is copied into place.

Export the latest image

./scripts/07_export_latest_png_to_svg_hpgl.sh my_first_export

Before: There must already be at least one image in ~/ai/ComfyUI/output.

After: You should get:

  • ~/ai/annplot_exports/my_first_export.svg
  • ~/ai/annplot_exports/my_first_export.hpgl
Included files

This module ships with helper scripts

  • 00_preflight_checks.sh
  • 01_install_base_deps.sh
  • 02_install_comfyui_manual.sh
  • 03_update_existing_comfyui.sh
  • 04_download_sdxl_turbo.sh
  • 05_run_comfyui.sh
  • 06_install_vector_pipeline.sh
  • 07_export_latest_png_to_svg_hpgl.sh
  • hp7470a_custom.toml