One layout on two models: a masthead, two cover lines, an issue label and the cover model, each in its own box. Every word is the one typed into its box.
Most image prompts describe a scene and leave the arrangement to the model. A layout prompt goes further. It lists each element with a box that says where it goes, and gives every piece of visible text its exact words. MOSS-Image and FLUX 3 Image Klein both read prompts like this, each in its own format.
Layout Composer is the editor for those prompts in Sogni. Draw and drag boxes on a canvas, type the words and descriptions, and press Imagine. Sogni writes the layout in the format the chosen model reads, and you can switch between MOSS-Image SFT, MOSS-Image 8-Step and FLUX 3 Image Klein without redrawing anything.
- Text where you put itTitles, taglines and labels get their own boxes, with your exact words and line breaks.
- Objects and charactersGive each one a box and describe how it looks.
- Three models, one layoutSwitch models and your layout comes along.
- TemplatesStart from a travel poster, a product campaign or a three-panel story.
- Compose with AIDescribe a scene and review a proposed layout before you use it.
- Save and reopenExport layout files, edit the raw JSON, or reopen any result with Adjust layout.
How it works
- Choose MOSS-Image SFT, MOSS-Image 8-Step or FLUX 3 Image Klein, then Compose by layout in the prompt box.
- Describe the whole picture in Scene: subject, style, light and background.
- Add an Object, Text or Character. Drag its box into place and drag a corner to resize it, or type exact pixel bounds.
- Type the exact words in Visible text (text boxes only), and how each element should look in Appearance.
- Press Imagine. Each result shows the boxes it was made from, and Adjust layout reopens them.
The ? beside the editor’s title explains each field, and switching models from the editor keeps everything you have placed.
One layout, three models
Each set below is a single layout, rendered twice on each model without moving a box. Switching models in Layout Composer keeps the layout and rewrites it in the new model’s format. Times are for both images, from pressing Imagine to the finished files, on Sogni’s Fast network.
Film poster
A title, a tagline and a release line in their own boxes, with the lighthouse and its keeper placed where the story needs them.
THE LAST / LIGHTHOUSETall white-and-red lighthouseAdult lighthouse keeperSome lights never go out.IN THEATERS DECEMBER
Two images per model, start to finish on Sogni: MOSS-Image SFT 29.7 s · MOSS-Image 8-Step 9.7 s · FLUX 3 Image Klein 10.3 s.
The prompts each model received
{
"summary": {
"meta": [
"movie poster"
],
"canvas": "single",
"caption": "A cinematic teal-and-amber movie poster at dusk: a lone lighthouse on a black cliff above a stormy sea, its beam cutting through slanting rain, an adult lighthouse keeper on the rocks below. Dramatic film-poster lighting, rich detail."
},
"items": [
{
"label": "text",
"caption": "Huge weathered silver serif title in two lines, centered.",
"text": "THE LAST\nLIGHTHOUSE",
"bbox": [
-560,
-350,
-352,
350
]
},
{
"label": "object",
"caption": "A tall white-and-red lighthouse on a dark cliff, its bright beam sweeping across the rain.",
"bbox": [
-341,
16,
244,
300
]
},
{
"label": "character",
"caption": "An adult lighthouse keeper in a yellow raincoat, seen from behind, standing on wet rocks and facing the sea.",
"bbox": [
24,
-300,
438,
-66
]
},
{
"label": "text",
"caption": "Small italic off-white tagline on one line.",
"text": "Some lights never go out.",
"bbox": [
456,
-283,
499,
283
]
},
{
"label": "text",
"caption": "Thin, widely spaced silver capitals on one line.",
"text": "IN THEATERS DECEMBER",
"bbox": [
516,
-250,
560,
250
]
}
],
"bbox_coord_space": "abs_centralized",
"bbox_format": "yxyx"
}A cinematic teal-and-amber movie poster at dusk: a lone lighthouse on a black cliff above a stormy sea, its beam cutting through slanting rain, an adult lighthouse keeper on the rocks below. Dramatic film-poster lighting, rich detail. Huge weathered silver serif title in two lines, centered. The visible text is "THE LAST\nLIGHTHOUSE". <Unknown_Text_1> A tall white-and-red lighthouse on a dark cliff, its bright beam sweeping across the rain. <object_2> An adult lighthouse keeper in a yellow raincoat, seen from behind, standing on wet rocks and facing the sea. <character_3> Small italic off-white tagline on one line. The visible text is "Some lights never go out.". <Unknown_Text_4> Thin, widely spaced silver capitals on one line. The visible text is "IN THEATERS DECEMBER". <Unknown_Text_5> [{"id":"Unknown_Text_1","bbox":[40,80,210,920],"desc":"Huge weathered silver serif title in two lines, centered. The visible text is \"THE LAST\\nLIGHTHOUSE\"."},{"id":"object_2","bbox":[220,520,700,860],"desc":"A tall white-and-red lighthouse on a dark cliff, its bright beam sweeping across the rain."},{"id":"character_3","bbox":[520,140,860,420],"desc":"An adult lighthouse keeper in a yellow raincoat, seen from behind, standing on wet rocks and facing the sea."},{"id":"Unknown_Text_4","bbox":[875,160,910,840],"desc":"Small italic off-white tagline on one line. The visible text is \"Some lights never go out.\"."},{"id":"Unknown_Text_5","bbox":[925,200,960,800],"desc":"Thin, widely spaced silver capitals on one line. The visible text is \"IN THEATERS DECEMBER\"."}]Magazine cover
A masthead, two cover lines and an issue label around the cover photo, with the model in her own box.
WANDERAdult womanSANTORINI / AFTER DARK48 hours in LisbonThe slow travel issue
Two images per model, start to finish on Sogni: MOSS-Image SFT 34.9 s · MOSS-Image 8-Step 8.5 s · FLUX 3 Image Klein 8.0 s.
The prompts each model received
{
"summary": {
"meta": [
"magazine cover"
],
"canvas": "single",
"caption": "A glossy travel magazine cover photographed in golden-hour light on a Mediterranean terrace with whitewashed walls, pink bougainvillea and the blue sea behind. Crisp editorial photography, natural skin texture."
},
"items": [
{
"label": "text",
"caption": "Huge bold white serif masthead across the top.",
"text": "WANDER",
"bbox": [
-572,
-367,
-413,
367
]
},
{
"label": "character",
"caption": "An adult woman in a white linen shirt and a wide straw hat, smiling, three-quarter view, from the waist up.",
"bbox": [
-377,
-134,
596,
375
]
},
{
"label": "text",
"caption": "Bold white sans-serif cover line in two lines, left aligned.",
"text": "SANTORINI\nAFTER DARK",
"bbox": [
-244,
-367,
-109,
-66
]
},
{
"label": "text",
"caption": "Small white sans-serif cover line, left aligned.",
"text": "48 hours in Lisbon",
"bbox": [
-61,
-367,
0,
-99
]
},
{
"label": "text",
"caption": "Small bright yellow sans-serif label, left aligned.",
"text": "The slow travel issue",
"bbox": [
474,
-367,
523,
-49
]
}
],
"bbox_coord_space": "abs_centralized",
"bbox_format": "yxyx"
}A glossy travel magazine cover photographed in golden-hour light on a Mediterranean terrace with whitewashed walls, pink bougainvillea and the blue sea behind. Crisp editorial photography, natural skin texture. Huge bold white serif masthead across the top. The visible text is "WANDER". <Unknown_Text_1> An adult woman in a white linen shirt and a wide straw hat, smiling, three-quarter view, from the waist up. <character_2> Bold white sans-serif cover line in two lines, left aligned. The visible text is "SANTORINI\nAFTER DARK". <Unknown_Text_3> Small white sans-serif cover line, left aligned. The visible text is "48 hours in Lisbon". <Unknown_Text_4> Small bright yellow sans-serif label, left aligned. The visible text is "The slow travel issue". <Unknown_Text_5> [{"id":"Unknown_Text_1","bbox":[30,60,160,940],"desc":"Huge bold white serif masthead across the top. The visible text is \"WANDER\"."},{"id":"character_2","bbox":[190,340,990,950],"desc":"An adult woman in a white linen shirt and a wide straw hat, smiling, three-quarter view, from the waist up."},{"id":"Unknown_Text_3","bbox":[300,60,410,420],"desc":"Bold white sans-serif cover line in two lines, left aligned. The visible text is \"SANTORINI\\nAFTER DARK\"."},{"id":"Unknown_Text_4","bbox":[450,60,500,380],"desc":"Small white sans-serif cover line, left aligned. The visible text is \"48 hours in Lisbon\"."},{"id":"Unknown_Text_5","bbox":[890,60,930,440],"desc":"Small bright yellow sans-serif label, left aligned. The visible text is \"The slow travel issue\"."}]Recipe spread
A dish, its title, a timing line and a five-line ingredient list, laid out like a magazine page.
StackLEMON RICOTTA / PANCAKESSERVES 4 · 25 MIN1 cup ricotta / 2 eggs / 1 lemon, zested / 1 cup flour / Honey to serveTwo whole lemons
Two images per model, start to finish on Sogni: MOSS-Image SFT 30.4 s · MOSS-Image 8-Step 8.9 s · FLUX 3 Image Klein 7.4 s.
The prompts each model received
{
"summary": {
"meta": [
"recipe card"
],
"canvas": "single",
"caption": "An overhead food-magazine recipe spread on a pale linen tablecloth in soft daylight. Fresh, bright and appetizing, with dark-green and terracotta typography."
},
"items": [
{
"label": "object",
"caption": "A stack of fluffy lemon ricotta pancakes on a white plate, topped with blueberries, lemon zest and a drizzle of honey, seen from above.",
"bbox": [
-317,
-560,
350,
-24
]
},
{
"label": "text",
"caption": "Large dark-green serif title in two lines, left aligned.",
"text": "LEMON RICOTTA\nPANCAKES",
"bbox": [
-350,
48,
-166,
560
]
},
{
"label": "text",
"caption": "Small widely spaced terracotta capitals on one line.",
"text": "SERVES 4 · 25 MIN",
"bbox": [
-142,
48,
-99,
487
]
},
{
"label": "text",
"caption": "A neat ingredient list in dark-green sans-serif, one item per line.",
"text": "1 cup ricotta\n2 eggs\n1 lemon, zested\n1 cup flour\nHoney to serve",
"bbox": [
-50,
48,
283,
365
]
},
{
"label": "object",
"caption": "Two whole lemons and a halved lemon with a sprig of mint.",
"bbox": [
99,
376,
367,
584
]
}
],
"bbox_coord_space": "abs_centralized",
"bbox_format": "yxyx"
}An overhead food-magazine recipe spread on a pale linen tablecloth in soft daylight. Fresh, bright and appetizing, with dark-green and terracotta typography. A stack of fluffy lemon ricotta pancakes on a white plate, topped with blueberries, lemon zest and a drizzle of honey, seen from above. <object_1> Large dark-green serif title in two lines, left aligned. The visible text is "LEMON RICOTTA\nPANCAKES". <Unknown_Text_2> Small widely spaced terracotta capitals on one line. The visible text is "SERVES 4 · 25 MIN". <Unknown_Text_3> A neat ingredient list in dark-green sans-serif, one item per line. The visible text is "1 cup ricotta\n2 eggs\n1 lemon, zested\n1 cup flour\nHoney to serve". <Unknown_Text_4> Two whole lemons and a halved lemon with a sprig of mint. <object_5> [{"id":"object_1","bbox":[120,40,920,480],"desc":"A stack of fluffy lemon ricotta pancakes on a white plate, topped with blueberries, lemon zest and a drizzle of honey, seen from above."},{"id":"Unknown_Text_2","bbox":[80,540,300,960],"desc":"Large dark-green serif title in two lines, left aligned. The visible text is \"LEMON RICOTTA\\nPANCAKES\"."},{"id":"Unknown_Text_3","bbox":[330,540,380,900],"desc":"Small widely spaced terracotta capitals on one line. The visible text is \"SERVES 4 · 25 MIN\"."},{"id":"Unknown_Text_4","bbox":[440,540,840,800],"desc":"A neat ingredient list in dark-green sans-serif, one item per line. The visible text is \"1 cup ricotta\\n2 eggs\\n1 lemon, zested\\n1 cup flour\\nHoney to serve\"."},{"id":"object_5","bbox":[620,810,940,980],"desc":"Two whole lemons and a halved lemon with a sprig of mint."}]Four-panel comic
Four panels and four speech bubbles, read in order, with the same astronaut and robot throughout.
Top-left panelRAMEN NIGHT!Top-right panel3 MINUTES...Bottom-left panelUH-OH.Bottom-right panelBEST NIGHT EVER.
Two images per model, start to finish on Sogni: MOSS-Image SFT 31.5 s · MOSS-Image 8-Step 9.9 s · FLUX 3 Image Klein 7.6 s.
The prompts each model received
{
"summary": {
"meta": [
"comic"
],
"canvas": "multi_panel",
"caption": "A four-panel anime comic page in clean line art with soft cel shading, read left to right and top to bottom, with white gutters. An adult astronaut woman with short silver hair and a small round white helper robot share a ramen night aboard a space station."
},
"items": [
{
"label": "character",
"caption": "Top-left panel: the adult astronaut with short silver hair floats into the station galley holding two cup-noodle bowls.",
"bbox": [
-492,
-492,
-10,
-10
]
},
{
"label": "text",
"caption": "White comic speech bubble with bold black lettering.",
"text": "RAMEN NIGHT!",
"bbox": [
-472,
-461,
-378,
-204
]
},
{
"label": "character",
"caption": "Top-right panel: the small round white robot pours hot water into the bowls as steam drifts in zero gravity.",
"bbox": [
-492,
10,
-10,
492
]
},
{
"label": "text",
"caption": "White comic speech bubble with bold black lettering.",
"text": "3 MINUTES...",
"bbox": [
-472,
51,
-378,
308
]
},
{
"label": "character",
"caption": "Bottom-left panel: noodles float out of the bowls in zero gravity while the astronaut and the robot look surprised.",
"bbox": [
10,
-492,
492,
-10
]
},
{
"label": "text",
"caption": "White comic speech bubble with bold black lettering.",
"text": "UH-OH.",
"bbox": [
30,
-461,
123,
-266
]
},
{
"label": "character",
"caption": "Bottom-right panel: both laugh and catch the floating noodles with chopsticks, the glowing Earth in the window behind them.",
"bbox": [
10,
10,
492,
492
]
},
{
"label": "text",
"caption": "White comic speech bubble with bold black lettering.",
"text": "BEST NIGHT EVER.",
"bbox": [
30,
51,
123,
390
]
}
],
"bbox_coord_space": "abs_centralized",
"bbox_format": "yxyx"
}A four-panel anime comic page in clean line art with soft cel shading, read left to right and top to bottom, with white gutters. An adult astronaut woman with short silver hair and a small round white helper robot share a ramen night aboard a space station. Top-left panel: the adult astronaut with short silver hair floats into the station galley holding two cup-noodle bowls. <character_1> White comic speech bubble with bold black lettering. The visible text is "RAMEN NIGHT!". <Unknown_Text_2> Top-right panel: the small round white robot pours hot water into the bowls as steam drifts in zero gravity. <character_3> White comic speech bubble with bold black lettering. The visible text is "3 MINUTES...". <Unknown_Text_4> Bottom-left panel: noodles float out of the bowls in zero gravity while the astronaut and the robot look surprised. <character_5> White comic speech bubble with bold black lettering. The visible text is "UH-OH.". <Unknown_Text_6> Bottom-right panel: both laugh and catch the floating noodles with chopsticks, the glowing Earth in the window behind them. <character_7> White comic speech bubble with bold black lettering. The visible text is "BEST NIGHT EVER.". <Unknown_Text_8> [{"id":"character_1","bbox":[20,20,490,490],"desc":"Top-left panel: the adult astronaut with short silver hair floats into the station galley holding two cup-noodle bowls."},{"id":"Unknown_Text_2","bbox":[40,50,130,300],"desc":"White comic speech bubble with bold black lettering. The visible text is \"RAMEN NIGHT!\"."},{"id":"character_3","bbox":[20,510,490,980],"desc":"Top-right panel: the small round white robot pours hot water into the bowls as steam drifts in zero gravity."},{"id":"Unknown_Text_4","bbox":[40,550,130,800],"desc":"White comic speech bubble with bold black lettering. The visible text is \"3 MINUTES...\"."},{"id":"character_5","bbox":[510,20,980,490],"desc":"Bottom-left panel: noodles float out of the bowls in zero gravity while the astronaut and the robot look surprised."},{"id":"Unknown_Text_6","bbox":[530,50,620,240],"desc":"White comic speech bubble with bold black lettering. The visible text is \"UH-OH.\"."},{"id":"character_7","bbox":[510,510,980,980],"desc":"Bottom-right panel: both laugh and catch the floating noodles with chopsticks, the glowing Earth in the window behind them."},{"id":"Unknown_Text_8","bbox":[530,550,620,880],"desc":"White comic speech bubble with bold black lettering. The visible text is \"BEST NIGHT EVER.\"."}]Fantasy book cover
A title, a dragon, a knight and the author line.
EMBER / CROWNHuge crimson dragonAdult knightA NOVEL BY MIRA VALE
Two images per model, start to finish on Sogni: MOSS-Image SFT 32.5 s · MOSS-Image 8-Step 6.9 s · FLUX 3 Image Klein 13.1 s.
The prompts each model received
{
"summary": {
"meta": [
"book cover"
],
"canvas": "single",
"caption": "An epic painted fantasy novel cover: a crimson dragon coiled around a ruined mountain tower under a stormy sky full of embers, with an adult armored knight on the cliff below. Rich oil-painting detail and dramatic light."
},
"items": [
{
"label": "text",
"caption": "Huge engraved gold fantasy title in two lines with a soft glow.",
"text": "EMBER\nCROWN",
"bbox": [
-560,
-350,
-328,
350
]
},
{
"label": "object",
"caption": "A huge crimson dragon with spread wings coiled around a ruined stone tower, breathing embers.",
"bbox": [
-317,
-267,
195,
375
]
},
{
"label": "character",
"caption": "An adult knight in dark steel armor and a red cloak, seen from behind, raising a glowing sword.",
"bbox": [
121,
-358,
499,
-83
]
},
{
"label": "text",
"caption": "Small widely spaced gold capitals on one line.",
"text": "A NOVEL BY MIRA VALE",
"bbox": [
522,
-250,
566,
250
]
}
],
"bbox_coord_space": "abs_centralized",
"bbox_format": "yxyx"
}An epic painted fantasy novel cover: a crimson dragon coiled around a ruined mountain tower under a stormy sky full of embers, with an adult armored knight on the cliff below. Rich oil-painting detail and dramatic light. Huge engraved gold fantasy title in two lines with a soft glow. The visible text is "EMBER\nCROWN". <Unknown_Text_1> A huge crimson dragon with spread wings coiled around a ruined stone tower, breathing embers. <object_2> An adult knight in dark steel armor and a red cloak, seen from behind, raising a glowing sword. <character_3> Small widely spaced gold capitals on one line. The visible text is "A NOVEL BY MIRA VALE". <Unknown_Text_4> [{"id":"Unknown_Text_1","bbox":[40,80,230,920],"desc":"Huge engraved gold fantasy title in two lines with a soft glow. The visible text is \"EMBER\\nCROWN\"."},{"id":"object_2","bbox":[240,180,660,950],"desc":"A huge crimson dragon with spread wings coiled around a ruined stone tower, breathing embers."},{"id":"character_3","bbox":[600,70,910,400],"desc":"An adult knight in dark steel armor and a red cloak, seen from behind, raising a glowing sword."},{"id":"Unknown_Text_4","bbox":[930,200,965,800],"desc":"Small widely spaced gold capitals on one line. The visible text is \"A NOVEL BY MIRA VALE\"."}]Festival banner
A 1536 × 640 banner with a title, a date line and a DJ.
Neon TidesAUG 14–16 · LONG BEACHAdult DJ
Two images per model, start to finish on Sogni: MOSS-Image SFT 35.0 s · MOSS-Image 8-Step 9.0 s · FLUX 3 Image Klein 6.1 s.
The prompts each model received
{
"summary": {
"meta": [
"banner"
],
"canvas": "single",
"caption": "A vibrant neon synthwave festival banner at night: a glowing magenta grid floor, palm silhouettes and a huge striped setting sun, washed in magenta and cyan light."
},
"items": [
{
"label": "text",
"caption": "Huge chrome-and-magenta retro script title with a neon glow.",
"text": "Neon Tides",
"bbox": [
-231,
-707,
39,
62
]
},
{
"label": "text",
"caption": "Bright cyan widely spaced capitals on one line.",
"text": "AUG 14–16 · LONG BEACH",
"bbox": [
76,
-707,
141,
62
]
},
{
"label": "character",
"caption": "An adult DJ with headphones and a glowing jacket behind turntables, lit by neon.",
"bbox": [
-256,
153,
295,
707
]
}
],
"bbox_coord_space": "abs_centralized",
"bbox_format": "yxyx"
}A vibrant neon synthwave festival banner at night: a glowing magenta grid floor, palm silhouettes and a huge striped setting sun, washed in magenta and cyan light. Huge chrome-and-magenta retro script title with a neon glow. The visible text is "Neon Tides". <Unknown_Text_1> Bright cyan widely spaced capitals on one line. The visible text is "AUG 14–16 · LONG BEACH". <Unknown_Text_2> An adult DJ with headphones and a glowing jacket behind turntables, lit by neon. <character_3> [{"id":"Unknown_Text_1","bbox":[140,40,560,540],"desc":"Huge chrome-and-magenta retro script title with a neon glow. The visible text is \"Neon Tides\"."},{"id":"Unknown_Text_2","bbox":[620,40,720,540],"desc":"Bright cyan widely spaced capitals on one line. The visible text is \"AUG 14–16 · LONG BEACH\"."},{"id":"character_3","bbox":[100,600,960,960],"desc":"An adult DJ with headphones and a glowing jacket behind turntables, lit by neon."}]Start from a template
Three templates cover common starting points (a travel poster, a product campaign and a three-panel story), each on a canvas shaped for it. Every element stays editable.
- Small ivory ceramic cup
- SAVOR / THE / MOMENT
- A little warmth, every day.
MOSS-Image SFT · 1216 × 832 · seed 854832372
Layout and prompt
Download this layout and open it with Import in Layout Composer.
{
"summary": {
"meta": [
"product campaign"
],
"canvas": "single",
"caption": "A studio campaign with a small ivory ceramic cup on the left, shown in full including its handle with space around it. A seamless warm background, soft side lighting, tactile speckled glaze and dark-green typography on the right."
},
"items": [
{
"label": "object",
"caption": "One small ivory ceramic cup with a rounded handle and speckled glaze, three-quarter view. The entire cup and handle fit within this region, with space around the silhouette and a soft contact shadow.",
"bbox": [
-184,
-487,
233,
-24
]
},
{
"label": "text",
"caption": "Dark-green serif title in three lines, generous spacing.",
"text": "SAVOR\nTHE\nMOMENT",
"bbox": [
-250,
85,
0,
536
]
},
{
"label": "text",
"caption": "Smaller dark-green supporting copy.",
"text": "A little warmth, every day.",
"bbox": [
99,
109,
208,
523
]
}
],
"bbox_coord_space": "abs_centralized",
"bbox_format": "yxyx"
}
- Left panel
- Middle panel
- Right panel
MOSS-Image SFT · 1536 × 640 · seed 2653686121
Layout and prompt
Download this layout and open it with Import in Layout Composer.
{
"summary": {
"meta": [
"comic"
],
"canvas": "multi_panel",
"caption": "A wordless three-panel watercolor comic, read from left to right with equal gutters. An adult gardener plants a seed, waters a sprout, then admires a sunflower. Consistent green overalls, straw hat and terracotta pot throughout."
},
"items": [
{
"label": "character",
"caption": "Left panel: an adult gardener in green overalls and a straw hat kneels beside a terracotta pot to plant a seed.",
"bbox": [
-269,
-730,
269,
-291
]
},
{
"label": "character",
"caption": "Middle panel: the same gardener waters a small sprout in the terracotta pot.",
"bbox": [
-269,
-220,
269,
220
]
},
{
"label": "character",
"caption": "Right panel: the same gardener stands beside a flowering sunflower in the terracotta pot.",
"bbox": [
-269,
291,
269,
730
]
}
],
"bbox_coord_space": "abs_centralized",
"bbox_format": "yxyx"
}Let AI draft the layout
Type a scene and press Compose with AI. Sogni’s prompt assistant turns it into elements and boxes for your canvas, keeping any words you quoted, and shows the proposal before anything changes: choose Use this layout or Keep mine. Then move any box or rewrite any element before you render. The assistant’s price shows before you run it.
“A warm vintage poster for a neighborhood bakery called "Golden Crumb", with a basket of fresh bread, a smiling adult baker, and the line "Fresh bread daily" at the bottom.”
From one sentence
Compose with AI placed the baker, the bread, the shop name and the tagline. Here is its layout on each model.
Smiling adult bakerWoven basketGolden CrumbFresh bread dailyWarm vintage background
Two images per model, start to finish on Sogni: MOSS-Image SFT 23.4 s · MOSS-Image 8-Step 7.6 s · FLUX 3 Image Klein 15.0 s.
The prompts each model received
{
"summary": {
"canvas": "single",
"caption": "A warm vintage poster for a neighborhood bakery called \"Golden Crumb\", with a basket of fresh bread, a smiling adult baker, and the line \"Fresh bread daily\" at the bottom."
},
"items": [
{
"label": "character",
"caption": "A smiling adult baker wearing a white apron and chef's hat, standing confidently with arms crossed or resting on the counter, warm friendly expression, vintage illustration style",
"bbox": [
-458,
-156,
142,
156
]
},
{
"label": "object",
"caption": "A woven basket overflowing with fresh golden-brown loaves of bread, crusty texture, placed prominently on the counter or held near the baker",
"bbox": [
-108,
-296,
242,
296
]
},
{
"label": "text",
"caption": "The bakery name 'Golden Crumb' in elegant vintage typography, large and prominent at the top of the poster",
"text": "Golden Crumb",
"bbox": [
-528,
-256,
-428,
256
]
},
{
"label": "text",
"caption": "The slogan 'Fresh bread daily' in clear readable vintage font at the bottom of the poster",
"text": "Fresh bread daily",
"bbox": [
342,
-200,
422,
200
]
},
{
"label": "object",
"caption": "Warm vintage background with soft beige and cream tones, subtle texture suggesting aged paper or poster board, cozy neighborhood bakery ambiance",
"bbox": [
-608,
-416,
608,
416
]
}
],
"bbox_coord_space": "abs_centralized",
"bbox_format": "yxyx"
}A warm vintage poster for a neighborhood bakery called "Golden Crumb", with a basket of fresh bread, a smiling adult baker, and the line "Fresh bread daily" at the bottom. A smiling adult baker wearing a white apron and chef's hat, standing confidently with arms crossed or resting on the counter, warm friendly expression, vintage illustration style <character_1> A woven basket overflowing with fresh golden-brown loaves of bread, crusty texture, placed prominently on the counter or held near the baker <object_2> The bakery name 'Golden Crumb' in elegant vintage typography, large and prominent at the top of the poster The visible text is "Golden Crumb". <Unknown_Text_3> The slogan 'Fresh bread daily' in clear readable vintage font at the bottom of the poster The visible text is "Fresh bread daily". <Unknown_Text_4> Warm vintage background with soft beige and cream tones, subtle texture suggesting aged paper or poster board, cozy neighborhood bakery ambiance <object_5> [{"id":"character_1","bbox":[123,312,617,688],"desc":"A smiling adult baker wearing a white apron and chef's hat, standing confidently with arms crossed or resting on the counter, warm friendly expression, vintage illustration style"},{"id":"object_2","bbox":[411,144,700,856],"desc":"A woven basket overflowing with fresh golden-brown loaves of bread, crusty texture, placed prominently on the counter or held near the baker"},{"id":"Unknown_Text_3","bbox":[65,192,149,808],"desc":"The bakery name 'Golden Crumb' in elegant vintage typography, large and prominent at the top of the poster The visible text is \"Golden Crumb\"."},{"id":"Unknown_Text_4","bbox":[781,259,848,741],"desc":"The slogan 'Fresh bread daily' in clear readable vintage font at the bottom of the poster The visible text is \"Fresh bread daily\"."},{"id":"object_5","bbox":[0,0,1000,1000],"desc":"Warm vintage background with soft beige and cream tones, subtle texture suggesting aged paper or poster board, cozy neighborhood bakery ambiance"}]It can also rework a layout you already have. Asked to turn the travel poster into a moody night scene with a full moon and a small cabin, it kept the EXPLORE title and its box and built the rest around it.
Reworking a layout
The travel poster template, re-planned as a moonlit night scene.
EXPLOREFull moonDark moody mountain silhouettesSmall cabinDark blue mountain illustration
Two images per model, start to finish on Sogni: MOSS-Image SFT 25.9 s · MOSS-Image 8-Step 8.8 s · FLUX 3 Image Klein 7.6 s.
The prompts each model received
{
"summary": {
"meta": [
"poster"
],
"canvas": "single",
"caption": "A moody night-time travel poster featuring a full moon above a mountain range and a small cabin with lit windows. The style is illustrative and atmospheric.",
"palette": [
"dark blue",
"black",
"white",
"yellow"
]
},
"items": [
{
"label": "text",
"caption": "Large dark-blue uppercase lettering centered at the top.",
"text": "EXPLORE",
"bbox": [
-494,
-312,
-342,
312
]
},
{
"label": "object",
"caption": "A full moon glowing brightly in the night sky above the mountain peaks.",
"bbox": [
-328,
-66,
-196,
66
]
},
{
"label": "object",
"caption": "Dark, moody mountain silhouettes rising behind the cabin and under the moon.",
"bbox": [
-308,
-416,
42,
416
]
},
{
"label": "object",
"caption": "A small cabin with warm yellow light glowing from its windows, nestled in the lower center foreground.",
"bbox": [
92,
-234,
342,
34
]
},
{
"label": "object",
"caption": "Dark blue mountain illustration in the lower center, forming the base landscape.",
"bbox": [
-228,
-234,
456,
234
]
}
],
"bbox_coord_space": "abs_centralized",
"bbox_format": "yxyx"
}A moody night-time travel poster featuring a full moon above a mountain range and a small cabin with lit windows. The style is illustrative and atmospheric. Palette: dark blue, black, white, yellow. Large dark-blue uppercase lettering centered at the top. The visible text is "EXPLORE". <Unknown_Text_1> A full moon glowing brightly in the night sky above the mountain peaks. <object_2> Dark, moody mountain silhouettes rising behind the cabin and under the moon. <object_3> A small cabin with warm yellow light glowing from its windows, nestled in the lower center foreground. <object_4> Dark blue mountain illustration in the lower center, forming the base landscape. <object_5> [{"id":"Unknown_Text_1","bbox":[93,125,219,875],"desc":"Large dark-blue uppercase lettering centered at the top. The visible text is \"EXPLORE\"."},{"id":"object_2","bbox":[230,420,339,580],"desc":"A full moon glowing brightly in the night sky above the mountain peaks."},{"id":"object_3","bbox":[246,0,535,1000],"desc":"Dark, moody mountain silhouettes rising behind the cabin and under the moon."},{"id":"object_4","bbox":[575,218,782,541],"desc":"A small cabin with warm yellow light glowing from its windows, nestled in the lower center foreground."},{"id":"object_5","bbox":[312,218,875,782],"desc":"Dark blue mountain illustration in the lower center, forming the base landscape."}]Which model for which layout
| Model | Choose it for | Canvas |
|---|---|---|
| MOSS-Image SFT | The most careful text and placement, in English or Chinese | 512–2048 px per side |
| MOSS-Image 8-Step | Fast drafts of the same layout | 512–2048 px per side |
| FLUX 3 Image Klein | Photoreal people and products, and larger canvases | Up to 4096 px per side |
In our tests on an RTX 4090 at 1024 × 1024 with the model already loaded, FLUX 3 Image Klein took about 2.3 seconds, MOSS-Image 8-Step 3.3 seconds and MOSS-Image SFT 23.8 seconds per image.
Under the hood: two layout formats
Each model reads its own format, and Layout Composer writes whichever one you need. Press JSON in the editor to see or edit the exact prompt. Here is the product campaign in both.
MOSS-Image
{
"summary": {
"meta": [
"product campaign"
],
"canvas": "single",
"caption": "A studio campaign with a small ivory ceramic cup on the left, shown in full including its handle with space around it. A seamless warm background, soft side lighting, tactile speckled glaze and dark-green typography on the right."
},
"items": [
{
"label": "object",
"caption": "One small ivory ceramic cup with a rounded handle and speckled glaze, three-quarter view. The entire cup and handle fit within this region, with space around the silhouette and a soft contact shadow.",
"bbox": [
-184,
-487,
233,
-24
]
},
{
"label": "text",
"caption": "Dark-green serif title in three lines, generous spacing.",
"text": "SAVOR\nTHE\nMOMENT",
"bbox": [
-250,
85,
0,
536
]
},
{
"label": "text",
"caption": "Smaller dark-green supporting copy.",
"text": "A little warmth, every day.",
"bbox": [
99,
109,
208,
523
]
}
],
"bbox_coord_space": "abs_centralized",
"bbox_format": "yxyx"
}FLUX 3 Image Klein
A studio campaign with a small ivory ceramic cup on the left, shown in full including its handle with space around it. A seamless warm background, soft side lighting, tactile speckled glaze and dark-green typography on the right. One small ivory ceramic cup with a rounded handle and speckled glaze, three-quarter view. The entire cup and handle fit within this region, with space around the silhouette and a soft contact shadow. <object_1> Dark-green serif title in three lines, generous spacing. The visible text is "SAVOR\nTHE\nMOMENT". <Unknown_Text_2> Smaller dark-green supporting copy. The visible text is "A little warmth, every day.". <Unknown_Text_3> [{"id":"object_1","bbox":[280,100,780,480],"desc":"One small ivory ceramic cup with a rounded handle and speckled glaze, three-quarter view. The entire cup and handle fit within this region, with space around the silhouette and a soft contact shadow."},{"id":"Unknown_Text_2","bbox":[200,570,500,940],"desc":"Dark-green serif title in three lines, generous spacing. The visible text is \"SAVOR\\nTHE\\nMOMENT\"."},{"id":"Unknown_Text_3","bbox":[620,590,750,930],"desc":"Smaller dark-green supporting copy. The visible text is \"A little warmth, every day.\"."}]- MOSS-Image takes one JSON object.
summarydescribes the whole image, anditemslists each element with alabel(object, text or character), acaption, the exacttextfor text, and abboxof[top, left, bottom, right]in pixels measured from the image’s center. - FLUX 3 Image Klein takes a caption that names each element with a tag such as
<object_1>, followed by a JSON list ofid,bboxanddesc. Its boxes are[top, left, bottom, right]on a 0–1000 grid, and visible words go intodescasThe visible text is "…".
Developers can send either format as the prompt through the Sogni API and SDKs. The MOSS-Image and FLUX 3 Image Klein model guides cover the details.
Tips for layouts that land
- Words go in Visible text. Describe the lettering, such as color, typeface style and alignment, in Appearance.
- Scene sets the look. Put style, lighting, palette and background there, once, for everything.
- One box per thing. Separate objects read better as separate boxes than as one large one.
- Size text for its words. A short word in a huge box comes out huge.
- Boxes guide, they don’t paste. Every take is a new image, so render a few and keep the best.
- Keep what works. Export a layout to reuse it, or reopen any result with Adjust layout.
