Play game
SUBJECT_03's itch.io pageResults
| Criteria | Rank | Score* | Raw Score |
| Technical Execution | #32 | 2.211 | 2.444 |
| Gameplay & Core Idea | #38 | 2.161 | 2.389 |
| Overall | #41 | 2.090 | 2.311 |
| Polish & Presentation | #42 | 2.161 | 2.389 |
| Creativity & Theme Interpretation | #55 | 2.010 | 2.222 |
| Scope & Completion | #73 | 1.910 | 2.111 |
Ranked from 18 ratings. Score is adjusted from raw score by the median number of ratings per game in the jam.
Team members and roles
IGOR YAKUNIN
How did you interpret the themes?
SUBJECT_03 is a genre collision on purpose. For 13 chambers it plays as a machine-learning evolution simulator: you build a creature from nodes and bones on a blueprint, and a real population search - genuinely simulated, generation by generation, at 1× speed - teaches your exact structure to walk, climb, jump and crawl. Then TEST 14 flips genres: the evolution is certified and locked, the lab starts demolishing the ceiling, and the game becomes an arcade dodge-em-up where you steer your own evolved creature through falling debris with a Balance Override. The simulation genre and the action genre collide inside one corrupted research terminal - which is itself the third collision: a 1994 lab-document aesthetic wrapped around a 2026 WebGL physics sim.
What are you entering?
Engine and tools used?
TypeScript, Three.js (WebGL2, custom ordered-dither / scanline / glitch post-processing), cannon-es (2D-locked rigid-body physics), Vite. All sound effects and music are synthesized at runtime with the Web Audio API - no audio files.
Road to GDC
I am not entering this prize track
Bezi Challenge
We are not entering this prize track
Synty Challenge
I am not entering this prize track
Which Synty assets did you use?
N/A
CrazyGames Challenge
Enter my game in the CrazyGames Challenge
Fan Art Artist
N/A
Fan Art Process Video
N/A
Where can we find your Fan Art?
N/A
Submission Confirmation
Leave a comment
Log in with itch.io to leave a comment.







Comments
Dont know how you guys managed to make a neural networkable thingamajig, but i love this work and its really simple to understand what to do
Watching a real evolution search figure out how my janky four-node creature should move is incredibly satisfying — my first fully-braced design failed with "STRUCTURE INADEQUATE", and pulling out one diagonal suddenly gave me a lateral-drag walker. Building the body and letting the sim teach it to walk is a brilliant pairing, and I love that the last chamber supposedly flips the whole thing on its head.
This one has me hooked. I'm definately gonna keep this one open in a browser, and revisit it over the next couple days. Hoping to be able to write a much more extensive impression/review then.
But bravo!