🔒 Nothing leaves your LAN 📦 No models bundled 🖼️ Batch + gallery 🌓 Light / dark · EN · 中文

1 · Pick photos

Select one or many images from your phone's gallery.

2 · Send to ComfyUI

Your server runs the saved workflow over the LAN, one image at a time.

3 · Save results

Results download to Pictures/ComfyUI and the in-app Gallery.

The app ships no AI models — it uses whatever workflow and models are already installed on your own ComfyUI server. You control the workflow, the model, and your data.

1

Prepare the ComfyUI server

The app talks to ComfyUI's HTTP API, so ComfyUI must accept cross-origin requests and listen on your LAN — not just localhost. Every command below is also shown, with a copy button, inside the app's Setup Guide.

1.1Import the example workflow

The app runs a saved workflow — it doesn't build one for you. Below is the example it expects: a FLUX.2 Klein 9B KV reference-conditioning graph that edits the picked photo from the text in the CLIP Text Encode node (e.g. "Change the color of clothes to red").

Example ComfyUI image-editing workflow
Example image-editing workflow — click to open full size.
⬇ Download the workflow image

Load it into ComfyUI and save it:

  1. Drag-and-drop the image onto the ComfyUI canvas (or Workflow → Open). The full node graph is embedded in the PNG, so ComfyUI rebuilds the whole workflow.
  2. Save the workflow (Workflow → Save) as image_flux2_klein_9b_kv_image_edit.json — the app's default workflow name, so it appears in the Workflow dropdown after you Connect.
  3. The image input is node #76 (Load Image) and the output is node #94 (Save Image → images) — the app's default Input/Output ports, so no extra configuration is needed.

The models referenced (flux-2-klein-9b-kv, qwen_3_8b CLIP, flux2-vae) must already be installed in your ComfyUI. Edit the prompt text to change what the edit does.

1.2Start ComfyUI for LAN access

python main.py --enable-cors-header --listen --port 8189

--port can be any free port; whatever you choose here is what you enter in the app.

1.3WSL on Windows: enable mirrored networking

If ComfyUI runs inside WSL, add this to C:\Users\YourName\.wslconfig:

[wsl2]
networkingMode=mirrored

Then run wsl --shutdown and restart ComfyUI.

1.4Find the host computer's LAN IP

Run the command for the OS hosting ComfyUI and use the IPv4 address of the active Wi-Fi/Ethernet adapter.

OSCommand
Windowsipconfig
Linuxhostname -I · ip -4 addr
macOSipconfig getifaddr en0 (or en1)

Using an emulator on the same machine? Use 127.0.0.1, or 10.0.2.2 from the Android emulator.

1.5Allow the port through the firewall

Windows

netsh advfirewall firewall add rule name="ComfyUI 8189" dir=in action=allow protocol=TCP localport=8189

Linux (UFW)

sudo ufw allow 8189/tcp

Linux (firewalld)

sudo firewall-cmd --add-port=8189/tcp --permanent
sudo firewall-cmd --reload

macOS — the firewall is app-based; allow the Python/Terminal app that starts ComfyUI.

2

Configure the app

Open the Editor tab and tap Configuration (the dialog also opens automatically on first launch).

  1. ComfyUI host IP and Port — enter your PC's LAN IP (step 1.4) and the --port from step 1.2. Tapping a field clears it for quick entry.
  2. Connect & load workflows — the app fetches the workflow list from the server; a status line shows how many were found, or the connection error.
  3. Workflow — pick your image-editing workflow from the dropdown.
  4. Input / Output port — pick the workflow's image input node (a LoadImage-style node) and its image output node. Selecting a workflow auto-fills sensible defaults.
  5. Save.

The chosen workflow must contain an uploadable image input (a widget named image/upload) and at least one IMAGE output — exactly the ports offered in the dropdowns.

3

Process photos

  1. In the Editor, tap Select Photos and choose one or more images.
  2. Tap Send to ComfyUI. Photos process one at a time with a Processing X / Y progress bar. Each cycle: upload → queue → poll (up to 15 min) → download → save to Pictures/ComfyUI.
  3. Each job card shows the original on the left and the ComfyUI result on the right, plus its status (Queued / Uploading / Editing / Saved / Failed). Failed jobs have a Retry button.
  4. Open the Gallery tab to review results; tap any image for full-screen.
Editor tab with two selected photos queued for processing
Queued photos, ready to send.
Photos processing one at a time — first result saved, second still editing
Processing one at a time — original left, result right.

Jobs live in memory only — if the app process is killed the job list and Gallery view clear, but images already saved to Pictures/ComfyUI stay on disk.

4

Settings

ThemeSystem · Light · Dark
LanguageSystem · English · 简体中文 · 繁體中文
Remove adsA monthly subscription or one-time lifetime purchase removes the banner and interstitial ads. Restore purchases from the upgrade sheet.
5

Troubleshooting

SymptomLikely cause / fix
"Cannot connect to ComfyUI"Wrong IP/port; ComfyUI not started with --listen; firewall blocking the port; phone not on the same LAN.
Workflow dropdown emptyConnected, but no .json workflows under workflows/, or you haven't tapped Connect & load yet.
Input port dropdown emptyThe workflow has no uploadable image input (image/upload widget). Pick a workflow that loads an image.
Job fails immediatelyThe workflow errored on the server — the ComfyUI message shows on the job card. Confirm it runs in the ComfyUI web UI first.
Result never arrivesLong-running workflow — the app polls up to 15 minutes before timing out.