Explore production-ready prompt templates and screenprint recipes tailored for 1-of-1 apparel. Switch between flat-lay garments and raw artwork, then make a print directly in the studio.
GPT Image rewards structure. Scene, Subject, Important details, Use case, Constraints, in that order, each heading on its own line. These are the same five headings the studio sends, so what you write here is what the model reads. Slot 5 is where unconstrained prompts fail silently, by letting the model invent a background you then have to print.
Scene: A standalone graphic on a 1:1 square canvas with a fully transparent background. Subject: Roaring timber wolf surrounded by electric static coils. Important details: 3-color screenprint (cadmium red, pitch black, bone cream), heavy halftone dot grain, 0.5pt vector contours, distressed vintage crackle finish. Use case: Chest center graphic for heavyweight streetwear tee. Constraints: Pure transparent background, zero rectangular bounding box, no watermarks, no extra text.
Zero vague praise tokens detected. Every slot contains measurable physical constraints that the diffusion model can draw cleanly.
Where the artwork lives. For a shirt graphic that is the canvas itself: a transparent field at your chosen ratio, plus only the environment that belongs inside the art. No room, no floor, no sky.
The hero and what it is doing. Form, pose, action, materials, expression. One focal thing, then at most three motifs that support it without crowding the silhouette.
Technique, named inks, outline weight, composition and lettering. Cadmium red and bone cream, 0.5pt contours, coarse halftone. This is also where quoted words become typography.
The artifact the model is drawing for. Here it is always a print-ready front graphic for direct-to-garment printing, which is what keeps the output flat art rather than a photo of a tee.
What must not drift. Transparent background, complete subject in frame, no watermark, no garment mockup, no extra words, no duplicate text.
Excitement does not render. Telling the model an idea is “epic”, “stunning”, or “8K” gives it no actionable pixels. Visual facts give the diffusion model concrete physical surfaces to draw.
“A stunning ultra-detailed cinematic masterpiece of a streetwear tiger t-shirt, beautiful, photoreal, 8K, award-winning, insane graphic design, trending on artstation.”
Battle-tested disciplines adapted from the official fal.ai GPT Image 2 specification to eliminate generative hallucinations and keep prints production-ready.
Adjectives like 'epic' or 'masterpiece' waste attention tokens. Replace praise with physical materials, lighting angles, and print textures.
Stacking buzzwords creates contradictory noise. Give abstract style tags concrete physical parameters like contrast, margin spacing, and stroke weight.
If the apparel graphic needs an embroidered patch, call it an embroidered patch. If you want a 1990s band tour tee, name the cracked wash explicitly.
Diffusion models re-roll unconstrained pixels. Every edit must declare what to change AND an explicit preserve checklist (pose, colors, typography).
Wrap literal words inside straight double quotes ("...") and mark them EXACT TEXT. State typeface family, weight, colour, size against the hero, and placement. Add 'no extra words, no duplicate text'. If a character keeps ghosting, spell the word out letter by letter after it.
Compound edits that request lighting, wardrobe, typography, and color changes in one go fail silently. Make one deliberate revision at a time.
Diffusion models allocate the highest attention weight to opening phrases. State your central character or object and its active pose before any style adjectives.
Only text inside "..." becomes lettering on the shirt. Never describe text in prose like 'with words saying XYZ' or it will render garbled gibberish.
Do not ask for skies, rooms, or landscape floors. If you describe an environment, the model draws a colored square box around your art, destroying the print silhouette.
Garments are viewed from a distance. Overly cluttered scenes lose punch and fill up negative space needed for breathability on combed cotton blanks.
Real apparel design workflows fall into three operational modes: text-to-image from scratch, single-variable edits that split what changes from what is locked, and multi-image compositing where every reference is labelled by role.
Original screenprint designs, custom typography stamps, risograph posters, and standalone apparel graphics.
Scene: A standalone graphic on a 1:1 square canvas with a fully transparent background. Subject: Cybernetic kabuto helmet flanked by dual cherry blossom branches. Important details: 2-color screenprint (cadmium red and bone cream), sharp vector lineart, 300 DPI halftone dot shadows, distressed cotton texture. Use case: Chest center graphic for oversized heavyweight streetwear tee. Constraints: Pure transparent background, zero bounding box, no watermarks, no extra text, no camera glare.
Colorway changes, typography adjustments, removing stray artifacts, or updating ink textures on existing art.
Change: Replace the red circular sun behind the mecha with a glowing crescent moon. Match the ink texture and halftone dot scale of the moon to the existing screenprint. Preserve: Keep the mecha character, the chest lettering "SHOGUN", the line weights, the palette and the transparent background exactly as they are. Constraints: Change only the sun. No extra objects, no redesign, no new text, no duplicate text.
Virtual try-on, applying raw artwork onto 3D garment mockups, style transfer, and character consistency.
Image 1: base person photo to preserve. Image 2: raw graphic artwork to print. Image 3: washed charcoal heavyweight tee reference. Instruction: Dress the person from Image 1 in the charcoal tee from Image 3 with the graphic from Image 2 printed crisp across the chest. Preserve the person's face, hair, posture, lighting, and background from Image 1 exactly. Fit the fabric naturally with realistic folds, drape, and contact shadows.
Precision parameters that take a raw creative prompt to a production-ready apparel print with 300 DPI fidelity.
Canvas dimensions dictate chest coverage, margin isolation, and how diffusion tokens allocate detail.
Unlike posters or flat prints, textile direct-to-garment (DTG) printing demands pure vector edges, controlled halftone grain, and high opacity ink so graphics remain vivid across 100% heavyweight combed cotton blanks.

Placing keywords inside straight double quotes converts plain phrases into bespoke typography integrated seamlessly into the illustration.

Every design is normalized through clean backdrop isolation, preventing colored bounding boxes from muddying the garment.
Multi-prompt composite generation and narrative style references tuned for screenprint dots.


Demonstrates complex lighting, circuit detailing, and verbatim dual-word lettering (“2077” and “SAMURAI”) rendered cleanly on black combed cotton.
A cybernetic samurai mask with intricate neon circuits, with "2077" and "SAMURAI" lettered in electric cyan.
Test the live regex extraction engine. Words placed inside straight double quotes ("...") become lettering on the shirt. Everything outside quotes is interpreted strictly as imagery description.
Isolated graphic centred on a fully transparent background, edge to edge, with no ground, field or backdrop of any kind behind it. Clean silhouette designed for direct-to-garment shirt printing with a sharp outer boundary; every part of the subject carries its own colour so it stays readable on any garment colour.
<role>
You are the graphic art director at a premium streetwear label. A maker hands you a rough t-shirt idea; you hand back one vivid, decisive brief that an image model will draw as the shirt's front graphic.
</role>
<guidelines>
- Use the exact section order Scene, Subject, Important details, Use case, Constraints, with each heading on its own line.
- Scene: state the transparent canvas and only the environment that belongs inside the artwork.
- Subject: open with the focal hero, its pose or action, expression and attitude.
- Add one to three secondary motifs that support the hero without crowding it.
- Name exact colour tones (oxblood, bone, faded teal, safety orange), never vague words like vibrant or colourful.
- Visual facts, never praise: no stunning, epic, masterpiece, gorgeous, 8K, award-winning. Write the surface, the light, the linework, the ink.
- Every style word needs a visual target. Not 'minimalist brutalist editorial', but the weight, the palette, the negative space and the one hero shape that make it so.
- Name the real thing. A transit kiosk, a boarding pass, a kabuto, a wing nut. Object names get drawn; mood words get interpreted.
- Typographic lockup only when wording strengthens the piece: keep any wording the maker put in double quotes exactly, put every intended phrase in double quotes, mark it EXACT TEXT, and treat it as typography — typeface family, weight, colour, size against the hero, and placement — saying how the lettering weaves into the graphic. If words add nothing, leave the piece wordless.
- Spell any invented, foreign or easily-mangled word letter by letter in brackets right after it, so the model cannot ghost a character: "KAIJU" (K-A-I-J-U).
- Write in the chosen style's own vocabulary and honour its direction.
- Use case: a print-ready front graphic for direct-to-garment t-shirt printing.
- Constraints: a fully transparent background, complete subject in frame, no watermark, no garment mockup, no unrequested text, no extra words and no duplicate text.
- Punchy, concrete, sensory. Cut corporate filler, cut adjectives that don't change the picture, and describe what to draw rather than listing things to avoid.
- Original work: no real logos, campaign artwork, brand names or copyrighted characters.
</guidelines>
<output_format>
Return only one JSON object: {"prompt": "Scene:\n...\n\nSubject:\n...\n\nImportant details:\n...\n\nUse case:\n...\n\nConstraints:\n..."}. Use 80-220 words, no markdown and no commentary.
</output_format>Battle-tested prompt recipes covering streetwear screenprints, transparent cutouts, verbatim lettering, virtual try-ons, and risograph panels. Copy with one click or launch directly into the studio.
Scene: A standalone graphic on a 1:1 square canvas with a fully transparent background. Subject: A wolf howling atop a satellite dish with electric static crackling off its fur. Important details: Vintage 3-color screenprint (cadmium red, pitch black, bone cream), coarse halftone dot grain, 0.5pt vector linework contour, distressed crackle texture. Use case: Chest center print for heavyweight 100% combed cotton streetwear tee. Constraints: Pure transparent background, zero rectangular bounding box, no watermarks, no extra text, no camera glare.
Change: Cut the garment out of Image 1 and re-issue it as a standalone product graphic with clean vector edges and a fully transparent background. Preserve: Keep the collar ribbing, fabric weave, seam stitching, chest print sharpness, garment proportions and colourway exactly as they sit in Image 1. Constraints: Transparent background, no halo, no white fringing along the contour, no restyling, no added objects, no watermark. A contact shadow only if it respects the alpha channel.
Scene: A standalone lettering graphic on a 4:3 landscape canvas with a fully transparent background. Subject: The headline "HYPER DRIVE" (H-Y-P-E-R D-R-I-V-E) as EXACT TEXT, set as the entire artwork. Important details: Futuristic distressed stencil lettering in a heavy geometric sans-serif, tight clean kerning, pitch black faces with a neon cyan and safety orange chromatic offset behind them. Set across the full chest width with generous negative space above and below. Use case: Print-ready front graphic for direct-to-garment t-shirt printing. Constraints: Render the wording verbatim, every character spelled correctly. Fully transparent background, no extra words, no duplicate text, no background box, no watermark.
Image 1: base subject photo to preserve. Image 2: graphic tee artwork reference. Image 3: washed charcoal heavyweight tee reference. Instruction: Dress the subject from Image 1 in the charcoal tee from Image 3 with the graphic from Image 2 centered across the chest. Preserve the subject's face, facial features, skin tone, hair, posture, background, and lighting exactly. Replace only the garment. Fit the fabric naturally with realistic drape, collar fold, seam contours, and contact shadows. Do not add accessories, jewelry, extra logos, or watermarks.
Change: Turn the pencil sketch in Image 1 into a production-ready vector screenprint graphic, rendered in opaque white and fluorescent orange plastisol with clean vector edges and 300 DPI micro-halftone dot shading. Preserve: Keep the exact layout, silhouette proportions, eye placement and contour lines of the sketch. The drawn layout is a contract, not a suggestion. Constraints: Fully transparent background, complete subject in frame, no added background elements, no buildings, no decorative clutter, no rectangular border, no watermark.
Change: Replace the orange ink in the tiger illustration with metallic gold ink. Match the ink sheen and micro-crack finish to the existing screenprint texture. Preserve: Keep the black contour lines, the bone cream base, the typography, the chest alignment and the transparent background exactly. Constraints: Change only the orange ink. No extra objects, no redesign, no new text, no duplicate text.
Scene: A standalone graphic on a 2:3 portrait canvas with a fully transparent background. Subject: Swiss brutalist monolithic concrete elevation with structural wireframe coordinates and technical cross-sections. Important details: Heavy black condensed sans-serif on a strict Swiss typographic grid, micro coordinates set in small monospace. The headline "ARCHITEKTUR" (A-R-C-H-I-T-E-K-T-U-R) as EXACT TEXT, stacked full-width above the elevation. High-contrast ink, orthogonal rules only. Use case: Print-ready oversized back print for direct-to-garment printing on heavyweight cream cotton. Constraints: Fully transparent background, strict orthogonal grid, no curved distortion, no extra words, no duplicate text, no watermark.
Scene: A standalone graphic on a 1:1 square canvas with a fully transparent background. Subject: Retro 1980s Japanese anime mecha pilot climbing into a cockpit. Important details: 3-colour risograph printing in fluorescent pink, cobalt blue and sunflower yellow, overlapping halftone rosettes, deliberate ink misregistration, paper-tooth grain kept inside the shapes. The headline "MECHA WARRIOR" as EXACT TEXT in heavy condensed sans-serif, arched over the cockpit. Use case: Print-ready front graphic for direct-to-garment printing on a vintage-washed tee. Constraints: Fully transparent background, no digital gradients, mechanical risograph artifacts only, no extra words, no duplicate text, no watermark.
Picking a style preset adjusts two distinct pipeline stages: the art direction injected after the subject, and the print execution parameters.
Versatile, true to the idea. clean, versatile illustration rendered true to the idea, no fixed genre.
Chiaroscuro, real anatomy. realistic rendering with anatomically accurate form and believable materials, lit with high-contrast chiaroscuro: one strong key light, deep resolved shadow.
Cel-shade, manga ink. anime illustration with dynamic manga line art of varied weight, cel-shading in two clean shadow steps, and vibrant ink fills.
Hand-drawn, layered fills. hand-drawn illustration with confident linework and flat layered colour fills.
Clay, octane light, dimension. stylized 3D render with tactile dimensional forms, studio clay and octane-style lighting, and soft ambient occlusion in the crevices.
One line, lots of air. minimal single-line or flat-shape composition with generous negative space.
Type-led, layered lockup. type-led graphic poster in which oversized display lettering and the illustration occupy the same space — the lettering set behind, around or through the subject so the two lock into one composition rather than reading as a picture with a caption under it.
Every creator account receives 7 free renders per week. Cost scales with quality tier and resolution.
| Tier | Cost (Renders) | Free Quota Policy | Best Used For |
|---|---|---|---|
| Low | 1 render | Covered by the free weekly quota at every size. | Drafts and rapid concept exploration |
| Medium | 1 to 2 renders | Covered by the free quota at every size except Wide 4K. | Everyday prototyping and silhouette testing |
| High | 3 to 7 renders | Never covered by the free quota. Always spends a render. | Production graphics and standard garment prints |
| X-High | 5 to 13 renders | Never covered by the free quota. Always spends a render. | High-detail apparel with fine lines and sharp lettering |
| Max | 11 to 28 renders | Never covered by the free quota. Always spends a render. | Archival gallery renders with microscopic fidelity |
Bring your own concepts to life, test typography formulas, and print 1-of-1 custom direct-to-garment apparel with zero minimum order quantities.