I've made artifacts/image.png: a 512 × 512 PNG of the IU_0453 bunny, drawn on a 32 × 32 grid and scaled up 16× so every pixel is a sharp square. A script check confirmed the size and format, that every 16 × 16 block is one solid colour (no smoothing, gradients or blur), and that only allowed colours appear. I also looked at the rendered image.

What's in it: one swamp-green (#7cc84a) bunny, front facing and centred, with its shoulders running off the bottom edge. Every shape has a thick black outline.

• Ears: long, with pink insides.
• Antenna: grey stalk with a bright green bulb on top.
• Eyes: 2 × 2 black dots, each with a one-pixel white highlight.
• Face: pink cheeks and nose, and a small "w" mouth.
• UI TEE: blue tee with a darker blue collar and sleeves, and a tiny white "UI" logo on the chest. That logo is the only text-like mark.
• Background: solid #0c1610 with six single-pixel dark-green specks.

Requirements not fully met:

• Black outline: #000000 isn't in the listed palette. I used it only because the brief asks for a thick black outline. The other 9 colours are the background and 8 palette colours.
• Collection match: I couldn't check that it fits the 787-piece set because no other pieces were available, so someone should review the style by eye.

I didn't use the image-generation tool. I drew the image with a Python script (Pillow) so the grid and palette come out exact, and it gives the same result every run. The script is tools/make_iu_0453.py and the README with these details is tools/README.md; both are committed. The image itself is left uncommitted, as instructed, for the daemon to upload.