What the gallery is
Open the skills menu in any chat and tap Custom Photo Inspiration. What loads is a grid of hand-picked photo ideas across categories — loungewear, beach, morning, undressing, sports fandom, cosplay, group shots, and everyday moments. Each tile shows a preview render of the scene plus the exact prompt beneath it, so you can see what you're getting before you commit.
Why specific beats generic
Default AI photos feel same-y because the underlying prompt is same-y. "Portrait of a woman in a cozy setting, warm lighting, cinematic" produces one narrow band of results across every generation. The catalog approach flips it: instead of the model picking the scene, the operator has hand-picked dozens of scenes with real detail — "sitting on the edge of the bed pulling her hair up," "on a rooftop pool deck in a linen wrap dress," "reaching back to unzip her dress." Specific inputs produce specific outputs.
How the render actually works


When you tap a tile, Sloane fires an image-to-image call to Grok Imagine using your companion's avatar as the identity reference. The tile's prompt drives the scene; the reference locks her face, hair color and texture, eye color, skin tone, and body type. The output is that scene rendered with HER identity — not a stand-in, not a random face, actually her.
The two shots below show it end-to-end. Left: the preview on the tile the user tapped — the composition rendered against a stand-in persona so you can see what the scene will look like. Right: the actual delivered photo, Monique in the same fitting room, the same biker jacket, but her identity throughout. Same scene, her face.
Categories in the gallery today
The current catalog spans loungewear (boyfriend-shirt, cotton pajamas, silk robe), beach (boat deck, linen sundress at a beach bar, walking barefoot on wet sand), morning (bathroom doorway with damp hair, edge of the bed pulling her hair up), undressing (unzipping her dress, loosening her tie, sliding one strap off her shoulder), sports fandom (team jersey at the stadium, courtside at a game), cosplay (Lara Croft in a jungle temple, Chun-Li mid-kick, Wonder Woman full costume), group shots (sorority composite, bachelorette in Vegas, bridesmaid line-up), and everyday moments (making pancakes in a band tee, reading in a window seat). Each category has multiple tested prompts.
Cost and gating
Custom Photo Inspiration is a Plus feature. Each generation costs 5 credits — the same as any custom photo request in chat. Free accounts can browse the gallery and preview any tile; generating with your companion's actual face requires Plus. Tap a tile as a Free user and you get the full preview plus a one-tap upgrade path.
What happens when a prompt gets blocked
Grok Imagine's moderation is stochastic on borderline scenes — the same prompt may pass on one attempt and get blocked on the next. Sloane retries up to three times before giving up. If all three attempts get moderated, you see a friendly note ("Our photo model vendor blocked this scene — no credits charged") and can pick a different tile. Credits are only deducted after a photo actually lands in your chat; a moderated request costs nothing.
How prompts get chosen for the catalog
The catalog is operator-curated. Each prompt is tested against multiple companions before being approved — different body types and skin tones can trigger the moderator differently, so anything unreliable gets pulled. The prompts on the live gallery are the ones that passed testing across at least three diverse companions on the first try. This means the failure rate you see in practice is low: most of the moderator-blocked wording has already been filtered out during curation.
