What Gladdihopper Actually Is and When You Should Use It
Gladdihopper is a tool built by members of the speedrunning community, specifically for people who hunt glitches in games. It automates the tedious process of frame-by-frame analysis that you'd otherwise do by hand. If you've ever spent three days trying to verify whether a certain sequence of inputs causes a buffer overflow in a game's hitbox detection, you already know why this exists. The basic workflow works like this: you record gameplay with frame data visible, feed it into Gladdihopper, and the tool scans through each frame looking for anomalies — things like sprites rendering in impossible positions, collision boxes shifting without input, or memory addresses being written to outside normal parameters. It doesn't guess. It reports what it finds and flags inconsistencies.
How to Get and Set Up Gladdihopper
You download it from the official Gladdihopper GitHub repository. The current version requires you to have Python 3.10 or higher installed, along with a few dependencies. I know that sounds like a barrier, but the dependency list is short — basically OpenCV and NumPy. Installation takes about four minutes on a decent connection. Once installed, you need to configure your input. Gladdihopper expects a video file with a frame counter overlay. Most people use RTSS or MSI Afterburner for this. The key detail nobody mentions in the README is that the frame counter needs to be in the top-left corner, white text, monospace font. If your overlay is in the bottom-right or uses a proportional font, the tool will misread frame numbers and your entire run gets corrupted. I wasted an afternoon on this before figuring it out. Run the executable, load your video file, set your region of interest coordinates, and hit start. The tool will process frame by frame. For a standard five-minute clip, expect roughly ten to fifteen minutes of processing time on a modern CPU. A dual-core processor from 2018 will take closer to forty minutes.
What Gladdihopper Misses and Where It Breaks
Here's the part most guides won't tell you: Gladdihopper only catches visual and memory-level glitches. It does not detect logical glitches — things like state machine desynchronization where the game's internal clock gets out of step with player input but nothing visually breaks. If you're hunting for frame-perfect strafe tricks or input-buffer exploits, this tool is useless to you. I ran into this limitation head-on while analyzing a particularly stubborn tool-assisted trick in an older Mario title. The glitch I was hunting didn't produce any anomalous visual data. Gladdihopper reported every frame as normal because, visually, everything was fine. The issue was purely in how the game's save-state rollback mechanism handled a specific memory write. What I ended up doing was pairing Gladdihopper output with a custom Lua script that monitored the game's RAM addresses directly. That combination caught the issue in about two hours instead of the three weeks I'd been spending on it manually. Another hard limitation: Gladdihopper struggles with emulators that don't provide deterministic frame output. If your emulator has frame skipping enabled, variable frame timing, or uses a different execution core than the one the tool expects, the frame counter becomes unreliable and the analysis breaks. Always run with frame locking on and use the exact emulator version listed in the documentation. I learned this the hard way when my initial results were completely inconsistent between two runs of the same clip.
Getting Meaningful Results Without Wasting Hours
The biggest mistake beginners make is feeding Gladdihopper an entire playthrough. The tool is designed to analyze specific segments, not whole runs. Your best results come from isolating the exact moment you suspect a glitch occurs and running a twenty-to-sixty-second clip around that timestamp. This cuts processing time dramatically and reduces false positives. Use the region-of-interest feature aggressively. You don't need the whole screen analyzed. Narrow your ROI to the area where the glitch would manifest — usually the character sprite and immediate collision radius. This can cut processing time by half and eliminates noise from background elements that might trigger false flagging. If you're working with a ROM hack or a modded version of a game, Gladdihopper's default signature database won't match. You'll need to generate your own signature file using the provided template. The documentation on this is sparse, but the process is straightforward: dump the relevant memory regions from your game, hash them, and swap the signature file in the config directory. I've seen people skip this step and spend hours wondering why the tool returns zero results. It's not broken — it just doesn't recognize your game's memory layout.
The tool outputs a CSV file with every flagged anomaly. Don't just scroll through it looking for dramatic entries. The useful data is often buried in rows with small numerical deviations. A sprite position shift of 0.03 pixels might look insignificant until you correlate it with an input frame and realize it lines up exactly with a known memory corruption pattern. Cross-reference the flagged frames with your input log. That's where the actual discovery happens.
