MossImage
Prompt library

MossImage Prompt Library: Tested Prompts That Render Text Correctly

38 copy-paste prompts for menus, posters, signage, packaging and UI. The showcase section is generated with MossImage (MOSS-Image-SFT); verified cards include the official MOSS-Image-Base image and published parameters; templates are labelled and shown without a fake preview.

Updated 2026-10-20Open weights · Apache-2.0Hosted: MOSS-Image SFT

Attribution

The eight MossImage showcase images were generated with MossImage (MOSS-Image-SFT) from the prompts shown on each card. The twenty verified images are official MOSS-Image-Base examples published by the OpenMOSS Team under Apache-2.0 (attribution: “OpenMOSS Team, MOSS-Image, Apache-2.0”). Prompt text and parameters for the verified cards come from the repository's per-image parameters.json.
MossImage showcase 8Chinese text & signage 6Posters & UI 5Long English text 4Ads & product 7Portraits & art 8

Diagnosis

Why most prompts garble text

Text rendering is not a font problem. It is a composition and conditioning problem.

The words are not quoted

A model can read “a sign about coffee” as a mood, not as a string. Put the exact copy in quotes or in a JSON text item, and keep punctuation exactly as it should appear.

One block carries too much

A paragraph in a single box has to fit, wrap and align all at once. Split headings, body columns, prices and captions into separate text items so each one has its own box.

Style words fight legibility

Glow, bloom, motion blur, heavy grain and extreme perspective all damage glyph shapes. Ask for a flat surface, soft even light, no glare and a clean typeface when the text matters.

Line breaks are missing

The model can wrap text on its own, but the result is unpredictable. Insert real newlines or \\n escapes at every line you care about, and say “preserve every line break”.

The prompt is too long

Text encoding is capped at 4096 tokens. Very long layouts get truncated, and the missing tail usually contains the words you care about. Shorten the prose before you shorten the text items.

Coordinates do not match the canvas

Layout boxes use centered pixel coordinates. If the JSON was written for 1024×1024 and you render 1536×1024, the boxes drift. Change the coordinates or render at the original size.

8 prompts

MossImage showcase

Cinematic key art poster for the fictional film The Last Signal, with a radio astronomer and a dense billing block
MossImage showcaseMOSS-Image-SFT

The Last Signal — film key art

A cinematic key art poster with a five-line billing block, festival laurel and title treatment printed into the artwork.

1536×2048 · 50 steps · CFG 3

Create a theatrical key art poster for a fictional science-fiction thriller titled THE LAST SIGNAL. Vertical 3:4 poster, printed on matte paper with a subtle film-grain texture, designed for a streaming service and a cinema lobby. Scene: a lone radio astronomer in a navy wool coat stands on the observation deck of a re…

Generated with MossImage (MOSS-Image-SFT)

COMMON magazine cover with a model in a knit sweater, cover lines and a barcode
MossImage showcaseMOSS-Image-SFT

COMMON — magazine cover

A masthead half-hidden behind the model, three cover lines, issue details, price and a real barcode.

1536×2048 · 50 steps · CFG 3

Create a cover for a fictional independent culture magazine called COMMON, issue 47, Autumn 2026, sold in the US and UK. Vertical 3:4, printed on heavy uncoated paper with a subtle paper fibre, scanned straight-on with no perspective distortion. Cover photograph: an androgynous model in a sculptural oatmeal knit sweate…

Generated with MossImage (MOSS-Image-SFT)

The Blue Counter diner menu with two columns of dishes, prices and a legal line
MossImage showcaseMOSS-Image-SFT

The Blue Counter — diner menu

Two columns of dishes and prices with dotted leaders, plus a tiny food-safety line at the bottom edge.

1536×2048 · 50 steps · CFG 3

Create a straight-on photograph of a printed diner menu for a fictional Portland restaurant called THE BLUE COUNTER. Vertical 3:4, the menu is a single cream card with a thin cobalt blue border, lying flat on a dark walnut table beside a glass of water and a small steel sugar caddy. Soft, even morning window light from…

Generated with MossImage (MOSS-Image-SFT)

Still Mountain osmanthus oolong tea pouch with Chinese and English label text
MossImage showcaseMOSS-Image-SFT

Still Mountain — bilingual tea pack

English front label, Chinese product name and seal, and a full ingredients and brewing block on the side panel.

1408×1408 · 50 steps · CFG 3

Create a premium product photograph of a fictional specialty tea brand called STILL MOUNTAIN, a 30 g stand-up pouch of osmanthus oolong, sold in a San Francisco boutique. Square 1:1, centered composition on a warm travertine surface with a soft sage-green plaster backdrop. A single matte kraft-and-sage pouch stands upr…

Generated with MossImage (MOSS-Image-SFT)

Pulse SaaS revenue dashboard with a chart, summary cards and a top accounts table
MossImage showcaseMOSS-Image-SFT

Pulse — SaaS dashboard UI

A full SaaS screen with sidebar navigation, chart labels, five table rows and a 10 px legal footer.

2048×1152 · 50 steps · CFG 3

Create a photorealistic screenshot of a fictional B2B analytics SaaS called PULSE, displayed full-frame inside a clean desktop browser window on a light grey background. Horizontal 16:9. The UI is a real product screen, not a concept illustration: crisp vector type, consistent 8-pixel grid, off-white canvas, charcoal t…

Generated with MossImage (MOSS-Image-SFT)

Northline GT billboard on a rainy Chicago street with headline and legal copy
MossImage showcaseMOSS-Image-SFT

Northline — rainy billboard

A rain-soaked Chicago billboard with perspective-correct headline type and 6-point legal copy.

1888×1056 · 50 steps · CFG 3

Create a cinematic photograph of a large roadside billboard in Chicago on a rainy November evening, shot from the sidewalk across the street at eye level. Horizontal 16:9. The billboard is mounted on a dark brick building, slightly angled away from camera, with wet asphalt in the foreground reflecting the billboard's l…

Generated with MossImage (MOSS-Image-SFT)

Northbound music festival poster with an orange risograph sunset and 27 artist names
MossImage showcaseMOSS-Image-SFT

Northbound — festival lineup

Twenty-seven fictional artist names in three tiers, from huge headliners to a four-column undercard.

1152×2048 · 50 steps · CFG 3

Create a screen-printed music festival poster for a fictional three-day event called NORTHBOUND, held August 14–16, 2026 at Fort Adams in Newport, Rhode Island. Vertical 9:16, designed as a physical 18 × 24 inch poster with a visible risograph texture: two-color print in fluorescent orange and black on warm off-white p…

Generated with MossImage (MOSS-Image-SFT)

Night Shift comic page with six panels, speech balloons and orange risograph ink
MossImage showcaseMOSS-Image-SFT

Night Shift — comic page

Six panels with hand-lettered captions, speech balloons, SFX and a page number in two-color riso.

1200×1600 · 50 steps · CFG 3

Create a six-panel comic page from a fictional independent graphic novel called NIGHT SHIFT, printed in a two-color risograph edition of black and fluorescent orange on off-white paper. Vertical 3:4 page, clean white margins, thin black panel borders, visible halftone dots in the shadows, slight ink misregistration. St…

Generated with MossImage (MOSS-Image-SFT)

6 prompts

Chinese text & signage

Humorous Chinese roadside notice sign — official MOSS-Image-Base example by the OpenMOSS Team
Verified exampleMOSS-Image-Base

Humorous Chinese roadside notice sign

Chinese heading plus two dense body columns, printed on a bent metal sign.

1184×1776 · 50 steps · CFG 3 · shift 5 · seed 111769238

{"scene_type":"natural roadside environmental photograph with a humorous Chinese notice sign","gen_brief":"A relaxed, photorealistic environmental photograph at pedestrian eye level along a quiet rural road. A battered dark green notice sign stands on the gravel-and-grass shoulder, naturally integrated into the landsca…

Source

OpenMOSS Team, MOSS-Image, Apache-2.0

Eight-panel Chinese weekend comic — official MOSS-Image-Base example by the OpenMOSS Team
Verified exampleMOSS-Image-Base

Eight-panel Chinese weekend comic

Eight sequential panels with hand-lettered Chinese dialogue.

512×4096 · 50 steps · CFG 5 · shift 5 · seed 42

{"scene_type":"Eight-panel weekend comic strip","gen_brief":"Create a finished Chinese slice-of-life comic in a very tall 1:8 format, using eight sequential panels and an expressive, lightly textured digital cartoon style. Follow one exhausted office worker whose weekend fishing escape becomes a comic misunderstanding …

Source

OpenMOSS Team, MOSS-Image, Apache-2.0

Paper-cut pop-up storybook scene — official MOSS-Image-Base example by the OpenMOSS Team
Verified exampleMOSS-Image-Base

Paper-cut pop-up storybook scene

Chinese storybook title and layered paper typography.

1536×1024 · 50 steps · CFG 5 · shift 5 · seed 2570432421

多层折叠式纸雕立体绘本，放在一张书桌上，背景纯净突出主题，比例 3:2 横版。翻开的跨页呈现「哪吒闹海」的场景：脚踏风火轮的少年、翻涌的层叠海浪与龙宫剪影，所有元素皆可精细折叠组合，呈现逼真细腻的纸张折叠质感与多层景深。构图统一采用正面视角，整体视觉梦幻唯美，色彩缤纷绚丽，充满奇幻生动的故事氛围。

Source

OpenMOSS Team, MOSS-Image, Apache-2.0

Chinese literature textbook page — official MOSS-Image-Base example by the OpenMOSS Team
Verified exampleMOSS-Image-Base

Chinese literature textbook page

Printed lesson text, headings and annotations on a photographed page.

1088×1456 · 50 steps · CFG 5 · shift 5 · seed 1946227359

{"scene_type":"Chinese literature textbook photograph","gen_brief":"A realistic editorial photograph of an open Chinese senior-school literature textbook on a warm wooden study desk. The printed lesson presents Wang Xizhi's Lanting preface in authentic simplified Chinese, with carefully typeset classical prose and comp…

Source

OpenMOSS Team, MOSS-Image, Apache-2.0

Text-only promptNo official example image for this one

Official exampleMOSS-Image-Base

Chinese sticky note on a refrigerator

The README Python example. Quote the Chinese text exactly, including line breaks.

1456×1456 · 50 steps · CFG 3 · shift 5 · seed 42

A wide view of a refrigerator beside a kitchen counter with a fruit bowl. A small sticky note on the refrigerator door is handwritten in Chinese: “记得买牛奶和鸡蛋！ 牛奶两盒，鸡蛋一打。 顺便带一袋面包。”

Source

Official MOSS-Image README example

Text-only promptNo official example image for this one

TemplateMOSS-Image-SFT

Storefront sign with Chinese and English

MossImage template. Quote both languages exactly and describe the physical material.

1536×1024 · 50 steps · CFG 3 · shift 5 · seed 42

A straight-on architectural photograph of a small neighborhood bakery at dusk, warm interior light spilling onto the sidewalk, brushed brass frame, cream enamel sign panel, realistic reflections and slight weathering. The sign reads exactly, on two lines: 麦香面包房 MOSS BAKERY The Chinese line is larger, in a bold serif; t…

MossImage prompt template

5 prompts

Posters & UI

Hanfu livestream with calligraphy fan — official MOSS-Image-Base example by the OpenMOSS Team
Verified exampleMOSS-Image-Base

Hanfu livestream with calligraphy fan

Chinese calligraphy on a round fan, wrapped in a live-stream UI overlay.

944×1680 · 50 steps · CFG 5 · shift 5 · seed 42

{"summary":{},"items":[{"bbox":[-824,-472,824,472],"label":"panel","palette":["#D4AF37","#FFFFFF","#B22222","#87CEEB","#2F4F4F"],"caption":"The entire composition is a single frame representing a live stream interface, blending a high-quality photographic portrait of a woman in Hanfu with dynamic social media UI overla…

Source

OpenMOSS Team, MOSS-Image, Apache-2.0

Elden Ring gameplay screenshot with HUD — official MOSS-Image-Base example by the OpenMOSS Team
Verified exampleMOSS-Image-Base

Elden Ring gameplay screenshot with HUD

Game UI labels, item names and a rune counter.

1680×944 · 50 steps · CFG 3 · shift 5 · seed 42

{"summary":{"meta":["screenshot","video_game_capture","dark_fantasy","elden_ring"],"canvas":"single","caption":"A cinematic 16:9 gameplay screenshot from Elden Ring, rendered with the game's richly textured dark-fantasy realism rather than as promotional artwork. A third-person view follows a lone Tarnished standing on…

Source

OpenMOSS Team, MOSS-Image, Apache-2.0

Black-and-white noir detective still — official MOSS-Image-Base example by the OpenMOSS Team
Verified exampleMOSS-Image-Base

Black-and-white noir detective still

Cinematic 4:3 still with directional blind shadows.

1184×880 · 50 steps · CFG 5 · shift 5 · seed 6

生成4:3黑白悬疑电影剧照，室内一位成年侦探坐在桌边。画面左侧可见半幅百叶窗，窗外强硬光透进来，在人物脸部、白衬衫和后墙上形成同一组方向一致的窄条阴影；条纹随表面转折与遮挡发生合理变化，不能像屏幕贴纸一样跨过人物外轮廓直接贴到所有物体上。人物一半面部明亮，另一半较暗，眼睛仍可见。桌上一只黑色有线电话，听筒放在座机上。无烟雾、无霓虹、无文字、无黑边。

Source

OpenMOSS Team, MOSS-Image, Apache-2.0

Text-only promptNo official example image for this one

Official exampleMOSS-Image-Base

EXPLORE travel poster (JSON layout)

The official layout example: a 1024x1024 canvas with centered pixel coordinates.

1024×1024 · 50 steps · CFG 3 · shift 5 · seed 42

{ "summary": { "meta": [ "poster" ], "canvas": "single", "caption": "A minimalist travel poster with a cream background, a blue mountain illustration, and the title EXPLORE at the top." }, "items": [ { "label": "text", "caption": "Large dark-blue uppercase title centered at the top.", "text": "EXPLORE", "bbox": [ -416,…

Source

Official MOSS-Image README and user guide example

Text-only promptNo official example image for this one

TemplateMOSS-Image-SFT

Event poster with date and venue

MossImage template. Put date, venue and ticket line in separate text items so none of them get swallowed.

1024×1536 · 50 steps · CFG 3 · shift 5 · seed 42

{ "summary": { "meta": [ "poster" ], "canvas": "single", "caption": "A risograph-style live music poster in deep ink blue and vermilion on warm paper. Abstract waveform shapes, visible paper grain, generous margins. Bold display type at the top, small mono type at the bottom." }, "items": [ { "label": "text", "caption"…

MossImage prompt template

4 prompts

Long English text

Chinese physics exam paper — official MOSS-Image-Base example by the OpenMOSS Team
Verified exampleMOSS-Image-Base

Chinese physics exam paper

A4 exam layout with numbered questions and equations.

1056×1488 · 50 steps · CFG 5 · shift 5 · seed 42

{"scene_type":"Chinese physics examination paper","gen_brief":"A finished Chinese university entrance examination practice paper, rendered as a crisp, straight-on scan of a single A4 sheet. The design follows formal examination typography, with restrained black ink, precise mathematical notation, generous writing margi…

Source

OpenMOSS Team, MOSS-Image, Apache-2.0

Chalkboard with long English text — official MOSS-Image-Base example by the OpenMOSS Team
Verified exampleMOSS-Image-Base

Chalkboard with long English text

Exact multi-line English copy in natural chalk handwriting.

1360×768 · 50 steps · CFG 3 · shift 6 · seed 2051794908

A candid classroom photograph, slightly off-center, with a worn dark green chalkboard filling most of the frame. Uneven window light, soft shadows, chalk dust, eraser streaks, subtle camera grain. Visible board frame and chalk tray, a blurred desk corner in the foreground. Natural, slightly irregular white chalk handwr…

Source

OpenMOSS Team, MOSS-Image, Apache-2.0

Text-only promptNo official example image for this one

Official exampleMOSS-Image-Base

Chinese whiteboard explainer with MOSI shirt

Official user-guide example with a long Chinese whiteboard explanation.

1024×1024 · 50 steps · CFG 3 · shift 5 · seed 42

一个穿着"MOSI"标志的T恤的中国美女正拿着黑色的马克笔面向镜头微笑。她身后的玻璃板上手写体写着 "一、MOSS-Image的技术路线：5B 的单流 DiT 模型，基于 Muon 训练。二、MOSS-Image的模型特色：1、复杂文字渲染。支持中英渲染、自动布局； 2、使用 Layout 进行精准图像控制，支持文字编辑和物体增减等自由编辑。三、MOSS-Image的后续发展：探索 R2I 参考生图、助力生成式AI发展。"

Source

Official MOSS-Image user guide example

Text-only promptNo official example image for this one

TemplateMOSS-Image-SFT

Packaging label with ingredients block

MossImage template. Small type is the hardest case: keep the block short and use a flat, well-lit surface.

1024×1024 · 50 steps · CFG 3 · shift 5 · seed 42

{ "summary": { "meta": [ "poster" ], "canvas": "single", "caption": "A flat-lay studio photograph of a matte kraft-paper tea pouch on a warm cream surface, soft top light, subtle paper fibers, no glare. A printed label occupies the upper two thirds; a small ingredients block sits below it." }, "items": [ { "label": "te…

MossImage prompt template

7 prompts

Ads & product

Nine-frame selfie contact sheet — official MOSS-Image-Base example by the OpenMOSS Team
Verified exampleMOSS-Image-Base

Nine-frame selfie contact sheet

A 3x3 contact sheet with varied framing and expressions.

768×1360 · 30 steps · CFG 4 · shift 5 · seed 42

{"scene_type":"Amateur selfie contact sheet","gen_brief":"Create a high-resolution photographic contact sheet of nine fictional amateur selfie mishaps featuring Donald Trump, presented as a coherent three-by-three grid on a portrait canvas. Recognizable facial features, swept blond hair, realistic skin texture, and ord…

Source

OpenMOSS Team, MOSS-Image, Apache-2.0

Chinese YouTube thumbnail celebration — official MOSS-Image-Base example by the OpenMOSS Team
Verified exampleMOSS-Image-Base

Chinese YouTube thumbnail celebration

Thumbnail-style Chinese headline and stadium crowd energy.

1536×1024 · 50 steps · CFG 5 · shift 5 · seed 42

一张极具 YouTube 缩略图风格的夸张宣传画面：场景设置在一座大型露天体育场或竞技场内，背景是明亮的蓝天、环形看台与巨型电子屏幕，大量年轻观众挤满画面，兴奋地欢呼、张嘴大笑，并高举写满数学公式、函数图像和几何草图的白纸，整体营造出“完成超大规模数学挑战后全场庆祝”的戏剧化氛围。 画面正中央是一位拟人化的 ChatGPT 形象，以低机位、中近景英雄式构图占据视觉中心。人物穿着宽松的黑色连帽卫衣，胸前印有青绿色的 OpenAI/ChatGPT 旋结标志，双臂交叉抱在胸前，姿态自信而具有挑战意味。人物的头部被替换成一个大型、圆角、具有明显 3D 体积感的青绿色方形 ChatGPT 图标，正面中央是白色 OpenAI 旋结 Logo；图…

Source

OpenMOSS Team, MOSS-Image, Apache-2.0

Gym trainers coaching a squat — official MOSS-Image-Base example by the OpenMOSS Team
Verified exampleMOSS-Image-Base

Gym trainers coaching a squat

Fitness brand photography with readable shirt and poster text.

1184×1776 · 50 steps · CFG 3 · shift 5 · seed 42

{"summary":{"meta":["photograph"],"canvas":"single","caption":"A high-quality indoor photograph captures a male and female fitness trainer guiding a male client through a weighted squat exercise in a modern gym. The scene is brightly lit with overhead industrial lights, emphasizing the texture of the gym equipment and …

Source

OpenMOSS Team, MOSS-Image, Apache-2.0

Rainy convenience store night scene — official MOSS-Image-Base example by the OpenMOSS Team
Verified exampleMOSS-Image-Base

Rainy convenience store night scene

Cinematic teal-and-amber storefront with no readable lettering.

1680×944 · 50 steps · CFG 3 · shift 5 · seed 42

A cinematic, photorealistic nighttime scene of a small neighborhood convenience store on a quiet, rain-soaked street. The warmly illuminated storefront occupies the right half of the frame, while the street recedes into soft mist on the left. A lone young adult wearing a charcoal coat stands just outside the entrance b…

Source

OpenMOSS Team, MOSS-Image, Apache-2.0

Text-only promptNo official example image for this one

TemplateMOSS-Image-SFT

Cafe menu board (legibility-first)

MossImage template. Name the dish, price and section order explicitly; keep one text block per column.

1024×1536 · 50 steps · CFG 3 · shift 5 · seed 42

A straight-on photograph of a cream-painted cafe menu board hanging on a warm plaster wall, soft morning window light from the left, subtle wood frame, shallow paper texture, no glare. The board reads exactly: MORNING MENU ESPRESSO — 3.50 CORTADO — 4.00 FLAT WHITE — 4.50 SEASONAL PLUM SODA — 5.00 Yuzu TONIC — 5.00 BAKE…

MossImage prompt template

Text-only promptNo official example image for this one

TemplateMOSS-Image-SFT

Product ad with headline and price

MossImage template. The layout JSON pins the headline, subhead and price to separate boxes.

1024×1024 · 50 steps · CFG 3 · shift 5 · seed 42

{ "summary": { "meta": [ "poster" ], "canvas": "single", "caption": "A clean studio product advertisement: a matte ceramic coffee cup on a warm cream background, soft shadow, one sprig of dried grass. Bold sans-serif headline, smaller subhead, price in a vermilion circle. Editorial, minimal, high-end." }, "items": [ { …

MossImage prompt template

Text-only promptNo official example image for this one

TemplateMOSS-Image-SFT

Square social ad with offer text

MossImage template. One short offer, one supporting line, one call to action.

1024×1024 · 50 steps · CFG 3 · shift 5 · seed 42

{ "summary": { "meta": [ "poster" ], "canvas": "single", "caption": "A bright editorial social ad for a skincare serum: a frosted glass bottle on a wet stone plinth, warm sunlight, sharp shadow, cream and vermilion palette, generous negative space in the upper half." }, "items": [ { "label": "text", "caption": "Short b…

MossImage prompt template

8 prompts

Portraits & art

Anime campus ensemble comedy — official MOSS-Image-Base example by the OpenMOSS Team
Verified exampleMOSS-Image-Base

Anime campus ensemble comedy

Ten named characters in one chaotic group scene.

1776×1184 · 50 steps · CFG 5 · shift 5 · seed 3072627948

{"scene_type":"chaotic anime ensemble action comedy","gen_brief":"A lively Blue Archive campus group photo collapsing into a harmless slapstick disaster, featuring exactly ten distinct named students from the elements, each appearing once. An asymmetric arch-shaped ensemble composition in one continuous sunlit courtyar…

Source

OpenMOSS Team, MOSS-Image, Apache-2.0

Anime girl at a Tokyo crosswalk — official MOSS-Image-Base example by the OpenMOSS Team
Verified exampleMOSS-Image-Base

Anime girl at a Tokyo crosswalk

Photoreal anime-style portrait with a crowded street backdrop.

1184×1776 · 50 steps · CFG 3 · shift 5 · seed 3537712559

A young anime woman, Kazusa from Blue Archive, stands still at the edge of a crowded Tokyo crosswalk outside a busy train station on a damp, overcast dusk. Her short dark hair with soft magenta inner highlights frames her face, cat ears perked slightly beneath her distinctive glowing pink halo. She wears an oversized c…

Source

OpenMOSS Team, MOSS-Image, Apache-2.0

Amiya cosplay character portrait — official MOSS-Image-Base example by the OpenMOSS Team
Verified exampleMOSS-Image-Base

Amiya cosplay character portrait

Short tag prompt with a costume logo and prop sword.

1024×1536 · 50 steps · CFG 5 · shift 5 · seed 0

general, 1girl, best quality, most attractive face, cosplay, amiya (arknights), with "Rhodes Island" on her clothes, carring a sword

Source

OpenMOSS Team, MOSS-Image, Apache-2.0

White marble Saber sculpture — official MOSS-Image-Base example by the OpenMOSS Team
Verified exampleMOSS-Image-Base

White marble Saber sculpture

Museum-style marble portrait with dramatic lighting.

768×1360 · 50 steps · CFG 3 · shift 5 · seed 42

An original, museum-quality white marble sculpture of Artoria Pendragon (Saber) from Fate, conceived with the anatomical discipline, monumental presence, and exquisite craftsmanship of Renaissance sculpture, without copying Michelangelo's David, its pose, or its composition. Preserve Saber's unmistakable character desi…

Source

OpenMOSS Team, MOSS-Image, Apache-2.0

Hanako cosplay hotel room photo — official MOSS-Image-Base example by the OpenMOSS Team
Verified exampleMOSS-Image-Base

Hanako cosplay hotel room photo

Warm lamp-lit cosplay portrait with detailed costume text.

880×1184 · 50 steps · CFG 3 · shift 5 · seed 42

A hotel room lit only by a warm bedside lamp. The white sheets on the double bed are rumpled, pillows stacked against the headboard. A cosplayer portraying Hanako from Blue Archive lies on her side along the edge of the bed, long pink hair spilling over one shoulder, a loose braid on the other, a white flower clip stil…

Source

OpenMOSS Team, MOSS-Image, Apache-2.0

Five-character summer beach panorama — official MOSS-Image-Base example by the OpenMOSS Team
Verified exampleMOSS-Image-Base

Five-character summer beach panorama

Ultra-wide 4:1 anime ensemble banner.

2880×720 · 50 steps · CFG 3 · shift 5 · seed 88

{"scene_type":"Anime-style summer beach ensemble with five characters","gen_brief":"Ultra-wide panoramic beach vacation illustration, aspect ratio 4:1. Recompose for a wide banner rather than stretching or vertically cropping a conventional image. Feature Shiroko, Hoshino, Yuuka, Hina, and Yuzu from Blue Archive, prese…

Source

OpenMOSS Team, MOSS-Image, Apache-2.0

JK uniform portrait — official MOSS-Image-Base example by the OpenMOSS Team
Verified exampleMOSS-Image-Base

JK uniform portrait

Short tag prompt with a quality-first ordering.

880×1184 · 50 steps · CFG 5 · shift 5 · seed 42

masterpiece, best quality, highres, most attractive face 生成一个穿JK短裙的少女，长得非常非常好看的那种类型的。

Source

OpenMOSS Team, MOSS-Image, Apache-2.0

Text-only promptNo official example image for this one

Official exampleMOSS-Image-Base

Tag-order character prompt

The official tag-order example: quality → score → meta → year → character → series → artist → general.

1024×1536 · 50 steps · CFG 5 · shift 5 · seed 42

best quality, highres, hatsune miku, vocaloid, @foo bar, 1girl, blue hair

Source

Official MOSS-Image user guide example

Style layer

Style prompts that do not break the text

Put style words in the scene caption, never inside a text item. Keep the text item concerned only with the exact words, position and type treatment.

Editorial print

risograph poster, warm paper stock, limited ink palette, visible paper grain, generous margins

Works for event posters and menu covers. Avoid “distressed” or “torn” if small type must stay readable.

Photoreal product

studio product photograph, softbox key light, cream seamless backdrop, subtle contact shadow, 85mm lens

Keep the product and the copy on separate planes. Flat lighting preserves label text.

Cinematic still

35mm lens aesthetic, teal-and-amber grade, soft highlight bloom, subtle film grain, shallow depth of field

Reduce bloom and grain when the image contains a sign or a screen with small text.

Anime key visual

anime key visual, clean linework, cel shading, dramatic rim light, detailed background

Ask for clean linework and flat color fills; painterly strokes are the usual cause of broken glyphs.

A safe prompt skeletonmono
[shot and setting]
[subject and action]
[lighting and material]
[exact text, in quotes, with line breaks]
[placement: top / center / lower third; type style]
[negative: no extra lettering, no watermark, no glare]

Next

Use a prompt now

Copy a prompt, then open the generator. The Try button saves the prompt in your browser session and takes you straight to the generator.

Open the MossImage generatorPaste the prompt, pick Draft or Final, and render. Two free credits per day.ComfyUI workflowRun the same model locally with the official graph and sampler settings.Layout control explainedCentered pixel coordinates, text items and the JSON format used by the templates.Browse the showcaseTwelve official examples with notes on dense text, hierarchy and layout.Official examplesBrowse the source images and parameters.json files in the OpenMOSS repository.

FAQ

Prompt library questions

Attribution, parameters and how to reuse the prompts.

Do these prompts work in the MossImage online generator?

Yes. Paste a prompt into the positive prompt field, or switch to the layout editor and paste a JSON layout. The eight showcase cards were generated with MossImage (MOSS-Image-SFT); the verified cards use MOSS-Image-Base, so the same prompt can produce a different but related image.

Are the images in the library generated by MossImage?

The eight MossImage showcase cards at the top were generated with MossImage (MOSS-Image-SFT) from the prompts shown on each card. The twenty verified cards use the official MOSS-Image-Base example images published by the OpenMOSS Team under Apache-2.0. The four text-only official examples and six templates have no example image.

Why do the official parameters vary between cards?

The official examples were generated over time with different choices. CFG ranges from 3 to 5, steps from 30 to 50, shift from 5 to 6, and canvas sizes differ. The published workflow defaults are 1024×1024, 50 steps, CFG 3, Euler / sgm_uniform, shift 5 and a layer-22 skip; use those when you want a consistent baseline.

How do I keep Chinese or English text from being garbled?

Quote the exact copy, keep every line break, and give each block its own text item instead of one long paragraph. Avoid perspective, motion blur, heavy glow and handwritten styles unless the reference itself is handwritten. Keep the whole prompt under the text encoder limit of 4096 tokens.

Can I use the copied prompts commercially?

The prompt text and the official example images are published under Apache-2.0. The images may still depict characters, brands or people with their own rights. Check the Terms of Service and clear any third-party rights before commercial use.

What is the difference between a showcase card, a verified example and a template?

A showcase card is a finished render generated with MossImage (MOSS-Image-SFT) from the prompt shown on the card. A verified example has an official image and a published parameters.json from the OpenMOSS repository. A template is a starting point written by MossImage that follows the same layout format but has no sample image.