⚔ Star Savior Substat Checker

Take a screenshot of your gear's substat panel in Star Savior, then either:

• Ctrl+V (or Cmd+V) anywhere on this page to paste from clipboard
• Click Paste from Clipboard button
• Drag and drop an image file onto the drop zone

OCR runs entirely in your browser — nothing is uploaded. After scanning, detected stats flash green in the manual entry grid. Always check the RAW OCR debug output if something looks wrong, and correct any misread values manually.

Set the Item Level (default 15) before or after pasting — every 3 levels adds 1 upgrade roll, so a max-level item has 5 upgrade rolls distributed randomly across its 4 substats.

At item level 15 this would be interpreted as: Crit Rate hit 1 upgrade (base 1–2%, upgraded to 2–4% range, rolled 3.20%), Crit DMG hit 1 upgrade (base 1.5–3%, upgraded to 3–6%, rolled 4.50%), ATK flat hit 2 upgrades (base 15–45, rolled near max each time), Effect Hit 0 upgrades (single roll, 1.40% is near minimum of 1.0–3.0%).

The tool estimates how many upgrade rolls landed on each stat by dividing the observed value by the single-roll maximum. For example, Crit Rate max per roll is 2.0% — a value of 3.20% implies floor(3.20 / 2.0) = 1 upgrade hit, so 2 total rolls.

You can override the inferred hit count using the small number input on each result card (it turns yellow when overridden). The upgrade summary line tracks how many of the total upgrade rolls are accounted for.

Each substat can only roll one of 6 discrete values per roll — the range is divided into exactly 6 equally-probable steps:

For a stat that rolled N times, the tool computes the exact probability distribution of the sum using convolution — multiplying the single-roll distribution with itself N times to enumerate all possible outcomes. The percentile is then P(random N-roll sum ≤ your value), smoothed by linear interpolation between discrete steps.

Example: Crit Rate 1.6% on 1 roll — exactly 3 of the 6 outcomes (1.0, 1.2, 1.4) are below it, so it scores at the 50th percentile. Crit Rate 3.20% on 2 rolls — the sum ranges from 2.0 to 4.0 across 36 combinations, and 3.20% sits around the 60th percentile of that distribution.

The overall score answers a different question: "compared to any random item at this upgrade level with these same 4 stats, how lucky is this one?"

This is computed by running 50,000 simulated items. Each simulation:

1. Distributes the upgrade rolls randomly — each roll has equal probability of landing on any of the 4 stats
2. Draws values from the discrete roll tables for each stat based on how many rolls it received
3. Computes that simulated item's average per-stat percentile

Your overall percentile = fraction of those 50,000 items that scored lower than yours.

This captures both dimensions of luck: getting upgrades concentrated on your best stat is itself lucky, and rolling high values on those upgrades is a separate layer of luck. An item where all 5 upgrades landed on Crit Rate and rolled 2.0% each time is genuinely rarer than one where upgrades spread evenly.

If you check 10 random items you should see roughly 1 in S tier, 1–2 in A tier, and 5 in B/C tier — the distribution is calibrated so the percentile reflects true rarity.