Turn Any Instagram Carousel Template Into Your Own

What Ribbi did

Ribbi starts with a structured breakdown of the reference — three vertical narrative panels (work → close-up → collapse), a dual-tone palette of deep maroon and teal-gray, hand-drawn linework with grain texture, that urban-youth exhaustion — then uses the original as the style source and the full breakdown as the generation brief, delivering a 1696x2528 recreation in one pass: panel structure, palette, stroke and mood all aligned, ready to post.

What Ribbi did

Ribbi condenses the ask into three variables (style, theme, count), then designs a three-panel storyline for each — "commute hell" with the subway crush and spilling crowds, "late-night convenience store" with fluorescent light and curbside noodles, "weekend hermit" from lying in to scrolling till dark — each with its own shot angles and warm-cold contrast plan. On approval, all three 1696x2528 pieces batch out at once, serial coherence and single-image polish both held.

What Ribbi did

A new batch of three themes arrives — dreamscape, medieval knight, forest adventure. Ribbi first confirms "same single color for all three, or one each?", locks in teal, and re-delivers: the asymmetric three-panel split, hand-drawn sketch lines and grain texture stay, the color system collapses into a teal monochrome scale, and the closed-in → close-up → release emotional camera language carries through, grouping three themes under one visual language.

What Ribbi did

Ribbi separates "regenerate" from "crop", first confirming the target is the teal monochrome batch, then switching to a crop-only editing path: three pieces go from 2:3 portrait to 2048x2048 square with composition weight preserved and zero change to the artwork — carousel-ready as is.

What Ribbi did

Ribbi reverse-engineers the reference into a complete English prompt, dimension by dimension — composition (asymmetric three-panel split, square format), style (hand-drawn sketch, crosshatching, indie comic feel), color (teal monochrome, heavy shadows, film grain) — then runs the user's named Midjourney v7 with a dual-image "story reference + style reference" input, generating four 1024x1024 variants to choose from: the same visual language, successfully reproduced on another model.