Image Crawler Troubleshooting

Troubleshooting

The page says "The model isn't running"

The AI model isn't loaded. Close the app's console window and launch it again from the launcher. If it happens again, read the console window and the log files (below).

The console window shows red text

Read the red lines. The common ones:

Message	What to do
The model isn't downloaded yet	The setup didn't finish downloading the model. In the launcher, open the app's ⋯ menu and click Repair (run setup again); the download resumes
No llama-server.exe found	The program that runs the model is missing. Reinstall with llama.cpp (CUDA 13 build) ticked, or install LM Studio
Python is not on PATH	Use Repair in the launcher, which sets up Python and the app's environment again
did not become ready within 180 s	The graphics card is busy or out of memory. Close games, other AI apps or models loaded in LM Studio, then launch again
GPU 1 is not available … continuing without it	Just a warning: that card was busy, so the app uses the others

The install stopped while downloading the model

The model is 9.7 GB. Use ⋯ ▸ Repair (run setup again): it carries on from where it stopped.

Nothing matches, or too much does

Edit the checklist (how): add unwanted things to EXCLUDE, or make REQUIRE less strict. Check Save if confidence ≥ and Min size too.

A website gives few or no pictures

• Some sites block automatic visitors or need you to log in; the app can't see those pictures.
• Respect robots.txt skips pages the site asks robots not to visit.
• Raise Max pages and Max link depth to go further into the site.
• Stay in this folder keeps the app out of pages above or beside the start address; untick it to roam the whole site.
• "Too many requests" under Problems means the site is slowing the app down; raise Delay between pages.

It's slow

About 1 to 2 seconds per picture is normal. A second NVIDIA card with about 16 GB doubles the speed. Min size skips small pictures before they reach the model.

Log files

In the app's folder under logs\: llama-server-gpu<N>.log.err (the model, one per card) and crawler-app.log (the app). Attach them when you report a problem.