▚ IMAGINE IMAGE 2.0 ─ NOT THE VIDEO ONE ─ THE STILL-IMAGE BEAST ▞
xAI’s Imagine Image 2.0 dropped Aug 7 — the new Quality Mode for stills + editing, live inside Grok. It landed #2 on both blind arenas (Image-Edit and Text-to-Image), built on the Aurora engine.
grok.com/imagine → — free tier gives a limited daily quota; SuperGrok / Premium+ removes the cap.
The gems most people miss:
Smart Resize is free outpainting — approve one image, then re-frame it across 9 ratios (1:1 → 9:16 → 16:9 → 2:1). It paints the new space instead of cropping: one render becomes a Story, a banner, a thumb. New image each time — scan for warped text before you ship.
Background removal → transparent PNG, plus magic-wand + segmentation edits: recolour the bottle without touching the label, table, or lighting.
5 reference images, each with a JOB — the pro move isn’t “combine these,” it’s role-tagging: img1 = face · img2 = jacket shape+colour · img3 = background · img4 = lighting only.
One-click templates hiding in the menu — headshots, mascots, game sprites, UI kits, emoji, merch, product posters.
Honest bit: xAI published no text-accuracy % or native resolution — “sharper small text” is their claim until you test it. API access still “coming soon.”
!