Feel From Footage: Measure Game Feel with FFmpeg

    1

    Settle game-feel arguments with numbers from timestamped FFmpeg tile strips of reference footage and your build.

    $5

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Feel From Footage: Measure Game Feel with FFmpeg

    Example session with this skill installed

    I have reference footage of a mining game in 'ref_mining.mp4'. I want to see how fast the pickaxe swings and how long the ore takes to drop after the final hit. Then compare it to 'my_build.mp4'.

    • Read your context and instructions
    • Compiled the feel from footage

    I have analyzed the metadata for ref_mining.mp4 and my_build.mp4. I generated 10 fps tile strips for both videos, focusing on a 4-second window of steady mining action. For the reference, I used a crop at 800:600:400:300 to isolate the character and the ore node.

    Game Feel Comparison Report

    EventReference (ref_mining.mp4)Your Build (my_build.mp4)DeltaLocation in Project
    Hit Interval (Median)0.4s (10 fps, 0:12.0-0:16.0)0.6s (10 fps, 0:05.0-0:09.0)+0.2s (Slower)Assets/Scripts/Mining/Config.json
    Hits to Break5 hits4 hits-1 hitMiningComponent.cs
    Hit to Drop Delay0.2s (2 tiles)0.5s (5 tiles)+0.3s (Slower)LootSpawner.cs
    Total Break Time2.2s2.4s+0.2sN/A

    Analysis Notes

    Reference: The reference footage shows a high-frequency "impact puff" every 4 frames. The drop appears almost instantly (200ms) after the final crack animation.

    Your Build: While you require fewer hits to break the object, the swing animation is 50% slower than the reference. The delay before the item spawns (500ms) creates a noticeable "dead gap" in feedback.

    • Verification: Reference wiki lists "Mining Speed: 2.5 hits/sec," which matches the observed 0.4s interval.

    Next steps

    1. Reduce swing_cooldown in Config.json from 0.6 to 0.4 to match the reference rhythm.
    2. Audit LootSpawner.cs for any WaitForSeconds calls that might be delaying the spawn effect.
    3. Re-run this analysis on the updated build to confirm the "snappiness" matches the target.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Stop arguing over whether a game mechanic "feels slow" based on memory or gut feeling. This skill uses frame-by-frame video analysis to extract concrete timing data from reference footage and your own game builds, turning subjective feedback into a table of numbers.

    What it does

    • Extract rhythm data by generating 10 fps or 30 fps tile strips from video files to visualize event timing.
    • Isolate action zones using FFmpeg crop filters to focus strictly on actors, targets, or UI elements.
    • Measure event intervals including hit speed, impact-to-feedback delay, and time-to-break counts.
    • Benchmarking your current build against reference footage using identical tile layouts for a direct comparison.
    • Verification of observed footage against official documentation or wikis to identify version drift.

    How it works

    1. Source Analysis: Point the skill to a local video file to probe metadata and create a low-fps overview contact sheet.
    2. Window Selection: Identify a 5-10 second window of steady action and define the pixel coordinates for the crop area.
    3. Strip Generation: Run FFmpeg commands to produce timestamped tile strips that serve as the "source of truth."
    4. Data Extraction: Count tiles to log specific event times (impact puffs, stage changes, item drops).
    5. Comparison: Measure your own build using the same process and generate a side-by-side comparison table.

    Frameworks & tools

    This skill requires ffmpeg and ffprobe installed on your system path. It processes local .mp4, .mov, or other standard video containers.

    Why this beats prompting it yourself

    Generic LLMs cannot "see" time or frame intervals accurately from a single image or description. This skill enforces a rigorous FFmpeg-based workflow that produces visual evidence (tile strips) for every claim, ensuring that timing measurements are mathematically tied to the video's framerate rather than hallucinated.

    Use cases

    • Tuning combat speed by measuring frame data from a genre-leading reference game.
    • Debugging input lag or feedback delay between an action and its visual effect.
    • Validating if a "hit to break" mechanic matches the intended balance design.
    • Documenting feel changes across different build versions during playtesting.

    Known limitations

    Requires FFmpeg installed on the command line. Cannot infer actual player input (e.g., button mashing vs. holding) without on-screen visual markers or external evidence.

    Requires local FFmpeg/ffprobe installation. Cannot definitively confirm player input intent (tap vs. hold) without visual input markers. Accurate to +/- 0.1s at 10 fps.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 7 days ago

    Permissions

    Write Files
    Read Files
    Terminal / Shell
    Listed7 days ago
    Updated6 days ago

    What's inside

    Frequently Asked Questions

    This skill eliminates subjective debate by using FFmpeg to convert video into frame-accurate tile strips, allowing you to count exact frames between an action and its visual feedback for objective benchmarking.

    You must have FFmpeg and FFprobe installed and accessible via your system's command line path, as the skill relies on these tools to process local video files like .mp4 and .mov.

    The skill provides a structured workflow and pre-configured commands to extract timestamped visual evidence, which prevents the hallucinations and timing inaccuracies common when asking generic AI to estimate video duration.

    The skill can analyze any standard video container supported by FFmpeg, but it cannot infer physical controller input unless your footage includes an on-screen input overlay or 'button-cam' visual markers.

    Your purchase includes the full agent skill logic, the specific FFmpeg filtering workflows for tile generation, and documentation on how to interpret the resulting frame data for game design.

    New in Game Development