Every tool gets a score from 0 to 100, computed by code from verified facts with
the published weights below. Hands-on test results are blended in at 50% once a test is published. Affiliate
links, sponsorships and ownership never change a score.
The factors
Accessibility: Can a beginner start for free without installing or coding?
Output & integration: Where can the results go: export formats, platforms, engines?
Commercial rights: Can you ship what it makes, and on which plans?
Value: How much does it cost to get real use out of it?
Maturity: Is it a stable product, a beta, or a research demo?
Game focus: Is it built for game development, or a general tool that can also make games?
Features: How much of its category's feature checklist does the vendor confirm it has?
Weights by category
Different jobs need different things. Commercial rights matter most for audio, and integrations matter most for coding tools.
Factor
AI game makers
Art, 3D & animation
Coding & engine tools
Audio, voice & NPCs
Accessibility
26%
17%
13%
21%
Output & integration
17%
21%
21%
17%
Commercial rights
13%
21%
13%
21%
Value
9%
9%
9%
9%
Maturity
9%
9%
13%
9%
Game focus
13%
9%
17%
9%
Features
15%
15%
15%
15%
Rules we follow
Facts come from primary sources. Each tool page lists its sources and a last-verified date.
Anything we haven't confirmed shows as "Not listed", "Not confirmed" or "See site for pricing", never as a yes or a no.
Unknown never helps a tool. An unverified fact scores below the midpoint, so verifying a tool
can only move it in the direction the facts point. The four key facts are free tier, price, commercial use and exports; a tool missing
some of them scores lower until they're confirmed. Vendors can send a primary source to fill a gap.
Features are checked, not assumed. Every tool has a feature checklist for its category (for
example pixel art, sprite sheets and 8-direction views for sprite tools). A feature is marked Yes, Partly or No
only when the vendor's own site, docs or pricing page says so; everything else shows as "Not confirmed". Features
are 15% of every score: Yes counts fully, Partly counts 60%, No counts nothing, and "Not confirmed" counts 40%, so
an unconfirmed feature never helps more than a limited one.
Scores are computed, not chosen. The scoring code is open in our repository, and nobody
adjusts a number by hand.
Hands-on tests are real. A hands-on score exists only after we publish the test, with the
prompt and outputs. Until then, pages say the score is facts-based.
Discontinued tools stay listed, but unranked, so people searching for them find the
replacement.
Freshness is automated. Agents re-check pricing, status and changelogs every week and propose
changes with sources. Automated checks catch bad or missing data before it goes live.
Ownership and money
This site is run by the Makko AI team, and Makko is scored by exactly the same rules. Some links are affiliate
links, and sponsored placements are labeled. See the full disclosure.