✦ how we test
Our Scoring Method
Same test for everyone, reviewers who can't see brand names, and rules public enough for you to check our math. Here's exactly how every score on this site is made.
What we measure
| Dimension | Weight | What it means |
|---|---|---|
| Image quality | 40% | Realism, composition and flaws, blind-scored across 50 prompts |
| Follows your prompt | 25% | Does it render exactly what you asked — counts, text, positions? |
| Editing control | 20% | Can you fix and iterate without the image falling apart? |
| Price & value | 15% | Real cost per image at production volume (re-checked quarterly) |
The test, step by step
Every tool gets the same 50 prompts at default settings. Three reviewers score the outputs blind — no tool names attached. The final score is the weighted average. You can recompute any score on this site from these rules.
Independence
We buy our own subscriptions. Companies don't see scores before we publish, and there are no sponsored spots. PickerPetal is part of a network of independent review publications — the full list of related publications is on our about page, and they do not represent independent confirmations of one another.