LevelFinder
How LevelFinder Works
LevelFinder combines AI-powered image recognition with a curated library of puzzle solutions to help you find the right walkthrough in seconds. Here is a step-by-step breakdown of how each part of the platform works.
Step 1: Upload a screenshot
The core feature of LevelFinder is its screenshot solver. When you upload a screenshot of your current puzzle board, the system analyzes the visual layout to identify the types and positions of game elements — vehicles, passengers, and platform features.
This analysis creates a compact representation of your board arrangement that acts as a fingerprint for your specific layout. The fingerprint captures what is on the board and where everything is positioned, regardless of your device or screen resolution.
For the best results, upload a screenshot taken at the very start of the level, before any moves have been made. The full board should be visible, and the image should not be cropped or have overlays covering the game area. Screenshots from the pause menu or results screen will not produce accurate matches.
Step 2: The system finds a match
Your board fingerprint is compared against every entry in the level library. Each library entry was built from a known walkthrough video using the same analysis process, so the comparison is consistent — the same features are measured in both your screenshot and the reference.
The system scores each library entry based on how closely it matches your board layout. The entry with the highest similarity score is selected as the best match. In most cases, the correct walkthrough stands out clearly from the alternatives.
The matching process typically completes in under five seconds, even on mobile connections.
Step 3: Watch the walkthrough
After matching, you are taken to a results page that shows the closest walkthrough video for your board. The video is embedded directly on the page so you can watch without leaving the site. You can also open the video on YouTube if you prefer.
If the match does not look right — which can happen if your board is partially obscured or if multiple levels share a similar layout — you can go back and upload a different screenshot. Taking a cleaner screenshot from the start of the level usually resolves mismatches.
Browsing the level guide
If you already know which level you are on, you can skip the screenshot solver entirely and use the level guide. The guide organizes every supported level into browsable groups of 100, with a search bar that lets you jump directly to any level number.
Each level has its own dedicated page with the walkthrough video, a difficulty badge, the content source, gameplay tips specific to that level range and difficulty, and navigation buttons to move to adjacent levels. This makes it easy to work through a sequence of levels without going back to the guide each time.
How the library is built
The level library is built from community walkthrough videos. Each video is processed by the same detection model used for screenshot matching, which extracts a reference payload for every level shown in the video. These payloads are stored alongside metadata like the level number, difficulty category, and video source.
The library is curated to avoid duplicates and to ensure that only clear, high-quality references are included. When multiple walkthroughs exist for the same level, the one with the clearest board view and most complete solution is selected. The Crowd Express library currently contains over 4,000 level references.
Privacy and data handling
LevelFinder processes your uploaded screenshots only for the purpose of finding a match. Images are sent to the detection server, processed, and discarded — they are not stored permanently or used for any other purpose. No account or personal information is required to use the solver.
The site uses session cookies to manage upload rate limits and prevent abuse, but these do not track your identity or browsing behavior across sessions. For more details, see the privacy policy.