Download

Get LayerShop

A small installer for the editor. The first time you open AI Gen, a wizard sets up local AI: ComfyUI, its Python, the models you pick and the workflows that drive them.

LayerShop 0.1.0 Preview

Windows 10 and 11, 64-bit. Free. Released October 2026.

installer .msiportable .zipSHA-256 checksums
Download for Windows

The download page lists the installer, a portable zip and their checksums. macOS is planned; Linux is undecided.

Install in three steps

  1. Run the installer. It installs the editor only, which takes a few seconds. Prefer not to install? Unzip the portable build anywhere and run layershop.exe; its settings stay in the folder beside it.
  2. Open AI Gen. Start LayerShop and choose Window › AI Gen. With nothing set up yet, the panel offers Set up local AI. That is the wizard.
  3. Pick a model set and wait for the download. The Starter set is 18.5 GB. When the wizard shows its test image, you are done: write a prompt and press Generate.

You can use the editor while models download. Closing LayerShop pauses the download; it carries on from where it stopped the next time you start.

The setup wizard

The wizard exists so that you do not have to know what ComfyUI, a virtual environment or a text encoder is. Click through the steps below to see what it does.

LayerShop Setup: Local AI

A walkthrough of the wizard. Click through the steps.

Checking your PC

Nothing is changed in this step. LayerShop looks at what you have and says what will work.

Windows 11, 64-bitsupported
NVIDIA GeForce RTX 4090Graphics driver is recent enough24 GB
Free disk spaceModels can go on any driveD: 412 GB
ComfyUI not foundLayerShop will install a private copy, with its own Pythonwill install
Download servers reachableonline
Already use ComfyUI? Choose Use my ComfyUI instead.

What each step does

  1. Your PC. Reads your Windows version, graphics card and its memory, free disk space, and whether a ComfyUI is already installed or running. Nothing is changed. If your card cannot run local models, the wizard says so here and offers cloud models instead.
  2. Location. Two folders: one for the engine (about 6 GB) and one for models (15 to 26 GB each). Models can live on another drive. Nothing is installed system-wide, and no administrator rights are needed.
  3. Models. Three presets, or tick models one by one. Each row shows its size and licence, and the wizard marks models your card is too small for. Files that several models share are counted once. A few models are published behind a licence click-through on Hugging Face; for those the wizard opens the page, you accept the licence there, and paste an access token. Skip that and the rest still installs.
  4. Install. Downloads a private Python, ComfyUI and the model files, checks each file against its published checksum, and writes the settings LayerShop needs. Interrupted downloads resume.
  5. Ready. Starts the engine, loads a model and renders a small test image, so a problem shows up now and not in the middle of your work.

The first step offers three routes. Set up local AI for me is the one described above. Use my ComfyUI and Cloud models only are described further down.

What goes where

Everything LayerShop adds to your computer is in one of these places.

WhatWhereSize
The editorC:\Program Files\LayerShopor wherever you unzipped the portable build~0.2 GB
Settingspreferences, model list, API keys, the engine's log%APPDATA%\LayerShopsmall
EngineComfyUI and its own Python%LOCALAPPDATA%\LayerShop\engineor the folder you chose~6 GB
ModelsThe folder you chose in the wizard15–26 GB each

The wizard does not install Python for Windows, does not change PATH, and does not touch a Python or ComfyUI you already have.

Already have ComfyUI?

Choose Use my ComfyUI on the wizard's first step. LayerShop then installs nothing of its own:

  • If your ComfyUI is running, LayerShop finds it and reads how it was started. If it is not, it looks for the installation ComfyUI Desktop recorded, or you point it at main.py and python.exe.
  • Scan Server lists a model for every model file your server has, with the recommended settings filled in.
  • For recommended models you are missing, the wizard offers to download just those files into your models folder.

A ComfyUI that is already running is used as it is and is never stopped by LayerShop. The portable build listens on port 8188 and ComfyUI Desktop on 8000; both are detected. More on the ComfyUI page.

Cloud models only

No suitable graphics card, or no wish to download 20 GB? Choose Cloud models only. Then open Edit › Preferences › Generative AI › API Keys and paste a key from OpenAI, Google, xAI, Meta or Black Forest Labs. Those models appear in the same list in the panel.

Cloud generations are billed by the provider to your account, and your prompt and picture are sent to them. The panel shows an estimate before you generate where the provider publishes prices, and what the last image cost.

Troubleshooting

Windows says "Windows protected your PC"

Preview builds are not code-signed yet, so SmartScreen does not recognise the publisher. Choose More info, then Run anyway. To be sure the file is the one that was published, compare its checksum with the one on the download page:

PowerShell
Get-FileHash .\LayerShop-0.1.0-windows-x64.msi -Algorithm SHA256

A download stopped or failed

Open the wizard again from Preferences › Generative AI › Local Server. Finished files are kept, a partial file continues from where it stopped, and a file that fails its checksum is fetched again.

Not enough disk space

Go back to the Location step and put the models folder on another drive, or pick the Starter set. The wizard will not start a download that cannot fit.

"Starting ComfyUI…" and then an error

The first start takes about half a minute. If it fails, the message quotes the last lines of the engine's output; the whole log is comfyui.log in the settings folder. The usual causes are an out-of-date graphics driver and another program holding the port.

ComfyUI Desktop will not start while LayerShop is open

Both want the same port. Press Stop Server in Preferences › Generative AI › Local Server, start Desktop, and LayerShop will use Desktop's server from then on.

LayerShop crashed and something is still using the GPU

An engine that LayerShop started keeps running if the app crashes. The next launch finds it and uses it. To stop it, end the Python process in Task Manager.

Anything else: report it, and include comfyui.log if the problem is with local models.

Updating and removing

  • Update the editor by installing the new version over the old one. Your settings, engine and models stay.
  • Add or remove models in Preferences › Generative AI › Manage Models. Removing a model there offers to delete its files as well, unless another model still needs them.
  • Remove local AI with the button of that name in Preferences › Generative AI › Local Server. It deletes the engine folder and, if you tick the box, the models folder, and tells you how much space comes back.
  • Uninstall the editor from Windows Settings › Apps. This leaves the engine and models in place, because they are large and you may want them back; remove them first, or delete the two folders above by hand.