GPT2 x Specimen Poetics Atlas
By xiaoxiaodong · P241
GPT2 x Specimen Poetics Atlas
By xiaoxiaodong · P241
Showcase
Description
Suited to turning an object into a poetic classification page styled like a museum publication: visual research, exhibition catalogs, and conceptual posters. The image shows only part of the subject as a precise black-and-white photograph; the missing portion is completed with a large expanse of untouched white space and a fragile, imperfect hand-drawn outline. A few square material samples serve as evidence. Feeling words label intangible qualities, not anatomical explanations. Extreme negative space, restrained asymmetry, small bold lowercase sans serif, a thin border line, black-and-white photographic grain, and slight printing imperfections. Quiet, cerebral, tender, archival. Avoid full-color illustration, dense catalog grids, cute stickers, smooth vector graphics, and explanatory infographics. The original text is in English and preserved verbatim in zh-CN.md.
Recommended by
淡苍
Why we love this skill
Transforms photographs into poetic black-and-white specimen plates, pairing partial images, unfinished outlines, material samples, and sensory text in a restrained editorial language.
Related Skills
View all
ImageGPT2 x Negative Monument
For turning everyday photos into minimalist independent-publication illustrations: exhibition studies, personal photography projects, and literary visuals. The upper half retains a recognizable version of the original photo; the lower half does not depict the subject, but instead builds a field of dense, slightly shaky, broken horizontal strokes, leaving the subject as a huge, quiet expanse of negative space. A tiny, naïve figure contrasts with, passes through, or accompanies this monument. Use no more than three colors: a warm white paper ground, one soft, bright ink color drawn from the source image, and black. The image must use flat graphic forms while retaining paper texture and slight misregistration. Avoid volumetric rendering, gradients, shadows, smooth digital rendering, and flashy multicolor treatments.
ImageGPT2 x Gothic Miniature
Ideal for turning strong or complex photos into fragile, understated picture-book metaphors for literary visuals, independent publishing, and exhibition pieces. The upper half retains a recognizable version of the original image; the lower half keeps only its most symbolic features, forming a small, isolated figure. Surround it with an enormous amount of empty space and, at most, one extremely faint cast shadow. Opaque gouache with thin watercolor, dry-brush marks, pencil, uneven blacks, and hesitant edges. The main colors are charcoal black, ivory white, and soft gray, with warm peach or pale apricot used only as tiny accents. A little dark but not frightening, a little awkward but not cutesy. Avoid smooth finishes, polished CG, commercial cartoons, complex backgrounds, and cinematic lighting.
ImageGPT2 x Indie Paper Object
For turning everyday photos into the kind of handmade paper piece you might find in an independent bookstore, design museum shop, or small art market: personal photo curation, independent publishing, exhibition exercises, and lifestyle brands. The upper half preserves a recognizable version of the original photo; the lower half does not apply a paper-cut filter, but removes irrelevant information and reworks it into a small paper-art object using naive hand-drawing, paper-cut color blocks, rough geometric shapes, and a few simple lines. The overall look is clean, friendly, optimistic, relaxed, and slightly quirky: a medium-to-small scale, generous warm-white space, matte paper and Risograph grain, and 2–3 spot colors sampled from the original image. Avoid realistic illustration, smooth vector graphics, a glossy 3D plastic feel, complex backgrounds, commercial packaging, children's craft-class aesthetics, and templated brand design.
Information
- Version
- v4
- Last updated
- Runtime credits
- Usage-based
- Models
- Auto
