[ { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Full-body character design render of Wren, a cheerful 11-year-old lantern courier girl, in the style of a premium 3D animated feature film: soft subsurface skin, stylized proportions with a slightly large head, expressive big hazel eyes, freckles, messy copper-red bob haircut with a small braid. She wears a mustard-yellow hooded raincoat with toggle buttons, a chunky knitted teal scarf, rolled navy trousers, and bright red rubber rain boots. An oversized worn brown leather messenger satchel stuffed with letters hangs across her body, and she holds a small glowing brass lantern in her right hand, arm slightly out from the body. Relaxed standing pose, feet apart, whole body visible from head to boots with generous margin. Camera: eye level, three-quarter view, the character turned about 30 degrees to her left from facing the camera. Plain seamless very light warm-gray studio background, soft even studio lighting from the upper left, soft contact shadow under the feet. Single character, no text, no props on the floor, crisp clean silhouette.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque", "numOutputs": 1 }, "referenceImageCount": 0, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Photorealistic studio product photograph of an antique deep-sea diving helmet, classic twelve-bolt design: hand-hammered copper dome with a warm aged patina and faint green verdigris in the seams, polished brass rims, a round front faceplate porthole plus two side portholes protected by brass grilles, wing nuts, a brass breastplate (corselet) with riveted straps, an air valve elbow at the back, small dents and scratches, a painted serial number plate on the breastplate. The whole helmet and breastplate fully visible and centered with generous margin, resting on nothing. Camera: slightly above eye level, three-quarter view, the helmet turned about 30 degrees to its left from facing the camera. Plain seamless light warm-gray backdrop, soft large softbox lighting from the upper left with gentle rim light, realistic metal reflections, soft contact shadow below. 85mm lens, sharp focus, high detail, no text overlay.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque", "numOutputs": 1 }, "referenceImageCount": 0, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Anime cel-shaded illustration of a retro-futurist delivery scooter with a sidecar, like a key visual from a hand-drawn Japanese animated film: rounded 1960s-style bodywork painted mint-teal and cream, chrome trim, a single round headlight, small windscreen, a tan leather saddle, a bubble-shaped sidecar carrying stacked wooden crates of oranges and a rolled-up rug, a little antenna with an orange pennant flag, a small hover-glow under the wheels. Clean confident ink linework, flat cel shading with two tones plus crisp highlights, saturated but gentle palette. The whole vehicle visible and centered with generous margin, no rider. Camera: slightly above eye level, three-quarter front view, the scooter turned about 30 degrees to its left from facing the camera. Plain seamless off-white background with a soft painted contact shadow under the wheels. No text, no background scenery.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque", "numOutputs": 1 }, "referenceImageCount": 0, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Hand-painted gouache storybook illustration of an old, gentle giant tortoise carrying a tiny two-storey wooden tea house on its shell: curved tiled roof with moss, round paper lantern hanging at the door, little balcony with potted plants, a thin chimney with a curl of smoke, a small wooden ladder down the side of the shell. The tortoise has a wrinkled friendly face with a half-smile, textured shell plates in ochre and olive, stubby elephant-like legs. Visible gouache brush texture, soft painterly edges, warm muted palette of ochre, sage green, terracotta and cream. The whole tortoise and house fully visible and centered with generous margin. Camera: slightly above eye level, three-quarter view, the tortoise turned about 30 degrees to its left from facing the camera, walking calmly. Plain textured cream paper background, soft painted shadow beneath. Single subject, no text, no scenery.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque", "numOutputs": 1 }, "referenceImageCount": 0, "label": "v1" }, { "model": "model_hitem-3d-3-0", "parameters": { "requestType": 1, "resolution": "2048quality", "face": 1000000 }, "referenceImageCount": 1, "label": "v1" }, { "model": "model_scenario-skybox-gpt", "parameters": { "prompt": "A small cobblestone square in a cozy seaside harbor town at blue-hour dusk just after rain, in the style of a premium 3D animated feature film. Wet cobblestones with soft reflections, crooked colorful half-timbered houses with warm glowing windows all around, a stone fountain, a lamppost with a warm light, bunting strung between buildings, distant harbor masts, gentle mist. Standing in the middle of an open empty area of the square at eye level, the ground clearly visible all around. No people, no characters, no text.", "width": 3840, "height": 1920, "quality": "high" }, "referenceImageCount": 1, "label": "v1" }, { "model": "model_scenario-skybox-gpt", "parameters": { "prompt": "Photorealistic interior of an old maritime salvage workshop inside a timber boathouse, late afternoon golden light streaming through tall dusty windows and open sea doors onto calm water. Weathered wide plank floor, coiled ropes, brass lanterns, a workbench with tools, old wooden crates, hanging nets, a vintage compressor, sea charts pinned on the walls, floating dust in the light beams. Viewpoint standing in the middle of an open empty area of the floor at eye level, the floor clearly visible all around. No people, no diving helmet, no text. Natural photographic detail.", "width": 3840, "height": 1920, "quality": "high" }, "referenceImageCount": 1, "label": "v1" }, { "model": "model_scenario-skybox-gpt", "parameters": { "prompt": "Hand-painted anime background art of a sunny Mediterranean-style hillside seaside town street on a summer afternoon, like a background painting from a Japanese animated film. Warm stone road with a painted crosswalk, whitewashed and pastel houses with flower boxes and awnings, a small bakery, laundry lines, a low wall overlooking a sparkling blue sea, tall cumulus clouds in a bright blue sky, a lighthouse on a distant cape. Viewpoint standing in the middle of an open empty stretch of the road at eye level, the road clearly visible all around. No people, no vehicles, no text. Clean linework and flat painted shading.", "width": 3840, "height": 1920, "quality": "high" }, "referenceImageCount": 1, "label": "v1" }, { "model": "model_scenario-skybox-gpt", "parameters": { "prompt": "Hand-painted gouache storybook illustration of a quiet misty meadow clearing at the edge of an old forest in early morning. Soft rolling grass with wildflowers, a winding dirt path, mossy boulders, a small stream, tall pines and birches fading into warm mist, distant blue hills, a pale golden sky. Visible gouache brush texture and soft painterly edges, warm muted palette of ochre, sage green, terracotta and cream. Viewpoint standing in the middle of an open empty patch of meadow at eye level, the ground clearly visible all around. No animals, no houses, no text.", "width": 3840, "height": 1920, "quality": "high" }, "referenceImageCount": 1, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 is the exact final frame layout: a plain gray clay 3D model placed in a finished background, seen from the exact camera angle wanted. Image 2 is the character reference. Repaint the gray clay model as the character from Image 2, as if the camera had walked around her: same identity, face, hair, colors, clothing, materials, props and premium 3D animated feature film finish as Image 2. Follow the clay model exactly: same pose, silhouette, proportions, position and size in frame, and the same camera angle, including which side of the body faces the camera and whether the face is visible. Where Image 2 does not show a part (back, sides), design it consistently with the character. Keep the background of Image 1: same place, composition and perspective, rendered at the same polished quality. Light the character with the scene: warm lamp light, blue dusk ambient, wet cobblestone reflections and a soft contact shadow under her boots. No gray clay left, no extra characters, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 is a plain gray clay 3D model seen from the exact camera angle wanted. Image 2 is the character reference. Repaint the gray clay model as the character from Image 2, as if the camera had walked around her: same identity, face, hair, colors, clothing, materials, props and premium 3D animated feature film finish as Image 2. Follow the clay model exactly: same pose, silhouette, proportions, position and size in frame, and the same camera angle, including which side of the body faces the camera and whether the face is visible. Where Image 2 does not show a part (back, sides), design it consistently with the character. Soft even studio lighting as in Image 2. Transparent background, the full character isolated, no floor, no shadow, no gray clay left, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "transparent" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 shows a plain gray clay 3D model seen from the exact camera angle wanted (ignore its background). Image 2 is the character reference. Repaint the gray clay model as the character from Image 2, as if the camera had walked around her: same identity, face, hair, colors, clothing, materials, props and premium 3D animated feature film finish as Image 2. Follow the clay model exactly: same pose, silhouette, proportions, position and size in frame, and the same camera angle, including which side of the body faces the camera and whether the face is visible. Where Image 2 does not show a part (back, sides), design it consistently with the character. Soft neutral studio lighting exactly like Image 2, no colored scene light. Transparent background: only the isolated character, no environment, no floor, no shadow, no gray clay left, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "transparent" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 is the exact final frame layout: a plain gray clay 3D model placed in a finished background, seen from the exact camera angle wanted. Image 2 is the object reference. Repaint the gray clay model as the antique diving helmet from Image 2, as if the camera had walked around it: same hand-hammered copper dome, aged patina and verdigris, polished brass rims, porthole grilles, bolts, wing nuts, valves and breastplate, photorealistic product photography finish. Follow the clay model exactly: same shape, silhouette, proportions, position and size in frame, and the same camera angle, including which portholes and fittings face the camera. Where Image 2 does not show a part (back, sides), design it consistently with the helmet. Keep the background of Image 1: same place, composition and perspective, photographic quality. Light the helmet with the scene: warm golden afternoon light from the windows, realistic reflections of the workshop in the metal, and a soft contact shadow on the plank floor. No gray clay left, no extra objects, no added text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 shows a plain gray clay 3D model seen from the exact camera angle wanted (ignore its background). Image 2 is the object reference. Repaint the gray clay model as the antique diving helmet from Image 2, as if the camera had walked around it: same hand-hammered copper dome, aged patina and verdigris, polished brass rims, porthole grilles, bolts, wing nuts, valves and breastplate, photorealistic product photography finish. Follow the clay model exactly: same shape, silhouette, proportions, position and size in frame, and the same camera angle, including which portholes and fittings face the camera. Where Image 2 does not show a part (back, sides), design it consistently with the helmet. Soft neutral studio lighting exactly like Image 2, no colored scene light. Transparent background: only the isolated helmet, no environment, no floor, no shadow, no gray clay left, no added text.", "width": 1024, "height": 1024, "quality": "high", "background": "transparent" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 is the exact final frame layout: a plain gray clay 3D model placed in a finished background, seen from the exact camera angle wanted. Image 2 is the vehicle reference. Repaint the gray clay model as the retro-futurist delivery scooter with sidecar from Image 2, as if the camera had walked around it: same mint-teal and cream bodywork, chrome trim, round headlight, windscreen, tan leather saddle, bubble sidecar with crates of oranges and a rolled rug, antenna with orange pennant, soft hover glow under the wheels, and the same anime cel-shaded finish with clean ink linework and flat two-tone shading. Follow the clay model exactly: same shape, silhouette, proportions, position and size in frame, and the same camera angle, including which side of the vehicle and which way its front faces the camera. Where Image 2 does not show a part (back, far side), design it consistently with the vehicle. Keep the background of Image 1: same place, composition and perspective, in the same hand-painted anime background style. Light the vehicle with the scene: bright summer sun, crisp cel shadows and a painted contact shadow under the wheels. No gray clay left, no rider, no extra vehicles, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 shows a plain gray clay 3D model seen from the exact camera angle wanted (ignore its background). Image 2 is the vehicle reference. Repaint the gray clay model as the retro-futurist delivery scooter with sidecar from Image 2, as if the camera had walked around it: same mint-teal and cream bodywork, chrome trim, round headlight, windscreen, tan leather saddle, bubble sidecar with crates of oranges and a rolled rug, antenna with orange pennant, soft hover glow under the wheels, and the same anime cel-shaded finish with clean ink linework and flat two-tone shading. Follow the clay model exactly: same shape, silhouette, proportions, position and size in frame, and the same camera angle, including which side of the vehicle and which way its front faces the camera. Where Image 2 does not show a part (back, far side), design it consistently with the vehicle. Soft even lighting like Image 2. Transparent background: only the isolated vehicle, no environment, no floor, no shadow, no gray clay left, no rider, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "transparent" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 is the exact final frame layout: a plain gray clay 3D model placed in a finished background, seen from the exact camera angle wanted. Image 2 is the creature reference. Repaint the gray clay model as the gentle giant tortoise carrying a tiny wooden tea house from Image 2, as if the camera had walked around it: same wrinkled friendly face, ochre and olive shell plates with moss, stubby legs, two-storey tea house with a curved mossy tiled roof, paper lantern, balcony plants, chimney and wooden ladder, and the same hand-painted gouache storybook finish with visible brush texture. Follow the clay model exactly: same shape, silhouette, proportions, position and size in frame, and the same camera angle, including which way the head faces and which side of the house faces the camera. Where Image 2 does not show a part (back, sides), design it consistently with the tortoise and house. Keep the background of Image 1: same place, composition and perspective, in the same painted gouache style. Light with the scene: soft misty morning light and a painted shadow on the grass. No gray clay left, no extra animals, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 shows a plain gray clay 3D model seen from the exact camera angle wanted (ignore its background). Image 2 is the creature reference. Repaint the gray clay model as the gentle giant tortoise carrying a tiny wooden tea house from Image 2, as if the camera had walked around it: same wrinkled friendly face, ochre and olive shell plates with moss, stubby legs, two-storey tea house with a curved mossy tiled roof, paper lantern, balcony plants, chimney and wooden ladder, and the same hand-painted gouache storybook finish with visible brush texture. Follow the clay model exactly: same shape, silhouette, proportions, position and size in frame, and the same camera angle, including which way the head faces and which side of the house faces the camera. Where Image 2 does not show a part (back, sides), design it consistently with the tortoise and house. Soft even light like Image 2. Transparent background: only the isolated tortoise, no environment, no ground, no shadow, no gray clay left, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "transparent" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 is the exact final frame layout: a plain gray clay 3D model placed in a finished background, seen from the exact camera angle wanted. The camera is high, looking down at her at about 35 degrees. Image 2 is the character reference. Repaint the gray clay model as the character from Image 2, as if the camera had moved around her: same identity, face, hair, colors, clothing, materials, props and premium 3D animated feature film finish as Image 2. Follow the clay model exactly: same pose, silhouette, proportions, position and size in frame, and the same camera angle and height, including which side of the body faces the camera and whether the face is visible. Where Image 2 does not show a part (back, top, sides), design it consistently with the character. Keep the background of Image 1: same place, composition and perspective, rendered at the same polished quality. Light the character with the scene: warm lamp light, blue dusk ambient, wet cobblestone reflections and a soft contact shadow under her boots. No gray clay left, no extra characters, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 is the exact final frame layout: a plain gray clay 3D model placed in a finished background, seen from the exact camera angle wanted. The camera is low near the ground, looking slightly up at her. Image 2 is the character reference. Repaint the gray clay model as the character from Image 2, as if the camera had moved around her: same identity, face, hair, colors, clothing, materials, props and premium 3D animated feature film finish as Image 2. Follow the clay model exactly: same pose, silhouette, proportions, position and size in frame, and the same camera angle and height, including which side of the body faces the camera and whether the face is visible. Where Image 2 does not show a part (back, underside, sides), design it consistently with the character. Keep the background of Image 1: same place, composition and perspective, rendered at the same polished quality. Light the character with the scene: warm lamp light, blue dusk ambient, wet cobblestone reflections and a soft contact shadow under her boots. No gray clay left, no extra characters, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 is the exact final frame layout: a plain gray clay 3D model placed in a finished background, seen from the exact camera angle wanted. The camera is directly overhead, looking straight down at the top of her head, shoulders, satchel and boots. Image 2 is the character reference. Repaint the gray clay model as the character from Image 2, as if the camera had moved above her: same identity, hair, colors, clothing, materials, props and premium 3D animated feature film finish as Image 2. Follow the clay model exactly: same pose, silhouette, proportions, position and size in frame, and the same top-down camera angle. Where Image 2 does not show a part (top, back, sides), design it consistently with the character. Keep the background of Image 1: the same wet cobblestone ground seen from above, rendered at the same polished quality. Light the character with the scene: warm lamp light, blue dusk ambient and a soft contact shadow on the cobbles. No gray clay left, no extra characters, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 shows a plain gray clay 3D model seen from the exact camera angle wanted (ignore its background). The camera is high, looking down at her at about 35 degrees. Image 2 is the character reference. Repaint the gray clay model as the character from Image 2, as if the camera had moved around her: same identity, face, hair, colors, clothing, materials, props and premium 3D animated feature film finish as Image 2. Follow the clay model exactly: same pose, silhouette, proportions, position and size in frame, and the same camera angle and height, including which side of the body faces the camera and whether the face is visible. Where Image 2 does not show a part (back, top, sides), design it consistently with the character. Soft neutral studio lighting exactly like Image 2, no colored scene light. Transparent background: only the isolated character, no environment, no floor, no shadow, no gray clay left, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "transparent" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 shows a plain gray clay 3D model seen from the exact camera angle wanted (ignore its background). The camera is low near the ground, looking slightly up at her. Image 2 is the character reference. Repaint the gray clay model as the character from Image 2, as if the camera had moved around her: same identity, face, hair, colors, clothing, materials, props and premium 3D animated feature film finish as Image 2. Follow the clay model exactly: same pose, silhouette, proportions, position and size in frame, and the same camera angle and height, including which side of the body faces the camera and whether the face is visible. Where Image 2 does not show a part (back, underside, sides), design it consistently with the character. Soft neutral studio lighting exactly like Image 2, no colored scene light. Transparent background: only the isolated character, no environment, no floor, no shadow, no gray clay left, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "transparent" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 shows a plain gray clay 3D model seen from the exact camera angle wanted (ignore its background). The camera is directly overhead, looking straight down at the top of her head, shoulders, satchel and boots. Image 2 is the character reference. Repaint the gray clay model as the character from Image 2, as if the camera had moved above her: same identity, hair, colors, clothing, materials, props and premium 3D animated feature film finish as Image 2. Follow the clay model exactly: same pose, silhouette, proportions, position and size in frame, and the same top-down camera angle. Where Image 2 does not show a part (top, back, sides), design it consistently with the character. Soft neutral studio lighting exactly like Image 2, no colored scene light. Transparent background: only the isolated character, no environment, no floor, no shadow, no gray clay left, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "transparent" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 is the exact final frame layout: a plain gray clay 3D model placed in a finished background, seen from the exact camera angle wanted. The camera is high, looking down at the helmet at about 35 degrees. Image 2 is the object reference. Repaint the gray clay model as the antique diving helmet from Image 2, as if the camera had moved around it: same hand-hammered copper dome, aged patina and verdigris, polished brass rims, porthole grilles, bolts, wing nuts, valves and breastplate, photorealistic product photography finish. Follow the clay model exactly: same shape, silhouette, proportions, position and size in frame, and the same camera angle and height, including which portholes and fittings face the camera. Where Image 2 does not show a part (back, top, sides), design it consistently with the helmet. Keep the background of Image 1: same place, composition and perspective, photographic quality. Light the helmet with the scene: warm golden afternoon light from the windows, realistic reflections of the workshop in the metal, and a soft contact shadow on the plank floor. No gray clay left, no extra objects, no added text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 is the exact final frame layout: a plain gray clay 3D model placed in a finished background, seen from the exact camera angle wanted. The camera is low near the floor, looking slightly up at the helmet. Image 2 is the object reference. Repaint the gray clay model as the antique diving helmet from Image 2, as if the camera had moved around it: same hand-hammered copper dome, aged patina and verdigris, polished brass rims, porthole grilles, bolts, wing nuts, valves and breastplate, photorealistic product photography finish. Follow the clay model exactly: same shape, silhouette, proportions, position and size in frame, and the same camera angle and height, including which portholes and fittings face the camera. Where Image 2 does not show a part (back, underside, sides), design it consistently with the helmet. Keep the background of Image 1: same place, composition and perspective, photographic quality. Light the helmet with the scene: warm golden afternoon light from the windows, realistic reflections of the workshop in the metal, and a soft contact shadow on the plank floor. No gray clay left, no extra objects, no added text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 is the exact final frame layout: a plain gray clay 3D model placed in a finished background, seen from the exact camera angle wanted. The camera is directly overhead, looking straight down at the top of the helmet dome and the breastplate around it. Image 2 is the object reference. Repaint the gray clay model as the antique diving helmet from Image 2, as if the camera had moved above it: same hand-hammered copper dome, aged patina and verdigris, polished brass rims, top grille, bolts, wing nuts, valves and breastplate, photorealistic product photography finish. Follow the clay model exactly: same shape, silhouette, proportions, position and size in frame, and the same top-down camera angle. Where Image 2 does not show a part (top, back, sides), design it consistently with the helmet. Keep the background of Image 1: the same wooden plank floor seen from above, photographic quality. Light the helmet with the scene: warm golden afternoon light and a soft contact shadow on the planks. No gray clay left, no extra objects, no added text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 shows a plain gray clay 3D model seen from the exact camera angle wanted (ignore its background). The camera is high, looking down at the helmet at about 35 degrees. Image 2 is the object reference. Repaint the gray clay model as the antique diving helmet from Image 2, as if the camera had moved around it: same hand-hammered copper dome, aged patina and verdigris, polished brass rims, porthole grilles, bolts, wing nuts, valves and breastplate, photorealistic product photography finish. Follow the clay model exactly: same shape, silhouette, proportions, position and size in frame, and the same camera angle and height, including which portholes and fittings face the camera. Where Image 2 does not show a part (back, top, sides), design it consistently with the helmet. Soft neutral studio lighting exactly like Image 2, no colored scene light. Transparent background: only the isolated helmet, no environment, no floor, no shadow, no gray clay left, no added text.", "width": 1024, "height": 1024, "quality": "high", "background": "transparent" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 shows a plain gray clay 3D model seen from the exact camera angle wanted (ignore its background). The camera is low near the floor, looking slightly up at the helmet. Image 2 is the object reference. Repaint the gray clay model as the antique diving helmet from Image 2, as if the camera had moved around it: same hand-hammered copper dome, aged patina and verdigris, polished brass rims, porthole grilles, bolts, wing nuts, valves and breastplate, photorealistic product photography finish. Follow the clay model exactly: same shape, silhouette, proportions, position and size in frame, and the same camera angle and height, including which portholes and fittings face the camera. Where Image 2 does not show a part (back, underside, sides), design it consistently with the helmet. Soft neutral studio lighting exactly like Image 2, no colored scene light. Transparent background: only the isolated helmet, no environment, no floor, no shadow, no gray clay left, no added text.", "width": 1024, "height": 1024, "quality": "high", "background": "transparent" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 shows a plain gray clay 3D model seen from the exact camera angle wanted (ignore its background). The camera is directly overhead, looking straight down at the top of the helmet dome and the breastplate around it. Image 2 is the object reference. Repaint the gray clay model as the antique diving helmet from Image 2, as if the camera had moved above it: same hand-hammered copper dome, aged patina and verdigris, polished brass rims, top grille, bolts, wing nuts, valves and breastplate, photorealistic product photography finish. Follow the clay model exactly: same shape, silhouette, proportions, position and size in frame, and the same top-down camera angle. Where Image 2 does not show a part (top, back, sides), design it consistently with the helmet. Soft neutral studio lighting exactly like Image 2, no colored scene light. Transparent background: only the isolated helmet, no environment, no floor, no shadow, no gray clay left, no added text.", "width": 1024, "height": 1024, "quality": "high", "background": "transparent" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 is the exact final frame layout: a plain gray clay 3D model placed in a finished background, seen from the exact camera angle wanted. The camera is high, looking down at the vehicle at about 35 degrees. Image 2 is the vehicle reference. Repaint the gray clay model as the retro-futurist delivery scooter with sidecar from Image 2, as if the camera had moved around it: same mint-teal and cream bodywork, chrome trim, round headlight, windscreen, tan leather saddle, bubble sidecar with crates of oranges and a rolled rug, antenna with orange pennant, soft hover glow under the wheels, and the same anime cel-shaded finish with clean ink linework and flat two-tone shading. Follow the clay model exactly: same shape, silhouette, proportions, position and size in frame, and the same camera angle and height, including which side of the vehicle and which way its front faces the camera. Where Image 2 does not show a part (back, top, far side), design it consistently with the vehicle. Keep the background of Image 1: same place, composition and perspective, in the same hand-painted anime background style. Light the vehicle with the scene: bright summer sun, crisp cel shadows and a painted contact shadow under the wheels. No gray clay left, no rider, no extra vehicles, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 is the exact final frame layout: a plain gray clay 3D model placed in a finished background, seen from the exact camera angle wanted. The camera is low near the road, looking slightly up at the vehicle. Image 2 is the vehicle reference. Repaint the gray clay model as the retro-futurist delivery scooter with sidecar from Image 2, as if the camera had moved around it: same mint-teal and cream bodywork, chrome trim, round headlight, windscreen, tan leather saddle, bubble sidecar with crates of oranges and a rolled rug, antenna with orange pennant, soft hover glow under the wheels, and the same anime cel-shaded finish with clean ink linework and flat two-tone shading. Follow the clay model exactly: same shape, silhouette, proportions, position and size in frame, and the same camera angle and height, including which side of the vehicle and which way its front faces the camera. Where Image 2 does not show a part (back, underside, far side), design it consistently with the vehicle. Keep the background of Image 1: same place, composition and perspective, in the same hand-painted anime background style. Light the vehicle with the scene: bright summer sun, crisp cel shadows and a painted contact shadow under the wheels. No gray clay left, no rider, no extra vehicles, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 is the exact final frame layout: a plain gray clay 3D model placed in a finished background, seen from the exact camera angle wanted. The camera is directly overhead, looking straight down at the top of the scooter, saddle, handlebars and the sidecar full of crates. Image 2 is the vehicle reference. Repaint the gray clay model as the retro-futurist delivery scooter with sidecar from Image 2, as if the camera had moved above it: same mint-teal and cream bodywork, chrome trim, windscreen, tan leather saddle, bubble sidecar with crates of oranges and a rolled rug, antenna with orange pennant, and the same anime cel-shaded finish with clean ink linework and flat two-tone shading. Follow the clay model exactly: same shape, silhouette, proportions, position and size in frame, and the same top-down camera angle. Where Image 2 does not show a part (top, back, far side), design it consistently with the vehicle. Keep the background of Image 1: the same paved road seen from above, in the same hand-painted anime background style. Bright summer sun and a crisp painted shadow on the road. No gray clay left, no rider, no extra vehicles, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 shows a plain gray clay 3D model seen from the exact camera angle wanted (ignore its background). The camera is high, looking down at the vehicle at about 35 degrees. Image 2 is the vehicle reference. Repaint the gray clay model as the retro-futurist delivery scooter with sidecar from Image 2, as if the camera had moved around it: same mint-teal and cream bodywork, chrome trim, round headlight, windscreen, tan leather saddle, bubble sidecar with crates of oranges and a rolled rug, antenna with orange pennant, soft hover glow under the wheels, and the same anime cel-shaded finish with clean ink linework and flat two-tone shading. Follow the clay model exactly: same shape, silhouette, proportions, position and size in frame, and the same camera angle and height, including which side of the vehicle and which way its front faces the camera. Where Image 2 does not show a part (back, top, far side), design it consistently with the vehicle. Soft even lighting like Image 2. Transparent background: only the isolated vehicle, no environment, no floor, no shadow, no gray clay left, no rider, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "transparent" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 shows a plain gray clay 3D model seen from the exact camera angle wanted (ignore its background). The camera is low near the road, looking slightly up at the vehicle. Image 2 is the vehicle reference. Repaint the gray clay model as the retro-futurist delivery scooter with sidecar from Image 2, as if the camera had moved around it: same mint-teal and cream bodywork, chrome trim, round headlight, windscreen, tan leather saddle, bubble sidecar with crates of oranges and a rolled rug, antenna with orange pennant, soft hover glow under the wheels, and the same anime cel-shaded finish with clean ink linework and flat two-tone shading. Follow the clay model exactly: same shape, silhouette, proportions, position and size in frame, and the same camera angle and height, including which side of the vehicle and which way its front faces the camera. Where Image 2 does not show a part (back, underside, far side), design it consistently with the vehicle. Soft even lighting like Image 2. Transparent background: only the isolated vehicle, no environment, no floor, no shadow, no gray clay left, no rider, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "transparent" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 shows a plain gray clay 3D model seen from the exact camera angle wanted (ignore its background). The camera is directly overhead, looking straight down at the top of the scooter, saddle, handlebars and the sidecar full of crates. Image 2 is the vehicle reference. Repaint the gray clay model as the retro-futurist delivery scooter with sidecar from Image 2, as if the camera had moved above it: same mint-teal and cream bodywork, chrome trim, windscreen, tan leather saddle, bubble sidecar with crates of oranges and a rolled rug, antenna with orange pennant, and the same anime cel-shaded finish with clean ink linework and flat two-tone shading. Follow the clay model exactly: same shape, silhouette, proportions, position and size in frame, and the same top-down camera angle. Where Image 2 does not show a part (top, back, far side), design it consistently with the vehicle. Soft even lighting like Image 2. Transparent background: only the isolated vehicle, no environment, no floor, no shadow, no gray clay left, no rider, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "transparent" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 is the exact final frame layout: a plain gray clay 3D model placed in a finished background, seen from the exact camera angle wanted. The camera is high, looking down at the tortoise at about 35 degrees. Image 2 is the creature reference. Repaint the gray clay model as the gentle giant tortoise carrying a tiny wooden tea house from Image 2, as if the camera had moved around it: same wrinkled friendly face, ochre and olive shell plates with moss, stubby legs, two-storey tea house with a curved mossy tiled roof, paper lantern, balcony plants, chimney and wooden ladder, and the same hand-painted gouache storybook finish with visible brush texture. Follow the clay model exactly: same shape, silhouette, proportions, position and size in frame, and the same camera angle and height, including which way the head faces and which side of the house faces the camera. Where Image 2 does not show a part (back, roof, sides), design it consistently with the tortoise and house. Keep the background of Image 1: same place, composition and perspective, in the same painted gouache style. Light with the scene: soft misty morning light and a painted shadow on the grass. No gray clay left, no extra animals, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 is the exact final frame layout: a plain gray clay 3D model placed in a finished background, seen from the exact camera angle wanted. The camera is low near the grass, looking slightly up at the tortoise. Image 2 is the creature reference. Repaint the gray clay model as the gentle giant tortoise carrying a tiny wooden tea house from Image 2, as if the camera had moved around it: same wrinkled friendly face, ochre and olive shell plates with moss, stubby legs, two-storey tea house with a curved mossy tiled roof, paper lantern, balcony plants, chimney and wooden ladder, and the same hand-painted gouache storybook finish with visible brush texture. Follow the clay model exactly: same shape, silhouette, proportions, position and size in frame, and the same camera angle and height, including which way the head faces and which side of the house faces the camera. Where Image 2 does not show a part (back, underside, sides), design it consistently with the tortoise and house. Keep the background of Image 1: same place, composition and perspective, in the same painted gouache style. Light with the scene: soft misty morning light and a painted shadow on the grass. No gray clay left, no extra animals, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 is the exact final frame layout: a plain gray clay 3D model placed in a finished background, seen from the exact camera angle wanted. The camera is directly overhead, looking straight down at the tea house roof, chimney, balcony and the tortoise shell, head and legs around it. Image 2 is the creature reference. Repaint the gray clay model as the gentle giant tortoise carrying a tiny wooden tea house from Image 2, as if the camera had moved above it: same ochre and olive shell plates with moss, stubby legs, curved mossy tiled roof, paper lantern, balcony plants, chimney and wooden ladder, and the same hand-painted gouache storybook finish with visible brush texture. Follow the clay model exactly: same shape, silhouette, proportions, position and size in frame, and the same top-down camera angle. Where Image 2 does not show a part (roof, back, sides), design it consistently with the tortoise and house. Keep the background of Image 1: the same meadow grass seen from above, in the same painted gouache style. Soft misty morning light and a painted shadow on the grass. No gray clay left, no extra animals, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 shows a plain gray clay 3D model seen from the exact camera angle wanted (ignore its background). The camera is high, looking down at the tortoise at about 35 degrees. Image 2 is the creature reference. Repaint the gray clay model as the gentle giant tortoise carrying a tiny wooden tea house from Image 2, as if the camera had moved around it: same wrinkled friendly face, ochre and olive shell plates with moss, stubby legs, two-storey tea house with a curved mossy tiled roof, paper lantern, balcony plants, chimney and wooden ladder, and the same hand-painted gouache storybook finish with visible brush texture. Follow the clay model exactly: same shape, silhouette, proportions, position and size in frame, and the same camera angle and height, including which way the head faces and which side of the house faces the camera. Where Image 2 does not show a part (back, roof, sides), design it consistently with the tortoise and house. Soft even light like Image 2. Transparent background: only the isolated tortoise, no environment, no ground, no shadow, no gray clay left, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "transparent" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 shows a plain gray clay 3D model seen from the exact camera angle wanted (ignore its background). The camera is low near the grass, looking slightly up at the tortoise. Image 2 is the creature reference. Repaint the gray clay model as the gentle giant tortoise carrying a tiny wooden tea house from Image 2, as if the camera had moved around it: same wrinkled friendly face, ochre and olive shell plates with moss, stubby legs, two-storey tea house with a curved mossy tiled roof, paper lantern, balcony plants, chimney and wooden ladder, and the same hand-painted gouache storybook finish with visible brush texture. Follow the clay model exactly: same shape, silhouette, proportions, position and size in frame, and the same camera angle and height, including which way the head faces and which side of the house faces the camera. Where Image 2 does not show a part (back, underside, sides), design it consistently with the tortoise and house. Soft even light like Image 2. Transparent background: only the isolated tortoise, no environment, no ground, no shadow, no gray clay left, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "transparent" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "parameters": { "prompt": "Image 1 shows a plain gray clay 3D model seen from the exact camera angle wanted (ignore its background). The camera is directly overhead, looking straight down at the tea house roof, chimney, balcony and the tortoise shell, head and legs around it. Image 2 is the creature reference. Repaint the gray clay model as the gentle giant tortoise carrying a tiny wooden tea house from Image 2, as if the camera had moved above it: same ochre and olive shell plates with moss, stubby legs, curved mossy tiled roof, paper lantern, balcony plants, chimney and wooden ladder, and the same hand-painted gouache storybook finish with visible brush texture. Follow the clay model exactly: same shape, silhouette, proportions, position and size in frame, and the same top-down camera angle. Where Image 2 does not show a part (roof, back, sides), design it consistently with the tortoise and house. Soft even light like Image 2. Transparent background: only the isolated tortoise, no environment, no ground, no shadow, no gray clay left, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "transparent" }, "referenceImageCount": 2, "label": "v1" }, { "model": "model_openai-gpt-image-2-5-sunburst", "label": "v2 source picture: dwarf (Brakka)", "parameters": { "prompt": "Hand-painted stylized fantasy game character art of Brakka, a stout dwarf blacksmith warrior, in the style of a premium hand-painted stylized fantasy video game: chunky exaggerated proportions, broad shoulders, big hands and boots, a long braided copper-red beard with small brass rings, a riveted steel helmet with short horns, a thick leather apron over chainmail, brass pauldrons, a heavy two-handed forge hammer held diagonally across the body in both hands, a round wooden shield with an iron rim strapped on the back, a pouch belt with tools. Painterly textures, crisp readable shapes, warm saturated palette of copper, brass, leather brown and steel blue. Relaxed confident standing pose, feet apart, the whole body visible from helmet to boots with generous margin. Camera: slightly above eye level, three-quarter view, the character turned about 30 degrees to his left from facing the camera. Plain seamless light warm-gray studio background, soft even lighting, soft contact shadow below. Single character, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque", "numOutputs": 1 }, "referenceImageCount": 0 }, { "model": "model_hitem-3d-3-0", "label": "v2 mesh: dwarf (untextured)", "parameters": { "images": [""], "requestType": 1, "resolution": "2048quality", "face": 1000000 }, "referenceImageCount": 1 }, { "model": "model_scenario-skybox-gpt", "label": "v2 panorama: dwarven forge hall", "parameters": { "prompt": "Hand-painted stylized fantasy game environment art of a great dwarven forge hall carved inside a mountain, warm glowing forges and lava channels, massive stone pillars with geometric dwarven carvings, brass pipes and chains, anvils and weapon racks along the walls, hanging lanterns, a wide flat flagstone floor, sparks and a little smoke in the air, warm orange firelight mixed with cool blue light from high openings. Painterly textures, crisp readable shapes, warm saturated palette. Viewpoint standing in the middle of an open empty area of the floor at eye level, the floor clearly visible all around. No people, no characters, no text.", "width": 3840, "height": 1920, "quality": "high" }, "referenceImageCount": 0 }, { "model": "model_openai-gpt-image-2-5-sunburst", "label": "v2 grounded repaint: scooter (25 jobs, references: [grounded layout, original picture, v1 studio view of the same camera])", "parameters": { "prompt": "Image 1 is the exact final frame: a plain gray clay 3D model standing on the ground of a finished background, with its real cast shadow, seen from the exact camera angle wanted. Image 2 is the design reference of the vehicle. Image 3 shows this same vehicle already painted from this exact camera angle on a plain background. Repaint the gray clay model as the retro-futurist delivery scooter with sidecar, using Image 3 for every detail visible from this angle and Image 2 for the overall design: mint-teal and cream bodywork, chrome trim, round headlight, windscreen, tan leather saddle, bubble sidecar with crates of oranges and a rolled rug, antenna with orange pennant, soft hover glow under the wheels, anime cel-shaded finish with clean ink linework and flat two-tone shading. Follow the clay model exactly: same silhouette, proportions, position and size in frame, and the same camera angle. The wheels touch the ground exactly where the clay touches it; keep the cast shadow of Image 1. Keep the background of Image 1 unchanged: same place, composition, perspective and horizon, in the same hand-painted anime background style, crisp and detailed. Light the vehicle with the scene: bright summer sun matching the shadow direction. No gray clay left, no rider, no extra vehicles, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 3 }, { "model": "model_openai-gpt-image-2-5-sunburst", "label": "v2 grounded repaint: wren (25 jobs, references: [grounded layout, original picture, v1 studio view of the same camera])", "parameters": { "prompt": "Image 1 is the exact final frame: a plain gray clay 3D model standing on the ground of a finished background, with its real cast shadow, seen from the exact camera angle wanted. Image 2 is the original design reference of the character. Image 3 shows the same character already painted from this exact camera angle on a plain background. Repaint the gray clay model as Wren, the cheerful young lantern courier girl, using Image 3 for the details visible from this angle and Image 2 for identity and design: copper-red messy bob with a small braid, freckles, big hazel eyes, mustard-yellow hooded raincoat with toggle buttons, chunky teal knitted scarf, rolled navy trousers, bright red rubber boots, worn brown leather satchel stuffed with letters, small glowing brass lantern in her hand. premium 3D animated feature film finish: soft subsurface skin, stylized proportions, rich materials. Follow the clay model exactly: same silhouette, pose, proportions, position and size in frame, and the same camera angle. Her boots stand on the wet cobblestones exactly where the clay feet touch the ground; keep the cast shadow of Image 1. Keep the background of Image 1: same place, composition, perspective and horizon, in the same premium 3D animated feature film rendering, crisp and detailed. Light the character with the scene: blue-hour dusk after rain: cool ambient light, warm glow from the lamps and from her lantern, soft reflections on the wet cobblestones. No gray clay left, no extra characters or objects, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 3 }, { "model": "model_openai-gpt-image-2-5-sunburst", "label": "v2 grounded repaint: helmet (25 jobs, references: [grounded layout, original picture, v1 studio view of the same camera])", "parameters": { "prompt": "Image 1 is the exact final frame: a plain gray clay 3D model standing on the ground of a finished background, with its real cast shadow, seen from the exact camera angle wanted. Image 2 is the original design reference of the prop. Image 3 shows the same prop already painted from this exact camera angle on a plain background. Repaint the gray clay model as the antique twelve-bolt deep-sea diving helmet, using Image 3 for the details visible from this angle and Image 2 for identity and design: hand-hammered copper dome with aged patina and faint verdigris, polished brass rims, front and side portholes with brass grilles, wing nuts, riveted brass breastplate, air valve elbow, small dents and scratches. photorealistic product photography, realistic metal reflections, sharp focus. Follow the clay model exactly: same silhouette, pose, proportions, position and size in frame, and the same camera angle. The breastplate rests on the wooden floor exactly where the clay touches it; keep the cast shadow of Image 1. Keep the background of Image 1: same place, composition, perspective and horizon, in the same natural photographic detail, crisp and detailed. Light the prop with the scene: late-afternoon golden light from the windows, warm highlights on the copper, soft bounce light from the wooden floor. No gray clay left, no extra characters or objects, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 3 }, { "model": "model_openai-gpt-image-2-5-sunburst", "label": "v2 grounded repaint: tortoise (25 jobs, references: [grounded layout, original picture, v1 studio view of the same camera])", "parameters": { "prompt": "Image 1 is the exact final frame: a plain gray clay 3D model standing on the ground of a finished background, with its real cast shadow, seen from the exact camera angle wanted. Image 2 is the original design reference of the creature. Image 3 shows the same creature already painted from this exact camera angle on a plain background. Repaint the gray clay model as the gentle giant tortoise carrying a tiny tea house on its shell, using Image 3 for the details visible from this angle and Image 2 for identity and design: wrinkled friendly face with a half-smile, ochre and olive shell plates, stubby legs, two-storey wooden tea house with a mossy curved tiled roof, round paper lantern, small balcony with potted plants, thin chimney with a curl of smoke, small wooden ladder down the shell. hand-painted gouache storybook illustration with visible brush texture and soft painterly edges, warm muted palette. Follow the clay model exactly: same silhouette, pose, proportions, position and size in frame, and the same camera angle. Its feet stand in the meadow grass exactly where the clay touches the ground; keep the cast shadow of Image 1. Keep the background of Image 1: same place, composition, perspective and horizon, in the same hand-painted gouache storybook style, crisp and detailed. Light the creature with the scene: soft misty dawn light, gentle warm glow from the paper lantern. No gray clay left, no extra characters or objects, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 3 }, { "model": "model_openai-gpt-image-2-5-sunburst", "label": "v2 grounded repaint: wren (2 jobs, references: [grounded layout, original picture])", "parameters": { "prompt": "Image 1 is the exact final frame: a plain gray clay 3D model standing on the ground of a finished background, with its real cast shadow, seen from the exact camera angle wanted. Image 2 is the original design reference of the character. Repaint the gray clay model as Wren, the cheerful young lantern courier girl, matching Image 2 for identity and design: copper-red messy bob with a small braid, mustard-yellow hooded raincoat with toggle buttons and its hood hanging down her back, chunky teal knitted scarf, rolled navy trousers, bright red rubber boots, one worn brown leather satchel stuffed with letters on a strap at her side, small glowing brass lantern in her hand. She carries exactly one bag, the brown satchel: no backpack, no second bag, no sack on her back; the rounded shape on her back is the yellow raincoat hood. Premium 3D animated feature film finish: soft subsurface skin, stylized proportions, rich materials. Follow the clay model exactly: same silhouette, pose, proportions, position and size in frame, and the same camera angle. Her boots stand on the wet cobblestones exactly where the clay feet touch the ground; keep the cast shadow of Image 1. Keep the background of Image 1: same place, composition, perspective and horizon, in the same premium 3D animated feature film rendering, crisp and detailed. Light the character with the scene: blue-hour dusk after rain: cool ambient light, warm glow from the lamps and from her lantern, soft reflections on the wet cobblestones. No gray clay left, no extra characters or objects, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2 }, { "model": "model_openai-gpt-image-2-5-sunburst", "label": "v2 grounded repaint: dwarf (25 jobs, references: [grounded layout, original picture])", "parameters": { "prompt": "Image 1 is the exact final frame: a plain gray clay 3D model standing on the ground of a finished background, with its real cast shadow, seen from the exact camera angle wanted. Image 2 is the original design reference of the character. Repaint the gray clay model as Brakka, the stout dwarf blacksmith warrior, matching Image 2 for identity, design, materials and colors: long braided copper-red beard with brass rings, riveted steel helmet with short horns, thick leather apron over chainmail, brass pauldrons with knotwork, heavy two-handed forge hammer with brass bands, round wooden shield with an iron rim on his back, belt with tongs, pouches and tools, big armored boots. Seen from behind, the round wooden shield with its iron rim covers his back; the hammer head is dark hammered steel with brass bands. Stylized fantasy game character render: chunky proportions, rich worn leather, hammered steel and brass, crisp readable shapes, warm saturated palette. Follow the clay model exactly: same silhouette, pose, proportions, position and size in frame, and the same camera angle. His boots stand on the flagstones exactly where the clay feet touch the ground; keep the cast shadow of Image 1. Keep the background of Image 1: same place, composition, perspective and horizon, in the same hand-painted stylized fantasy game environment style, crisp and detailed. Light the character with the scene: warm orange firelight from the forges with a cool blue fill from above. No gray clay left, no extra characters or objects, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2 }, { "model": "model_openai-gpt-image-2-5-sunburst", "label": "v2 source picture: robot (Pip)", "parameters": { "prompt": "Stop-motion claymation character photograph of Pip, a small friendly retro robot gardener, in the style of a handcrafted stop-motion animated film: visible fingerprint texture in the plasticine, slightly imperfect hand-sculpted shapes, felt and painted wood details. A round dented tin-can body painted pastel mint green with a small round porthole window on the chest showing a sprouting seedling inside, big round glass eyes with brass rims, a single antenna ending in a little yellow flower, stubby jointed arms holding a small orange watering can, short legs on two rubber wheels, a tiny canvas tool belt with a trowel. Relaxed standing pose, the whole robot visible from antenna to wheels with generous margin. Camera: slightly above eye level, three-quarter view, the robot turned about 30 degrees to its left from facing the camera. Plain seamless light warm-gray studio background, soft even lighting like a miniature film set, soft contact shadow below. Single character, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque", "numOutputs": 1 }, "referenceImageCount": 0 }, { "model": "model_hitem-3d-3-0", "label": "v2 mesh: robot (untextured)", "parameters": { "images": [""], "requestType": 1, "resolution": "2048quality", "face": 1000000 }, "referenceImageCount": 1 }, { "model": "model_scenario-skybox-gpt", "label": "v2 panorama: stop-motion greenhouse", "parameters": { "prompt": "Handcrafted stop-motion miniature film set of a cozy Victorian glass greenhouse, as in a claymation animated film: plasticine and felt plants, terracotta pots, seed trays and hanging baskets on wooden benches, curved white iron frames with slightly foggy glass panes, a watering hose coiled on a hook, warm late-afternoon sunlight streaming through the glass with soft dust in the air. Visible handmade textures and fingerprints on the clay. Viewpoint standing in the middle of an open empty area of the floor at eye level, a wide flat terracotta tile floor clearly visible all around. No people, no characters, no text.", "width": 3840, "height": 1920, "quality": "high" }, "referenceImageCount": 0 }, { "model": "model_openai-gpt-image-2-5-sunburst", "label": "v2 source picture: lighthouse islet", "parameters": { "prompt": "Stylized low-poly 3D diorama render of a small red and white striped lighthouse standing on a compact rocky islet, in the style of a polished low-poly isometric game asset: clean faceted geometry, flat-shaded facets, soft pastel palette. The lighthouse has a white and red banded tower, a black iron gallery with railing near the top, a glowing yellow lamp room with a dark red domed cap, a small arched blue door at the base. Beside it a tiny white keeper's cottage with a red roof and a chimney, a short wooden jetty with a little rowing boat, a few low-poly pine trees and grassy tufts on the grey rocks. The islet is a compact self-contained base, the whole diorama visible with generous margin. Camera: slightly above eye level, three-quarter view, turned about 30 degrees from facing the camera. Plain seamless light warm-gray studio background, soft even lighting, soft contact shadow below. No water around the base, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque", "numOutputs": 1 }, "referenceImageCount": 0 }, { "model": "model_hitem-3d-3-0", "label": "v2 mesh: lighthouse islet (untextured)", "parameters": { "images": [""], "requestType": 1, "resolution": "2048quality", "face": 1000000 }, "referenceImageCount": 1 }, { "model": "model_scenario-skybox-gpt", "label": "v2 panorama: low-poly beach at golden hour", "parameters": { "prompt": "Stylized low-poly 3D game environment of a calm coastal bay at golden hour: clean faceted geometry, flat-shaded facets, soft pastel palette. A wide flat pale sandy beach at low tide with a few faceted grey rocks, tide pools and low-poly seashells, gentle turquoise low-poly sea with faceted waves in the distance, distant faceted green headlands and cliffs, a few low-poly pine trees, puffy faceted clouds and a warm peach sky. Viewpoint standing in the middle of the open empty flat sand at eye level, the flat beach floor clearly visible all around. No people, no buildings, no boats, no text.", "width": 3840, "height": 1920, "quality": "high" }, "referenceImageCount": 0 }, { "model": "model_openai-gpt-image-2-5-sunburst", "label": "v2 grounded repaint: robot (25 jobs, references: [grounded layout, original picture])", "parameters": { "prompt": "Image 1 is the exact final frame: a plain gray clay 3D model standing on the ground of a finished background, with its real cast shadow, seen from the exact camera angle wanted. Image 2 is the original design reference of the character. Repaint the gray clay model as Pip, the small friendly retro robot gardener, matching Image 2 for identity, design, materials and colors: round dented tin-can body painted pastel mint green with rust spots, a round porthole window on the chest with a sprouting seedling inside, big round glass eyes with brass rims, a single antenna ending in a small yellow flower, stubby jointed arms holding a small orange watering can, short legs on two rubber wheels, a canvas tool belt with a trowel. Handcrafted stop-motion claymation look: hand-sculpted plasticine with visible fingerprints, painted metal and felt details, miniature film-set lighting. Follow the clay model exactly: same silhouette, pose, proportions, position and size in frame, and the same camera angle. Its wheels stand on the terracotta floor tiles exactly where the clay wheels touch the ground; keep the cast shadow of Image 1. Keep the background of Image 1: same place, composition, perspective and horizon, in the same handcrafted stop-motion miniature set style, crisp and detailed. Light the character with the scene: warm late-afternoon sunlight through the greenhouse glass with soft fill. No gray clay left, no extra characters or objects, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2 }, { "model": "model_openai-gpt-image-2-5-sunburst", "label": "v2 grounded repaint: lighthouse (25 jobs, references: [grounded layout, original picture])", "parameters": { "prompt": "Image 1 is the exact final frame: a plain gray clay 3D model standing on the ground of a finished background, with its real cast shadow, seen from the exact camera angle wanted. Image 2 is the original design reference of the building. Repaint the gray clay model as the small red and white striped lighthouse diorama on its rocky islet, matching Image 2 for identity, design, materials and colors: white and red banded tower, black iron gallery with railing, glowing yellow lamp room with a dark red domed cap, small arched blue door at the base, a tiny white keeper's cottage with a red roof and chimney, a short wooden jetty with a little white rowing boat, low-poly pine trees and grass tufts on faceted grey rocks. Stylized low-poly 3D game asset: clean faceted geometry, flat-shaded facets, soft pastel palette. Follow the clay model exactly: same silhouette, pose, proportions, position and size in frame, and the same camera angle. The rocky base sits on the sand exactly where the clay base touches the ground; keep the cast shadow of Image 1. Keep the background of Image 1: same place, composition, perspective and horizon, in the same stylized low-poly flat-shaded style, crisp and detailed. Light the building with the scene: warm golden-hour sunlight with a soft peach sky fill. No gray clay left, no extra characters or objects, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2 }, { "model": "model_birefnet-background-removal", "label": "v2 No background: cutout of each v2 grounded repaint (175 jobs)", "parameters": { "image": "" }, "referenceImageCount": 0 }, { "model": "model_openai-gpt-image-2-5-sunburst", "label": "v3 structural fix: lighthouse 03 (1 job, references: [grounded layout, original picture]; the added lines name what this camera can and cannot see)", "parameters": { "prompt": "Image 1 is the exact final frame: a plain gray clay 3D model standing on the ground of a finished background, with its real cast shadow, seen from the exact camera angle wanted. Image 2 is the original design reference of the building. Repaint the gray clay model as the small red and white striped lighthouse diorama on its rocky islet, matching Image 2 for identity, design, materials and colors: white and red banded tower, black iron gallery with railing, glowing yellow lamp room with a dark red domed cap, small arched blue door at the base, a tiny white keeper's cottage with a red roof and chimney, a short wooden jetty with a little white rowing boat, low-poly pine trees and grass tufts on faceted grey rocks. The clay model in Image 1 is the truth for which side of the building faces the camera: paint only what the clay shows from this angle, even where Image 2 shows another side. Follow the clay: the cottage sits in front of the tower, the jetty is at the far left edge of the rocks, and the blue door is not visible. Stylized low-poly 3D game asset: clean faceted geometry, flat-shaded facets, soft pastel palette. Follow the clay model exactly: same silhouette, pose, proportions, position and size in frame, and the same camera angle. The rocky base sits on the sand exactly where the clay base touches the ground; keep the cast shadow of Image 1. Keep the background of Image 1: same place, composition, perspective and horizon, in the same stylized low-poly flat-shaded style, crisp and detailed. Light the building with the scene: warm golden-hour sunlight with a soft peach sky fill. No gray clay left, no extra characters or objects, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2 }, { "model": "model_openai-gpt-image-2-5-sunburst", "label": "v3 structural fix: lighthouse 05, 06, 07, 08, 09, high_08, low_08 (7 jobs, references: [grounded layout, original picture]; the added lines name what this camera can and cannot see)", "parameters": { "prompt": "Image 1 is the exact final frame: a plain gray clay 3D model standing on the ground of a finished background, with its real cast shadow, seen from the exact camera angle wanted. Image 2 is the original design reference of the building. Repaint the gray clay model as the small red and white striped lighthouse diorama on its rocky islet, matching Image 2 for identity, design, materials and colors: white and red banded tower, black iron gallery with railing, glowing yellow lamp room with a dark red domed cap, small arched blue door at the base, a tiny white keeper's cottage with a red roof and chimney, a short wooden jetty with a little white rowing boat, low-poly pine trees and grass tufts on faceted grey rocks. The clay model in Image 1 is the truth for which side of the building faces the camera: paint only what the clay shows from this angle, even where Image 2 shows another side. This camera sees the BACK of the islet. The blue door, the stone stairs, the wooden jetty and the rowing boat are on the far side and must not appear. Paint the back of the striped tower, the back of the cottage, the pines and the rocks exactly where the clay has them. Stylized low-poly 3D game asset: clean faceted geometry, flat-shaded facets, soft pastel palette. Follow the clay model exactly: same silhouette, pose, proportions, position and size in frame, and the same camera angle. The rocky base sits on the sand exactly where the clay base touches the ground; keep the cast shadow of Image 1. Keep the background of Image 1: same place, composition, perspective and horizon, in the same stylized low-poly flat-shaded style, crisp and detailed. Light the building with the scene: warm golden-hour sunlight with a soft peach sky fill. No gray clay left, no extra characters or objects, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2 }, { "model": "model_openai-gpt-image-2-5-sunburst", "label": "v3 structural fix: lighthouse 10 (1 job, references: [grounded layout, original picture]; the added lines name what this camera can and cannot see)", "parameters": { "prompt": "Image 1 is the exact final frame: a plain gray clay 3D model standing on the ground of a finished background, with its real cast shadow, seen from the exact camera angle wanted. Image 2 is the original design reference of the building. Repaint the gray clay model as the small red and white striped lighthouse diorama on its rocky islet, matching Image 2 for identity, design, materials and colors: white and red banded tower, black iron gallery with railing, glowing yellow lamp room with a dark red domed cap, small arched blue door at the base, a tiny white keeper's cottage with a red roof and chimney, a short wooden jetty with a little white rowing boat, low-poly pine trees and grass tufts on faceted grey rocks. The clay model in Image 1 is the truth for which side of the building faces the camera: paint only what the clay shows from this angle, even where Image 2 shows another side. Follow the clay: the cottage is on the left of the tower, pines stand in front, and there is no jetty, no boat and no door on this side. Stylized low-poly 3D game asset: clean faceted geometry, flat-shaded facets, soft pastel palette. Follow the clay model exactly: same silhouette, pose, proportions, position and size in frame, and the same camera angle. The rocky base sits on the sand exactly where the clay base touches the ground; keep the cast shadow of Image 1. Keep the background of Image 1: same place, composition, perspective and horizon, in the same stylized low-poly flat-shaded style, crisp and detailed. Light the building with the scene: warm golden-hour sunlight with a soft peach sky fill. No gray clay left, no extra characters or objects, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2 }, { "model": "model_openai-gpt-image-2-5-sunburst", "label": "v3 structural fix: lighthouse low_04 (1 job, references: [grounded layout, original picture]; the added lines name what this camera can and cannot see)", "parameters": { "prompt": "Image 1 is the exact final frame: a plain gray clay 3D model standing on the ground of a finished background, with its real cast shadow, seen from the exact camera angle wanted. Image 2 is the original design reference of the building. Repaint the gray clay model as the small red and white striped lighthouse diorama on its rocky islet, matching Image 2 for identity, design, materials and colors: white and red banded tower, black iron gallery with railing, glowing yellow lamp room with a dark red domed cap, small arched blue door at the base, a tiny white keeper's cottage with a red roof and chimney, a short wooden jetty with a little white rowing boat, low-poly pine trees and grass tufts on faceted grey rocks. The clay model in Image 1 is the truth for which side of the building faces the camera: paint only what the clay shows from this angle, even where Image 2 shows another side. Follow the clay: pines and rocks are in front, the jetty is only at the far left edge, and the blue door is not visible. Stylized low-poly 3D game asset: clean faceted geometry, flat-shaded facets, soft pastel palette. Follow the clay model exactly: same silhouette, pose, proportions, position and size in frame, and the same camera angle. The rocky base sits on the sand exactly where the clay base touches the ground; keep the cast shadow of Image 1. Keep the background of Image 1: same place, composition, perspective and horizon, in the same stylized low-poly flat-shaded style, crisp and detailed. Light the building with the scene: warm golden-hour sunlight with a soft peach sky fill. No gray clay left, no extra characters or objects, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2 }, { "model": "model_openai-gpt-image-2-5-sunburst", "label": "v3 structural fix: robot 05, 07, 08, high_08 (4 jobs, references: [grounded layout, original picture]; the added lines name what this camera can and cannot see)", "parameters": { "prompt": "Image 1 is the exact final frame: a plain gray clay 3D model standing on the ground of a finished background, with its real cast shadow, seen from the exact camera angle wanted. Image 2 is the original design reference of the character. Repaint the gray clay model as Pip, the small friendly retro robot gardener, matching Image 2 for identity, design, materials and colors: round dented tin-can body painted pastel mint green with rust spots, a round porthole window on the chest with a sprouting seedling inside, big round glass eyes with brass rims, a single antenna ending in a small yellow flower, stubby jointed arms holding a small orange watering can, short legs on two rubber wheels, a canvas tool belt with a trowel. The clay model in Image 1 is the truth for which side of the character faces the camera: paint only what the clay shows from this angle, even where Image 2 shows another side. This camera sees Pip from behind: the back of the round head with no eyes and no face, the back of the body as plain dented mint tin with rust spots, with no porthole and no seedling on the back, and the tool belt seen from behind with no front pockets or trowel. The watering can is exactly where the clay shows it. Handcrafted stop-motion claymation look: hand-sculpted plasticine with visible fingerprints, painted metal and felt details, miniature film-set lighting. Follow the clay model exactly: same silhouette, pose, proportions, position and size in frame, and the same camera angle. Its wheels stand on the terracotta floor tiles exactly where the clay wheels touch the ground; keep the cast shadow of Image 1. Keep the background of Image 1: same place, composition, perspective and horizon, in the same handcrafted stop-motion miniature set style, crisp and detailed. Light the character with the scene: warm late-afternoon sunlight through the greenhouse glass with soft fill. No gray clay left, no extra characters or objects, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2 }, { "model": "model_openai-gpt-image-2-5-sunburst", "label": "v3 structural fix: robot 09 (1 job, references: [grounded layout, original picture]; the added lines name what this camera can and cannot see)", "parameters": { "prompt": "Image 1 is the exact final frame: a plain gray clay 3D model standing on the ground of a finished background, with its real cast shadow, seen from the exact camera angle wanted. Image 2 is the original design reference of the character. Repaint the gray clay model as Pip, the small friendly retro robot gardener, matching Image 2 for identity, design, materials and colors: round dented tin-can body painted pastel mint green with rust spots, a round porthole window on the chest with a sprouting seedling inside, big round glass eyes with brass rims, a single antenna ending in a small yellow flower, stubby jointed arms holding a small orange watering can, short legs on two rubber wheels, a canvas tool belt with a trowel. The clay model in Image 1 is the truth for which side of the character faces the camera: paint only what the clay shows from this angle, even where Image 2 shows another side. This camera sees Pip from behind and slightly to one side: the back of the head with only the edge of one brass eye rim, no face, and no porthole or seedling on the back. The watering can is exactly where the clay shows it. Handcrafted stop-motion claymation look: hand-sculpted plasticine with visible fingerprints, painted metal and felt details, miniature film-set lighting. Follow the clay model exactly: same silhouette, pose, proportions, position and size in frame, and the same camera angle. Its wheels stand on the terracotta floor tiles exactly where the clay wheels touch the ground; keep the cast shadow of Image 1. Keep the background of Image 1: same place, composition, perspective and horizon, in the same handcrafted stop-motion miniature set style, crisp and detailed. Light the character with the scene: warm late-afternoon sunlight through the greenhouse glass with soft fill. No gray clay left, no extra characters or objects, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2 }, { "model": "model_openai-gpt-image-2-5-sunburst", "label": "v3 structural fix: scooter 07 (1 job, references: [grounded layout, original picture]; the added lines name what this camera can and cannot see)", "parameters": { "prompt": "Image 1 is the exact final frame: a plain gray clay 3D model standing on the ground of a finished background, with its real cast shadow, seen from the exact camera angle wanted. Image 2 is the original design reference of the vehicle. Repaint the gray clay model as the retro-futurist delivery scooter with sidecar, matching Image 2 for identity, design, materials and colors: mint-teal and cream bodywork, chrome trim, round headlight, windscreen, tan leather saddle, bubble sidecar with crates of oranges and a rolled rug, antenna with an orange pennant, soft hover glow under the wheels. The clay model in Image 1 is the truth for which side of the vehicle faces the camera: paint only what the clay shows from this angle, even where Image 2 shows another side. This camera sees the scooter and the sidecar from BEHIND: the rear of the scooter with its round taillight, rear fender and the back of the saddle. The headlight, the front shield, the front wheel and the front of the handlebars are not visible. anime cel-shaded finish with clean ink linework and flat two-tone shading. Follow the clay model exactly: same silhouette, pose, proportions, position and size in frame, and the same camera angle. The wheels touch the stone road exactly where the clay touches it; keep the cast shadow of Image 1. Keep the background of Image 1: same place, composition, perspective and horizon, in the same hand-painted anime background style, crisp and detailed. Light the vehicle with the scene: bright summer sun matching the shadow direction, crisp cel shadows. No gray clay left, no extra characters or objects, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2 }, { "model": "model_openai-gpt-image-2-5-sunburst", "label": "v3 structural fix: scooter 08 (1 job, references: [grounded layout, original picture]; the added lines name what this camera can and cannot see)", "parameters": { "prompt": "Image 1 is the exact final frame: a plain gray clay 3D model standing on the ground of a finished background, with its real cast shadow, seen from the exact camera angle wanted. Image 2 is the original design reference of the vehicle. Repaint the gray clay model as the retro-futurist delivery scooter with sidecar, matching Image 2 for identity, design, materials and colors: mint-teal and cream bodywork, chrome trim, round headlight, windscreen, tan leather saddle, bubble sidecar with crates of oranges and a rolled rug, antenna with an orange pennant, soft hover glow under the wheels. The clay model in Image 1 is the truth for which side of the vehicle faces the camera: paint only what the clay shows from this angle, even where Image 2 shows another side. This camera sees the scooter and the sidecar from behind and slightly to the side: rear fender, taillight and the back of the saddle face the camera. The headlight and the front shield are not visible. anime cel-shaded finish with clean ink linework and flat two-tone shading. Follow the clay model exactly: same silhouette, pose, proportions, position and size in frame, and the same camera angle. The wheels touch the stone road exactly where the clay touches it; keep the cast shadow of Image 1. Keep the background of Image 1: same place, composition, perspective and horizon, in the same hand-painted anime background style, crisp and detailed. Light the vehicle with the scene: bright summer sun matching the shadow direction, crisp cel shadows. No gray clay left, no extra characters or objects, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2 }, { "model": "model_openai-gpt-image-2-5-sunburst", "label": "v3 structural fix: tortoise 07 (1 job, references: [grounded layout, original picture]; the added lines name what this camera can and cannot see)", "parameters": { "prompt": "Image 1 is the exact final frame: a plain gray clay 3D model standing on the ground of a finished background, with its real cast shadow, seen from the exact camera angle wanted. Image 2 is the original design reference of the creature. Repaint the gray clay model as the gentle giant tortoise carrying a tiny tea house on its shell, matching Image 2 for identity, design, materials and colors: wrinkled friendly face with a half-smile, ochre and olive shell plates, stubby legs, two-storey wooden tea house with a mossy curved tiled roof, round paper lantern, small balcony with potted plants, thin chimney with a curl of smoke, small wooden ladder down the shell. The clay model in Image 1 is the truth for which side of the creature faces the camera: paint only what the clay shows from this angle, even where Image 2 shows another side. This camera sees the tortoise from BEHIND: the rear of the shell and the short tail face the camera, and the head is hidden on the far side. There is no head and no face in this view. hand-painted gouache storybook illustration with visible brush texture and soft painterly edges, warm muted palette. Follow the clay model exactly: same silhouette, pose, proportions, position and size in frame, and the same camera angle. Its feet stand in the meadow grass exactly where the clay touches the ground; keep the cast shadow of Image 1. Keep the background of Image 1: same place, composition, perspective and horizon, in the same hand-painted gouache storybook style, crisp and detailed. Light the creature with the scene: soft misty dawn light, gentle warm glow from the paper lantern. No gray clay left, no extra characters or objects, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2 }, { "model": "model_openai-gpt-image-2-5-sunburst", "label": "v3 structural fix: wren 11, 12, 13 (3 jobs, references: [grounded layout, original picture]; the added lines name what this camera can and cannot see)", "parameters": { "prompt": "Image 1 is the exact final frame: a plain gray clay 3D model standing on the ground of a finished background, with its real cast shadow, seen from the exact camera angle wanted. Image 2 is the original design reference of the character. Repaint the gray clay model as Wren, the cheerful young lantern courier girl, matching Image 2 for identity, design, materials and colors: copper-red messy bob with a small braid, freckles, big hazel eyes, mustard-yellow hooded raincoat with toggle buttons, chunky teal knitted scarf, rolled navy trousers, bright red rubber boots, worn brown leather satchel stuffed with letters, small glowing brass lantern in her hand. The clay model in Image 1 is the truth for which side of the character faces the camera: paint only what the clay shows from this angle, even where Image 2 shows another side. The rounded shape on her back in the clay is the hood of her raincoat hanging down, not a bag. She carries exactly one satchel, on her left hip with the strap across her chest, and nothing on her back. premium 3D animated feature film finish: soft subsurface skin, stylized proportions, rich materials. Follow the clay model exactly: same silhouette, pose, proportions, position and size in frame, and the same camera angle. Her boots stand on the wet cobblestones exactly where the clay feet touch the ground; keep the cast shadow of Image 1. Keep the background of Image 1: same place, composition, perspective and horizon, in the same premium 3D animated feature film rendering, crisp and detailed. Light the character with the scene: blue-hour dusk after rain: cool ambient light, warm glow from the lamps and from her lantern, soft reflections on the wet cobblestones. No gray clay left, no extra characters or objects, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2 }, { "model": "model_openai-gpt-image-2-5-sunburst", "label": "v3 structural fix: dwarf 11, 12, low_12 (3 jobs, references: [grounded layout, original picture]; the added lines name what this camera can and cannot see)", "parameters": { "prompt": "Image 1 is the exact final frame: a plain gray clay 3D model standing on the ground of a finished background, with its real cast shadow, seen from the exact camera angle wanted. Image 2 is the original design reference of the character. Repaint the gray clay model as Brakka, the stout dwarf blacksmith warrior, matching Image 2 for identity, design, materials and colors: long braided copper-red beard with brass rings, riveted steel helmet with short horns, thick leather apron over chainmail, brass pauldrons with knotwork, heavy two-handed forge hammer with brass bands, round wooden shield with an iron rim on his back, belt with tongs, pouches and tools, big armored boots. The clay model in Image 1 is the truth for which side of the character faces the camera: paint only what the clay shows from this angle, even where Image 2 shows another side. He holds ONE hammer: its heavy head is hidden behind the round shield, and the part sticking out on the right is only the end of the long wooden handle with a small brass cap, not a second hammer head. Seen from behind, the round wooden shield with its iron rim covers his back; the hammer head is dark hammered steel with brass bands. Stylized fantasy game character render: chunky proportions, rich worn leather, hammered steel and brass, crisp readable shapes, warm saturated palette. Follow the clay model exactly: same silhouette, pose, proportions, position and size in frame, and the same camera angle. His boots stand on the flagstones exactly where the clay feet touch the ground; keep the cast shadow of Image 1. Keep the background of Image 1: same place, composition, perspective and horizon, in the same hand-painted stylized fantasy game environment style, crisp and detailed. Light the character with the scene: warm orange firelight from the forges with a cool blue fill from above. No gray clay left, no extra characters or objects, no text.", "width": 1024, "height": 1024, "quality": "high", "background": "opaque" }, "referenceImageCount": 2 }, { "model": "model_birefnet-background-removal", "label": "v3 structural fix: No background cutouts of the 24 fixed repaints", "parameters": { "image": "" }, "referenceImageCount": 0 } ]