Wide angle

GPT Image 2.5 Flare

Use the uploaded image as the sole subject reference. Recreate the same object, product, character, creature, sculpture, logo or lettering with its recognizable identity, underlying structure, silhouette, colors, materials, surface texture, branding and every defining detail faithfully preserved. Change only the camera perspective. Photograph the subject from an extremely close dynamic three-quarter viewpoint using an 8–10mm ultra-wide close-focus lens, positioning the camera only a few centimeters from one naturally prominent part of the subject. Let this closest area dominate the foreground at a dramatically exaggerated scale while the remaining form rapidly tapers and recedes into depth, creating intense forced perspective, powerful foreshortening and a bold sculptural silhouette. Select the foreground anchor according to the subject’s own geometry and pose so the result remains instantly recognizable and compositionally strong. Keep the distortion optical and perspective-driven while maintaining coherent anatomy, construction and connections between every part. Fill most of the frame with one isolated subject and build the composition along a pronounced near-to-far depth axis. For a flat logo, graphic or lettering, create a clean volumetric version while preserving its exact contours, typography and colors. Place the subject on a pure seamless white background with high-key diffused studio lighting, soft frontal fill, bright backlight, luminous rim glow, subtle bloom and airy overexposed highlights. Render the closest material details with tactile realism and create a gentle depth-of-field falloff toward the distant parts. Contemporary premium editorial campaign, clean art direction, playful monumental scale, polished photorealism, high resolution. Deliver a clean final image containing only the transformed subject and the white background.

copied

Seedream 5.0 Pro

Use the uploaded image as the visual source and preserve the object’s exact recognizable shape, construction, proportions, colors, materials, textures, patterns, logos, inscriptions and placement of every detail. Rebuild the composition around extreme object-specific foreshortening. Analyze the silhouette and identify the two most widely separated structural points, the principal three-dimensional axis and the broadest, heaviest or most distinctive endpoint. Direct this individual axis almost straight toward the camera, placing the selected endpoint only 5–15 cm from the lens while the opposite endpoint recedes deeply away from it. The nearest section occupies 45–60% of the composition and appears three to five times larger than the farthest section. For a compact object, construct the depth axis between its most distinctive asymmetrical feature and the opposite contour point. The frontal surface turns into a sharply receding plane while every identifying feature remains readable. Use a 10 mm rectilinear ultra-wide perspective with clean rectangular edge-to-edge coverage, uninterrupted corners and the lens completely outside the photograph. Prioritize severe near-to-far scale change, elongated depth and dramatic perspective shortening over a balanced frontal product view. Levitate the object freely in the original environment with visible empty space beneath and around it, zero support, zero surface contact and zero cast or contact shadow. Keep the entire object and every attached component inside the frame with a narrow margin of breathing room. Preserve the original background, lighting, colors and atmosphere, transformed consistently by the new camera position. The chosen foreground endpoint, depth direction, rotation and camera height must arise from the unique geometry of each uploaded object, creating a different dramatic viewpoint for every input.

copied

Nano Banana Pro

Use the uploaded image as the visual source and preserve the object’s exact recognizable shape, construction, proportions, colors, materials, textures, patterns, logos, inscriptions and placement of every detail. Analyze this specific object before composing the shot: identify its strongest asymmetrical feature, dominant spatial axis, direction of visual weight, most expressive contour and the angle from which its construction looks most distinctive. Build the camera position from these individual characteristics so every new object receives its own viewpoint. A protruding feature creates an axial view along its direction, a flat construction creates a grazing lateral view, a vertically weighted form creates an upper or lower oblique view, and a compact symmetrical form rotates toward its most distinctive side while its identifying front details remain readable. Position the object across several depth planes, with one structurally meaningful area close to the camera and the opposite area receding sharply, creating an obvious difference in scale through real near-to-far perspective. Use a 10–12 mm rectilinear ultra-wide lens with full-sensor edge-to-edge coverage, a clean rectangular photographic frame and uninterrupted background extending naturally into all four corners. The lens, glass, housing, circular border and dark rim remain completely outside the image. Levitate the object freely in the original environment with a clear air gap around and beneath it, zero surface contact, zero support and zero cast or contact shadow below. Give the object an input-specific spatial rotation around its natural construction axis rather than a standard centered product orientation. Keep the complete object and every attached element fully inside the frame with breathing room around the silhouette. Preserve the original background, lighting, colors and atmosphere, transformed consistently by the new camera position. The result feels sculptural, dynamic and editorial, with strong wide-angle perspective created by proximity and depth while the object remains completely recognizable.

copied

GPT Image 2.5 Flare

Use the uploaded image as the sole subject reference. Recreate the same object, product, character, creature, sculpture, logo or lettering with its recognizable identity, underlying structure, silhouette, colors, materials, surface texture, branding and every defining detail faithfully preserved. Change only the camera perspective. Photograph the subject from an extremely close dynamic three-quarter viewpoint using an 8–10mm ultra-wide close-focus lens, positioning the camera only a few centimeters from one naturally prominent part of the subject. Let this closest area dominate the foreground at a dramatically exaggerated scale while the remaining form rapidly tapers and recedes into depth, creating intense forced perspective, powerful foreshortening and a bold sculptural silhouette. Select the foreground anchor according to the subject’s own geometry and pose so the result remains instantly recognizable and compositionally strong. Keep the distortion optical and perspective-driven while maintaining coherent anatomy, construction and connections between every part. Fill most of the frame with one isolated subject and build the composition along a pronounced near-to-far depth axis. For a flat logo, graphic or lettering, create a clean volumetric version while preserving its exact contours, typography and colors. Place the subject on a pure seamless white background with high-key diffused studio lighting, soft frontal fill, bright backlight, luminous rim glow, subtle bloom and airy overexposed highlights. Render the closest material details with tactile realism and create a gentle depth-of-field falloff toward the distant parts. Contemporary premium editorial campaign, clean art direction, playful monumental scale, polished photorealism, high resolution. Deliver a clean final image containing only the transformed subject and the white background.

copied

Seedream 5 Pro

Use the uploaded image as the authoritative full-scene reference for the object and its environment. Preserve the exact number of objects and each object’s recognizable silhouette, rigid construction, proportions, number and placement of parts, colors, materials, textures, patterns, finish, logos, inscriptions and every identifying detail. Preserve the same recognizable background, location, surrounding objects, architecture or landscape, lighting, colors and atmosphere. Analyze the silhouette, principal three-dimensional axis, two most widely separated structural points and the broadest, heaviest or most distinctive volume, then choose a unique camera side, height, tilt and rotation derived from this specific construction. Treat the object as an unchanged rigid volume and create all exaggeration through camera position and perspective. For an elongated object, point its length almost directly toward the camera; for a flat or compact object, turn its main face 55–75 degrees away from the image plane and direct its strongest edge or corner toward the camera; for an asymmetrical object, align its most distinctive volume with the opposite contour point. Position an extreme rectilinear ultra-wide close-focus virtual camera closer to the foreground surface than the object’s own front-to-back depth. The nearest section occupies 50–65% of the composition and appears four to six times larger than the farthest section. Create a continuous scale cascade across the whole object through dramatically expanded near geometry, rapid reduction into depth, severe foreshortening, peripheral stretching and strongly converging structural lines. Build this perspective transformation first, then extend the surrounding canvas until the complete distorted silhouette fits inside the image with 8–12% breathing room. Levitate the object freely within the original environment, with visible open air beneath and around it, complete separation from surrounding surfaces and shadowless suspension. Reproject the same identifiable background consistently from the new viewpoint. The image fills the complete rectangular canvas edge to edge with natural background imagery in all four corners; optical hardware, circular image boundaries, black rims and vignetting are entirely absent.

copied

GPT Image 2.5 Flare

Edit the uploaded image itself rather than generating a new scene. Use it as the sole source for the object and environment. Preserve the exact number of objects and each object’s recognizable silhouette, rigid construction, proportions, number and placement of parts, colors, materials, textures, patterns, finish, branding, inscriptions and small identifying details. Change only the spatial projection, object orientation and viewpoint. Analyze the object’s geometry, dominant three-dimensional axis, two most widely separated structural points and strongest asymmetrical feature, then select a viewing direction unique to that construction. Direct the length of an elongated form deeply toward the viewpoint; turn the main surface of a flat or compact form 60–75 degrees across the depth axis and bring its strongest edge or corner forward; align a projecting functional part with the opposite contour of an irregular form. Position the virtual viewpoint closer to the nearest surface than the object’s own front-to-back depth. The nearest structural zone occupies 50–65% of the composition and appears four to six times larger than the farthest zone. Create a continuous perspective gradient through the complete object, with expanded near geometry, rapid scale reduction, severe foreshortening, stretched peripheral space and strongly converging structural lines. Form the extreme perspective first, then expand the surrounding composition until the entire distorted silhouette and every attached component fit inside the image with 8–12% breathing room. Suspend the object freely in the same recognizable environment with visible open air beneath and around it, complete separation from surrounding surfaces and clean shadowless levitation. Retain the source background elements, location, colors, lighting and atmosphere, applying the same spatial projection coherently across the scene. Present the result as a standard rectangular edge-to-edge image with continuous environment reaching all four corners, intact construction, readable branding and detailed photorealism.

copied

GPT Image 2.5 Flare

Use the uploaded image as the visual source and preserve the object’s exact recognizable shape, construction, proportions, colors, materials, textures, patterns, logos, inscriptions and placement of every detail. Rebuild the composition around extreme object-specific foreshortening. Analyze the silhouette and identify the two most widely separated structural points, the principal three-dimensional axis and the broadest, heaviest or most distinctive endpoint. Direct this individual axis almost straight toward the camera, placing the selected endpoint only 5–15 cm from the lens while the opposite endpoint recedes deeply away from it. The nearest section occupies 45–60% of the composition and appears three to five times larger than the farthest section. For a compact object, construct the depth axis between its most distinctive asymmetrical feature and the opposite contour point. The frontal surface turns into a sharply receding plane while every identifying feature remains readable. Use a 10 mm rectilinear ultra-wide perspective with clean rectangular edge-to-edge coverage, uninterrupted corners and the lens completely outside the photograph. Prioritize severe near-to-far scale change, elongated depth and dramatic perspective shortening over a balanced frontal product view. Levitate the object freely in the original environment with visible empty space beneath and around it, zero support, zero surface contact and zero cast or contact shadow. Keep the entire object and every attached component inside the frame with a narrow margin of breathing room. Preserve the original background, lighting, colors and atmosphere, transformed consistently by the new camera position. The chosen foreground endpoint, depth direction, rotation and camera height must arise from the unique geometry of each uploaded object, creating a different dramatic viewpoint for every input.

copied

Nano Banana Pro

Use the uploaded image as the fixed visual source. Preserve the object with very high fidelity: the same recognizable silhouette, proportions, construction, number and placement of parts, colors, materials, surface textures, patterns, labels, logos and small identifying details. Do not redesign, simplify, melt or invent parts of the object.Before generating, silently analyze the object’s specific geometry. Identify its dominant axis, most recognizable feature, strongest surface detail and most expressive side. Based on that analysis, select a unique camera position that reveals this particular object most dramatically. The viewpoint must be determined by the object’s individual shape, not by a reusable standard composition.Avoid the default centered front view and ordinary three-quarter product angle. Depending on the object’s geometry, choose an extreme axial view along its longest dimension, a low underside diagonal, a high oblique orbit, an edge-leading perspective, a tilted lateral sweep or another unexpected viewpoint that best exaggerates its distinctive construction. Do not apply the same camera angle to every object.Photograph the object from very close range using an extreme 6–10 mm ultra-wide lens. The wide-angle effect must be unmistakable: the nearest defining part becomes optically enlarged, the rest of the object recedes dramatically into depth, parallel spatial lines converge, and the background stretches through perspective. Use strong but coherent optical distortion and controlled barrel distortion, without turning the object into an unrecognizable shape.The object must levitate freely in space with a subtle dynamic rotation appropriate to its form. It must not stand, sit or rest on any surface. No pedestal, holder, visible support, floor contact, contact shadow or cast shadow beneath it. Environmental light and reflections may remain on its materials, but there must be no visual indication that the object is supported by the ground.The complete object and every attached component must fit entirely inside the image. Nothing may be cropped or touch the frame edges. Leave generous breathing room around the full silhouette while maintaining strong close-range perspective. The composition should be asymmetrical, spatial and energetic rather than a centered catalog packshot.Retain the original location and background from the uploaded image: the same recognizable environment, background elements, colors, textures, lighting and atmosphere. Reconstruct it consistently from the selected ultra-wide viewpoint. Do not replace it with a white studio or an unrelated background unless the original image already contains one.The final image should feel like an experimental high-fashion object photograph: sculptural, weightless, visually tense and strongly distorted through real ultra-wide perspective, while the original object remains completely recognizable.Hard exclusions: floor contact, pedestal, support, contact shadow, cast shadow beneath the object, cropped edges, missing parts, duplicated parts, changed design, melted geometry, centered catalog composition, ordinary eye-level three-quarter view, weak perspective, identical camera treatment for different objects, white replacement background.

copied

Seedream 5.0 Pro

Use the uploaded image as the authoritative full-scene reference for the object and its environment. Preserve the exact number of objects and each object’s recognizable silhouette, rigid construction, proportions, number and placement of parts, colors, materials, textures, patterns, finish, logos, inscriptions and every identifying detail. Preserve the same recognizable background, location, surrounding objects, architecture or landscape, lighting, colors and atmosphere. Analyze the silhouette, principal three-dimensional axis, two most widely separated structural points and the broadest, heaviest or most distinctive volume, then choose a unique camera side, height, tilt and rotation derived from this specific construction. Treat the object as an unchanged rigid volume and create all exaggeration through camera position and perspective. For an elongated object, point its length almost directly toward the camera; for a flat or compact object, turn its main face 55–75 degrees away from the image plane and direct its strongest edge or corner toward the camera; for an asymmetrical object, align its most distinctive volume with the opposite contour point. Position an extreme rectilinear ultra-wide close-focus virtual camera closer to the foreground surface than the object’s own front-to-back depth. The nearest section occupies 50–65% of the composition and appears four to six times larger than the farthest section. Create a continuous scale cascade across the whole object through dramatically expanded near geometry, rapid reduction into depth, severe foreshortening, peripheral stretching and strongly converging structural lines. Build this perspective transformation first, then extend the surrounding canvas until the complete distorted silhouette fits inside the image with 8–12% breathing room. Levitate the object freely within the original environment, with visible open air beneath and around it, complete separation from surrounding surfaces and shadowless suspension. Reproject the same identifiable background consistently from the new viewpoint. The image fills the complete rectangular canvas edge to edge with natural background imagery in all four corners; optical hardware, circular image boundaries, black rims and vignetting are entirely absent.

copied

GPT Image 2.5 Flare

Use the uploaded image as the visual source and preserve the object’s exact recognizable shape, construction, proportions, colors, materials, textures, patterns, logos, inscriptions and placement of every detail. Rebuild the composition around extreme object-specific foreshortening. Analyze the silhouette and identify the two most widely separated structural points, the principal three-dimensional axis and the broadest, heaviest or most distinctive endpoint. Direct this individual axis almost straight toward the camera, placing the selected endpoint only 5–15 cm from the lens while the opposite endpoint recedes deeply away from it. The nearest section occupies 45–60% of the composition and appears three to five times larger than the farthest section. For a compact object, construct the depth axis between its most distinctive asymmetrical feature and the opposite contour point. The frontal surface turns into a sharply receding plane while every identifying feature remains readable. Use a 10 mm rectilinear ultra-wide perspective with clean rectangular edge-to-edge coverage, uninterrupted corners and the lens completely outside the photograph. Prioritize severe near-to-far scale change, elongated depth and dramatic perspective shortening over a balanced frontal product view. Levitate the object freely in the original environment with visible empty space beneath and around it, zero support, zero surface contact and zero cast or contact shadow. Keep the entire object and every attached component inside the frame with a narrow margin of breathing room. Preserve the original background, lighting, colors and atmosphere, transformed consistently by the new camera position. The chosen foreground endpoint, depth direction, rotation and camera height must arise from the unique geometry of each uploaded object, creating a different dramatic viewpoint for every input.

copied

GPT Image 2.5 Flare

Use the uploaded image as the fixed visual source. Preserve the object with very high fidelity: the same recognizable silhouette, proportions, construction, number and placement of parts, colors, materials, surface textures, patterns, labels, logos and small identifying details. Do not redesign, simplify, melt or invent parts of the object.Before generating, silently analyze the object’s specific geometry. Identify its dominant axis, most recognizable feature, strongest surface detail and most expressive side. Based on that analysis, select a unique camera position that reveals this particular object most dramatically. The viewpoint must be determined by the object’s individual shape, not by a reusable standard composition.Avoid the default centered front view and ordinary three-quarter product angle. Depending on the object’s geometry, choose an extreme axial view along its longest dimension, a low underside diagonal, a high oblique orbit, an edge-leading perspective, a tilted lateral sweep or another unexpected viewpoint that best exaggerates its distinctive construction. Do not apply the same camera angle to every object.Photograph the object from very close range using an extreme 6–10 mm ultra-wide lens. The wide-angle effect must be unmistakable: the nearest defining part becomes optically enlarged, the rest of the object recedes dramatically into depth, parallel spatial lines converge, and the background stretches through perspective. Use strong but coherent optical distortion and controlled barrel distortion, without turning the object into an unrecognizable shape.The object must levitate freely in space with a subtle dynamic rotation appropriate to its form. It must not stand, sit or rest on any surface. No pedestal, holder, visible support, floor contact, contact shadow or cast shadow beneath it. Environmental light and reflections may remain on its materials, but there must be no visual indication that the object is supported by the ground.The complete object and every attached component must fit entirely inside the image. Nothing may be cropped or touch the frame edges. Leave generous breathing room around the full silhouette while maintaining strong close-range perspective. The composition should be asymmetrical, spatial and energetic rather than a centered catalog packshot.Retain the original location and background from the uploaded image: the same recognizable environment, background elements, colors, textures, lighting and atmosphere. Reconstruct it consistently from the selected ultra-wide viewpoint. Do not replace it with a white studio or an unrelated background unless the original image already contains one.The final image should feel like an experimental high-fashion object photograph: sculptural, weightless, visually tense and strongly distorted through real ultra-wide perspective, while the original object remains completely recognizable.Hard exclusions: floor contact, pedestal, support, contact shadow, cast shadow beneath the object, cropped edges, missing parts, duplicated parts, changed design, melted geometry, centered catalog composition, ordinary eye-level three-quarter view, weak perspective, identical camera treatment for different objects, white replacement background.

copied

Nano Banana Pro

Use the uploaded image as the authoritative full-scene reference for the object and its environment. Preserve the exact number of objects and each object’s recognizable silhouette, rigid construction, proportions, number and placement of parts, colors, materials, textures, patterns, finish, logos, inscriptions and every identifying detail. Preserve the same recognizable background, location, surrounding objects, architecture or landscape, lighting, colors and atmosphere. Analyze the silhouette, principal three-dimensional axis, two most widely separated structural points and the broadest, heaviest or most distinctive volume, then choose a unique camera side, height, tilt and rotation derived from this specific construction. Treat the object as an unchanged rigid volume and create all exaggeration through camera position and perspective. For an elongated object, point its length almost directly toward the camera; for a flat or compact object, turn its main face 55–75 degrees away from the image plane and direct its strongest edge or corner toward the camera; for an asymmetrical object, align its most distinctive volume with the opposite contour point. Position an extreme rectilinear ultra-wide close-focus virtual camera closer to the foreground surface than the object’s own front-to-back depth. The nearest section occupies 50–65% of the composition and appears four to six times larger than the farthest section. Create a continuous scale cascade across the whole object through dramatically expanded near geometry, rapid reduction into depth, severe foreshortening, peripheral stretching and strongly converging structural lines. Build this perspective transformation first, then extend the surrounding canvas until the complete distorted silhouette fits inside the image with 8–12% breathing room. Levitate the object freely within the original environment, with visible open air beneath and around it, complete separation from surrounding surfaces and shadowless suspension. Reproject the same identifiable background consistently from the new viewpoint. The image fills the complete rectangular canvas edge to edge with natural background imagery in all four corners; optical hardware, circular image boundaries, black rims and vignetting are entirely absent.

copied

Nano Banana Pro

Use the uploaded image as the sole subject reference. Recreate the same object, product, character, creature, sculpture, logo or lettering with its recognizable identity, underlying structure, silhouette, colors, materials, surface texture, branding and every defining detail faithfully preserved. Change only the camera perspective. Photograph the subject from an extremely close dynamic three-quarter viewpoint using an 8–10mm ultra-wide close-focus lens, positioning the camera only a few centimeters from one naturally prominent part of the subject. Let this closest area dominate the foreground at a dramatically exaggerated scale while the remaining form rapidly tapers and recedes into depth, creating intense forced perspective, powerful foreshortening and a bold sculptural silhouette. Select the foreground anchor according to the subject’s own geometry and pose so the result remains instantly recognizable and compositionally strong. Keep the distortion optical and perspective-driven while maintaining coherent anatomy, construction and connections between every part. Fill most of the frame with one isolated subject and build the composition along a pronounced near-to-far depth axis. For a flat logo, graphic or lettering, create a clean volumetric version while preserving its exact contours, typography and colors. Place the subject on a pure seamless white background with high-key diffused studio lighting, soft frontal fill, bright backlight, luminous rim glow, subtle bloom and airy overexposed highlights. Render the closest material details with tactile realism and create a gentle depth-of-field falloff toward the distant parts. Contemporary premium editorial campaign, clean art direction, playful monumental scale, polished photorealism, high resolution. Deliver a clean final image containing only the transformed subject and the white background.

copied

Nano Banana Pro

Edit the uploaded image itself rather than generating a new scene. Use it as the sole source for the object and environment. Preserve the exact number of objects and each object’s recognizable silhouette, rigid construction, proportions, number and placement of parts, colors, materials, textures, patterns, finish, branding, inscriptions and small identifying details. Change only the spatial projection, object orientation and viewpoint. Analyze the object’s geometry, dominant three-dimensional axis, two most widely separated structural points and strongest asymmetrical feature, then select a viewing direction unique to that construction. Direct the length of an elongated form deeply toward the viewpoint; turn the main surface of a flat or compact form 60–75 degrees across the depth axis and bring its strongest edge or corner forward; align a projecting functional part with the opposite contour of an irregular form. Position the virtual viewpoint closer to the nearest surface than the object’s own front-to-back depth. The nearest structural zone occupies 50–65% of the composition and appears four to six times larger than the farthest zone. Create a continuous perspective gradient through the complete object, with expanded near geometry, rapid scale reduction, severe foreshortening, stretched peripheral space and strongly converging structural lines. Form the extreme perspective first, then expand the surrounding composition until the entire distorted silhouette and every attached component fit inside the image with 8–12% breathing room. Suspend the object freely in the same recognizable environment with visible open air beneath and around it, complete separation from surrounding surfaces and clean shadowless levitation. Retain the source background elements, location, colors, lighting and atmosphere, applying the same spatial projection coherently across the scene. Present the result as a standard rectangular edge-to-edge image with continuous environment reaching all four corners, intact construction, readable branding and detailed photorealism.

copied

GPT Image 2

Use the uploaded image as the sole subject reference. Recreate the same object, product, character, creature, sculpture, logo or lettering with its recognizable identity, underlying structure, silhouette, colors, materials, surface texture, branding and every defining detail faithfully preserved. Change only the camera perspective. Photograph the subject from an extremely close dynamic three-quarter viewpoint using an 8–10mm ultra-wide close-focus lens, positioning the camera only a few centimeters from one naturally prominent part of the subject. Let this closest area dominate the foreground at a dramatically exaggerated scale while the remaining form rapidly tapers and recedes into depth, creating intense forced perspective, powerful foreshortening and a bold sculptural silhouette. Select the foreground anchor according to the subject’s own geometry and pose so the result remains instantly recognizable and compositionally strong. Keep the distortion optical and perspective-driven while maintaining coherent anatomy, construction and connections between every part. Fill most of the frame with one isolated subject and build the composition along a pronounced near-to-far depth axis. For a flat logo, graphic or lettering, create a clean volumetric version while preserving its exact contours, typography and colors. Place the subject on a pure seamless white background with high-key diffused studio lighting, soft frontal fill, bright backlight, luminous rim glow, subtle bloom and airy overexposed highlights. Render the closest material details with tactile realism and create a gentle depth-of-field falloff toward the distant parts. Contemporary premium editorial campaign, clean art direction, playful monumental scale, polished photorealism, high resolution. Deliver a clean final image containing only the transformed subject and the white background.

copied

GPT Image 2.5 Flare

Use the uploaded image as the fixed visual source. Preserve the object with very high fidelity: the same recognizable silhouette, proportions, construction, number and placement of parts, colors, materials, surface textures, patterns, labels, logos and small identifying details. Do not redesign, simplify, melt or invent parts of the object.Before generating, silently analyze the object’s specific geometry. Identify its dominant axis, most recognizable feature, strongest surface detail and most expressive side. Based on that analysis, select a unique camera position that reveals this particular object most dramatically. The viewpoint must be determined by the object’s individual shape, not by a reusable standard composition.Avoid the default centered front view and ordinary three-quarter product angle. Depending on the object’s geometry, choose an extreme axial view along its longest dimension, a low underside diagonal, a high oblique orbit, an edge-leading perspective, a tilted lateral sweep or another unexpected viewpoint that best exaggerates its distinctive construction. Do not apply the same camera angle to every object.Photograph the object from very close range using an extreme 6–10 mm ultra-wide lens. The wide-angle effect must be unmistakable: the nearest defining part becomes optically enlarged, the rest of the object recedes dramatically into depth, parallel spatial lines converge, and the background stretches through perspective. Use strong but coherent optical distortion and controlled barrel distortion, without turning the object into an unrecognizable shape.The object must levitate freely in space with a subtle dynamic rotation appropriate to its form. It must not stand, sit or rest on any surface. No pedestal, holder, visible support, floor contact, contact shadow or cast shadow beneath it. Environmental light and reflections may remain on its materials, but there must be no visual indication that the object is supported by the ground.The complete object and every attached component must fit entirely inside the image. Nothing may be cropped or touch the frame edges. Leave generous breathing room around the full silhouette while maintaining strong close-range perspective. The composition should be asymmetrical, spatial and energetic rather than a centered catalog packshot.Retain the original location and background from the uploaded image: the same recognizable environment, background elements, colors, textures, lighting and atmosphere. Reconstruct it consistently from the selected ultra-wide viewpoint. Do not replace it with a white studio or an unrelated background unless the original image already contains one.The final image should feel like an experimental high-fashion object photograph: sculptural, weightless, visually tense and strongly distorted through real ultra-wide perspective, while the original object remains completely recognizable.Hard exclusions: floor contact, pedestal, support, contact shadow, cast shadow beneath the object, cropped edges, missing parts, duplicated parts, changed design, melted geometry, centered catalog composition, ordinary eye-level three-quarter view, weak perspective, identical camera treatment for different objects, white replacement background.

copied

Seedream 5.0 Pro

Use the uploaded image as the authoritative full-scene reference for the object and its environment. Preserve the exact number of objects and each object’s recognizable silhouette, rigid construction, proportions, number and placement of parts, colors, materials, textures, patterns, finish, logos, inscriptions and every identifying detail. Preserve the same recognizable background, location, surrounding objects, architecture or landscape, lighting, colors and atmosphere. Analyze the silhouette, principal three-dimensional axis, two most widely separated structural points and the broadest, heaviest or most distinctive volume, then choose a unique camera side, height, tilt and rotation derived from this specific construction. Treat the object as an unchanged rigid volume and create all exaggeration through camera position and perspective. For an elongated object, point its length almost directly toward the camera; for a flat or compact object, turn its main face 55–75 degrees away from the image plane and direct its strongest edge or corner toward the camera; for an asymmetrical object, align its most distinctive volume with the opposite contour point. Position an extreme rectilinear ultra-wide close-focus virtual camera closer to the foreground surface than the object’s own front-to-back depth. The nearest section occupies 50–65% of the composition and appears four to six times larger than the farthest section. Create a continuous scale cascade across the whole object through dramatically expanded near geometry, rapid reduction into depth, severe foreshortening, peripheral stretching and strongly converging structural lines. Build this perspective transformation first, then extend the surrounding canvas until the complete distorted silhouette fits inside the image with 8–12% breathing room. Levitate the object freely within the original environment, with visible open air beneath and around it, complete separation from surrounding surfaces and shadowless suspension. Reproject the same identifiable background consistently from the new viewpoint. The image fills the complete rectangular canvas edge to edge with natural background imagery in all four corners; optical hardware, circular image boundaries, black rims and vignetting are entirely absent.

copied

Nano Banana Pro

Edit the uploaded image itself rather than generating a new scene. Use it as the sole source for the object and environment. Preserve the exact number of objects and each object’s recognizable silhouette, rigid construction, proportions, number and placement of parts, colors, materials, textures, patterns, finish, branding, inscriptions and small identifying details. Change only the spatial projection, object orientation and viewpoint. Analyze the object’s geometry, dominant three-dimensional axis, two most widely separated structural points and strongest asymmetrical feature, then select a viewing direction unique to that construction. Direct the length of an elongated form deeply toward the viewpoint; turn the main surface of a flat or compact form 60–75 degrees across the depth axis and bring its strongest edge or corner forward; align a projecting functional part with the opposite contour of an irregular form. Position the virtual viewpoint closer to the nearest surface than the object’s own front-to-back depth. The nearest structural zone occupies 50–65% of the composition and appears four to six times larger than the farthest zone. Create a continuous perspective gradient through the complete object, with expanded near geometry, rapid scale reduction, severe foreshortening, stretched peripheral space and strongly converging structural lines. Form the extreme perspective first, then expand the surrounding composition until the entire distorted silhouette and every attached component fit inside the image with 8–12% breathing room. Suspend the object freely in the same recognizable environment with visible open air beneath and around it, complete separation from surrounding surfaces and clean shadowless levitation. Retain the source background elements, location, colors, lighting and atmosphere, applying the same spatial projection coherently across the scene. Present the result as a standard rectangular edge-to-edge image with continuous environment reaching all four corners, intact construction, readable branding and detailed photorealism.

copied

GPT Image 2.5 Flare

Use the uploaded image as the authoritative reference for both the object and the complete environment. Preserve the object one-to-one: exact silhouette, proportions, construction, number and placement of parts, materials, textures, surface finish, colors, transparency, reflections, branding, graphics, labels and correctly spelled text. Preserve the exact number of objects and keep grouped elements visually connected. Analyze the object’s geometry and identify the viewing axis that creates the strongest silhouette and greatest near-to-far depth. Move the camera around the object and orient the complete unchanged form so one distinctive structural feature points toward the lens while the opposite side recedes deeply into the scene. Photograph from point-blank range with an 8mm rectilinear ultra-wide close-focus lens. The closest part appears three to four times larger than the farthest part, creating an expanded foreground, rapid scale reduction into depth, powerful foreshortening and stretched peripheral geometry. The transformation reads entirely as optical perspective while the original design remains intact. Keep the complete silhouette inside the frame with every edge, tip, handle, component and letter fully visible and a 5–8% margin of the original environment around it. Preserve the object’s original physical relationship with the scene, including its support, placement, contact, suspension or interaction with other elements. Retain the original background as the same identifiable place with its surfaces, architecture, furniture, landscape, colors, textures and lighting. Reconstruct the environment consistently from the new camera position so its perspective responds naturally to the ultra-wide lens. Match the original shadows, reflections and color temperature. The final image contains the exact original object or object group within the same recognizable environment, captured as a bold high-end ultra-wide editorial photograph.

copied

GPT Image 2.5 Flare

Use the uploaded image as the visual source and preserve the object’s exact recognizable shape, construction, proportions, colors, materials, textures, patterns, logos, inscriptions and placement of every detail. Analyze this specific object before composing the shot: identify its strongest asymmetrical feature, dominant spatial axis, direction of visual weight, most expressive contour and the angle from which its construction looks most distinctive. Build the camera position from these individual characteristics so every new object receives its own viewpoint. A protruding feature creates an axial view along its direction, a flat construction creates a grazing lateral view, a vertically weighted form creates an upper or lower oblique view, and a compact symmetrical form rotates toward its most distinctive side while its identifying front details remain readable. Position the object across several depth planes, with one structurally meaningful area close to the camera and the opposite area receding sharply, creating an obvious difference in scale through real near-to-far perspective. Use a 10–12 mm rectilinear ultra-wide lens with full-sensor edge-to-edge coverage, a clean rectangular photographic frame and uninterrupted background extending naturally into all four corners. The lens, glass, housing, circular border and dark rim remain completely outside the image. Levitate the object freely in the original environment with a clear air gap around and beneath it, zero surface contact, zero support and zero cast or contact shadow below. Give the object an input-specific spatial rotation around its natural construction axis rather than a standard centered product orientation. Keep the complete object and every attached element fully inside the frame with breathing room around the silhouette. Preserve the original background, lighting, colors and atmosphere, transformed consistently by the new camera position. The result feels sculptural, dynamic and editorial, with strong wide-angle perspective created by proximity and depth while the object remains completely recognizable.

copied

Nano Banana Pro

Use the uploaded image as the authoritative full-scene reference for the object and its environment. Preserve the exact number of objects and each object’s recognizable silhouette, rigid construction, proportions, number and placement of parts, colors, materials, textures, patterns, finish, logos, inscriptions and every identifying detail. Preserve the same recognizable background, location, surrounding objects, architecture or landscape, lighting, colors and atmosphere. Analyze the silhouette, principal three-dimensional axis, two most widely separated structural points and the broadest, heaviest or most distinctive volume, then choose a unique camera side, height, tilt and rotation derived from this specific construction. Treat the object as an unchanged rigid volume and create all exaggeration through camera position and perspective. For an elongated object, point its length almost directly toward the camera; for a flat or compact object, turn its main face 55–75 degrees away from the image plane and direct its strongest edge or corner toward the camera; for an asymmetrical object, align its most distinctive volume with the opposite contour point. Position an extreme rectilinear ultra-wide close-focus virtual camera closer to the foreground surface than the object’s own front-to-back depth. The nearest section occupies 50–65% of the composition and appears four to six times larger than the farthest section. Create a continuous scale cascade across the whole object through dramatically expanded near geometry, rapid reduction into depth, severe foreshortening, peripheral stretching and strongly converging structural lines. Build this perspective transformation first, then extend the surrounding canvas until the complete distorted silhouette fits inside the image with 8–12% breathing room. Levitate the object freely within the original environment, with visible open air beneath and around it, complete separation from surrounding surfaces and shadowless suspension. Reproject the same identifiable background consistently from the new viewpoint. The image fills the complete rectangular canvas edge to edge with natural background imagery in all four corners; optical hardware, circular image boundaries, black rims and vignetting are entirely absent.

copied

GPT Image 2.5 Flare

Use the uploaded image as the authoritative reference for both the object and the complete environment. Preserve the object one-to-one: exact silhouette, proportions, construction, number and placement of parts, materials, textures, surface finish, colors, transparency, reflections, branding, graphics, labels and correctly spelled text. Preserve the exact number of objects and keep grouped elements visually connected. Analyze the object’s geometry and identify the viewing axis that creates the strongest silhouette and greatest near-to-far depth. Move the camera around the object and orient the complete unchanged form so one distinctive structural feature points toward the lens while the opposite side recedes deeply into the scene. Photograph from point-blank range with an 8mm rectilinear ultra-wide close-focus lens. The closest part appears three to four times larger than the farthest part, creating an expanded foreground, rapid scale reduction into depth, powerful foreshortening and stretched peripheral geometry. The transformation reads entirely as optical perspective while the original design remains intact. Keep the complete silhouette inside the frame with every edge, tip, handle, component and letter fully visible and a 5–8% margin of the original environment around it. Preserve the object’s original physical relationship with the scene, including its support, placement, contact, suspension or interaction with other elements. Retain the original background as the same identifiable place with its surfaces, architecture, furniture, landscape, colors, textures and lighting. Reconstruct the environment consistently from the new camera position so its perspective responds naturally to the ultra-wide lens. Match the original shadows, reflections and color temperature. The final image contains the exact original object or object group within the same recognizable environment, captured as a bold high-end ultra-wide editorial photograph.

copied

Seedream 5.0 Pro

Use the uploaded image as the authoritative full-scene reference for the object and its environment. Preserve the exact number of objects and each object’s recognizable silhouette, rigid construction, proportions, number and placement of parts, colors, materials, textures, patterns, finish, logos, inscriptions and every identifying detail. Preserve the same recognizable background, location, surrounding objects, architecture or landscape, lighting, colors and atmosphere. Analyze the silhouette, principal three-dimensional axis, two most widely separated structural points and the broadest, heaviest or most distinctive volume, then choose a unique camera side, height, tilt and rotation derived from this specific construction. Treat the object as an unchanged rigid volume and create all exaggeration through camera position and perspective. For an elongated object, point its length almost directly toward the camera; for a flat or compact object, turn its main face 55–75 degrees away from the image plane and direct its strongest edge or corner toward the camera; for an asymmetrical object, align its most distinctive volume with the opposite contour point. Position an extreme rectilinear ultra-wide close-focus virtual camera closer to the foreground surface than the object’s own front-to-back depth. The nearest section occupies 50–65% of the composition and appears four to six times larger than the farthest section. Create a continuous scale cascade across the whole object through dramatically expanded near geometry, rapid reduction into depth, severe foreshortening, peripheral stretching and strongly converging structural lines. Build this perspective transformation first, then extend the surrounding canvas until the complete distorted silhouette fits inside the image with 8–12% breathing room. Levitate the object freely within the original environment, with visible open air beneath and around it, complete separation from surrounding surfaces and shadowless suspension. Reproject the same identifiable background consistently from the new viewpoint. The image fills the complete rectangular canvas edge to edge with natural background imagery in all four corners; optical hardware, circular image boundaries, black rims and vignetting are entirely absent.

copied

GPT Image 2.5 Flare

Use the uploaded image as the visual source and preserve the object’s exact recognizable shape, construction, proportions, colors, materials, textures, patterns, logos, inscriptions and placement of every detail. Rebuild the composition around extreme object-specific foreshortening. Analyze the silhouette and identify the two most widely separated structural points, the principal three-dimensional axis and the broadest, heaviest or most distinctive endpoint. Direct this individual axis almost straight toward the camera, placing the selected endpoint only 5–15 cm from the lens while the opposite endpoint recedes deeply away from it. The nearest section occupies 45–60% of the composition and appears three to five times larger than the farthest section. For a compact object, construct the depth axis between its most distinctive asymmetrical feature and the opposite contour point. The frontal surface turns into a sharply receding plane while every identifying feature remains readable. Use a 10 mm rectilinear ultra-wide perspective with clean rectangular edge-to-edge coverage, uninterrupted corners and the lens completely outside the photograph. Prioritize severe near-to-far scale change, elongated depth and dramatic perspective shortening over a balanced frontal product view. Levitate the object freely in the original environment with visible empty space beneath and around it, zero support, zero surface contact and zero cast or contact shadow. Keep the entire object and every attached component inside the frame with a narrow margin of breathing room. Preserve the original background, lighting, colors and atmosphere, transformed consistently by the new camera position. The chosen foreground endpoint, depth direction, rotation and camera height must arise from the unique geometry of each uploaded object, creating a different dramatic viewpoint for every input.

copied

Seedream 5.0 Pro

Use the uploaded image as the sole subject reference. Recreate the same object, product, character, creature, sculpture, logo or lettering with its recognizable identity, underlying structure, silhouette, colors, materials, surface texture, branding and every defining detail faithfully preserved. Change only the camera perspective. Photograph the subject from an extremely close dynamic three-quarter viewpoint using an 8–10mm ultra-wide close-focus lens, positioning the camera only a few centimeters from one naturally prominent part of the subject. Let this closest area dominate the foreground at a dramatically exaggerated scale while the remaining form rapidly tapers and recedes into depth, creating intense forced perspective, powerful foreshortening and a bold sculptural silhouette. Select the foreground anchor according to the subject’s own geometry and pose so the result remains instantly recognizable and compositionally strong. Keep the distortion optical and perspective-driven while maintaining coherent anatomy, construction and connections between every part. Fill most of the frame with one isolated subject and build the composition along a pronounced near-to-far depth axis. For a flat logo, graphic or lettering, create a clean volumetric version while preserving its exact contours, typography and colors. Place the subject on a pure seamless white background with high-key diffused studio lighting, soft frontal fill, bright backlight, luminous rim glow, subtle bloom and airy overexposed highlights. Render the closest material details with tactile realism and create a gentle depth-of-field falloff toward the distant parts. Contemporary premium editorial campaign, clean art direction, playful monumental scale, polished photorealism, high resolution. Deliver a clean final image containing only the transformed subject and the white background.

copied

GPT Image 2.5 Flare

Use the uploaded image as the visual source and preserve the object’s exact recognizable shape, construction, proportions, colors, materials, textures, patterns, logos, inscriptions and placement of every detail. Analyze this specific object before composing the shot: identify its strongest asymmetrical feature, dominant spatial axis, direction of visual weight, most expressive contour and the angle from which its construction looks most distinctive. Build the camera position from these individual characteristics so every new object receives its own viewpoint. A protruding feature creates an axial view along its direction, a flat construction creates a grazing lateral view, a vertically weighted form creates an upper or lower oblique view, and a compact symmetrical form rotates toward its most distinctive side while its identifying front details remain readable. Position the object across several depth planes, with one structurally meaningful area close to the camera and the opposite area receding sharply, creating an obvious difference in scale through real near-to-far perspective. Use a 10–12 mm rectilinear ultra-wide lens with full-sensor edge-to-edge coverage, a clean rectangular photographic frame and uninterrupted background extending naturally into all four corners. The lens, glass, housing, circular border and dark rim remain completely outside the image. Levitate the object freely in the original environment with a clear air gap around and beneath it, zero surface contact, zero support and zero cast or contact shadow below. Give the object an input-specific spatial rotation around its natural construction axis rather than a standard centered product orientation. Keep the complete object and every attached element fully inside the frame with breathing room around the silhouette. Preserve the original background, lighting, colors and atmosphere, transformed consistently by the new camera position. The result feels sculptural, dynamic and editorial, with strong wide-angle perspective created by proximity and depth while the object remains completely recognizable.

copied

GPT Image 2.5 Flare

Use the uploaded image as the sole content source and preserve one complete object with its exact recognizable silhouette, rigid construction, proportions, number and placement of parts, colors, materials, textures, patterns, logos, inscriptions and small identifying details. Create an extreme close-proximity forced-perspective photograph using 6 mm rectilinear ultra-wide optics at minimum focusing distance. Treat the object as an unchanged rigid three-dimensional volume; all visual exaggeration comes entirely from the camera position and viewing direction. Analyze the specific geometry and choose an individual camera axis: follow the length of an elongated form, approach a flat form edge-first, use the strongest asymmetrical contour of a compact form, or look along the functional direction of a projecting part. Rotate the object freely in pitch, yaw and roll until this geometry-specific axis points deeply into the camera, placing one meaningful surface extremely close to the lens while the opposite structure recedes rapidly through space. Create an unmistakable continuous scale gradient across the whole object: nearby surfaces expand dramatically, repeated details shrink quickly with distance, parallel structural lines converge and the far end becomes visibly smaller while every part remains correctly connected. Build this extreme perspective first, then extend the surrounding canvas until the entire already-distorted object fits inside the image with clear breathing room. The image contains one object only, fully visible and structurally intact. Levitate it freely in the original environment with visible empty space beneath it, zero support, zero surface contact and zero cast or contact shadow. Preserve the original background, lighting, colors and atmosphere, reprojected consistently from the new camera position. Use a clean full-bleed rectangular photograph with uninterrupted background in all four corners and the physical lens outside the image. The final viewpoint must arise from this object’s unique construction rather than from a standard frontal or three-quarter product composition.

copied

Seedream 5 Pro

Use the uploaded image as the visual source and preserve the object’s exact recognizable shape, construction, proportions, colors, materials, textures, patterns, logos, inscriptions and placement of every detail. Rebuild the composition around extreme object-specific foreshortening. Analyze the silhouette and identify the two most widely separated structural points, the principal three-dimensional axis and the broadest, heaviest or most distinctive endpoint. Direct this individual axis almost straight toward the camera, placing the selected endpoint only 5–15 cm from the lens while the opposite endpoint recedes deeply away from it. The nearest section occupies 45–60% of the composition and appears three to five times larger than the farthest section. For a compact object, construct the depth axis between its most distinctive asymmetrical feature and the opposite contour point. The frontal surface turns into a sharply receding plane while every identifying feature remains readable. Use a 10 mm rectilinear ultra-wide perspective with clean rectangular edge-to-edge coverage, uninterrupted corners and the lens completely outside the photograph. Prioritize severe near-to-far scale change, elongated depth and dramatic perspective shortening over a balanced frontal product view. Levitate the object freely in the original environment with visible empty space beneath and around it, zero support, zero surface contact and zero cast or contact shadow. Keep the entire object and every attached component inside the frame with a narrow margin of breathing room. Preserve the original background, lighting, colors and atmosphere, transformed consistently by the new camera position. The chosen foreground endpoint, depth direction, rotation and camera height must arise from the unique geometry of each uploaded object, creating a different dramatic viewpoint for every input.

copied

Seedream 5.0 Pro

Use the uploaded image as the visual source and preserve the object’s exact recognizable shape, construction, proportions, colors, materials, textures, patterns, logos, inscriptions and placement of every detail. Rebuild the composition around extreme object-specific foreshortening. Analyze the silhouette and identify the two most widely separated structural points, the principal three-dimensional axis and the broadest, heaviest or most distinctive endpoint. Direct this individual axis almost straight toward the camera, placing the selected endpoint only 5–15 cm from the lens while the opposite endpoint recedes deeply away from it. The nearest section occupies 45–60% of the composition and appears three to five times larger than the farthest section. For a compact object, construct the depth axis between its most distinctive asymmetrical feature and the opposite contour point. The frontal surface turns into a sharply receding plane while every identifying feature remains readable. Use a 10 mm rectilinear ultra-wide perspective with clean rectangular edge-to-edge coverage, uninterrupted corners and the lens completely outside the photograph. Prioritize severe near-to-far scale change, elongated depth and dramatic perspective shortening over a balanced frontal product view. Levitate the object freely in the original environment with visible empty space beneath and around it, zero support, zero surface contact and zero cast or contact shadow. Keep the entire object and every attached component inside the frame with a narrow margin of breathing room. Preserve the original background, lighting, colors and atmosphere, transformed consistently by the new camera position. The chosen foreground endpoint, depth direction, rotation and camera height must arise from the unique geometry of each uploaded object, creating a different dramatic viewpoint for every input.

copied

GPT Image 2.5 Flare

Use the uploaded image as the authoritative full-scene reference for the object and its environment. Preserve the exact number of objects and each object’s recognizable silhouette, rigid construction, proportions, number and placement of parts, colors, materials, textures, patterns, finish, logos, inscriptions and every identifying detail. Preserve the same recognizable background, location, surrounding objects, architecture or landscape, lighting, colors and atmosphere. Analyze the silhouette, principal three-dimensional axis, two most widely separated structural points and the broadest, heaviest or most distinctive volume, then choose a unique camera side, height, tilt and rotation derived from this specific construction. Treat the object as an unchanged rigid volume and create all exaggeration through camera position and perspective. For an elongated object, point its length almost directly toward the camera; for a flat or compact object, turn its main face 55–75 degrees away from the image plane and direct its strongest edge or corner toward the camera; for an asymmetrical object, align its most distinctive volume with the opposite contour point. Position an extreme rectilinear ultra-wide close-focus virtual camera closer to the foreground surface than the object’s own front-to-back depth. The nearest section occupies 50–65% of the composition and appears four to six times larger than the farthest section. Create a continuous scale cascade across the whole object through dramatically expanded near geometry, rapid reduction into depth, severe foreshortening, peripheral stretching and strongly converging structural lines. Build this perspective transformation first, then extend the surrounding canvas until the complete distorted silhouette fits inside the image with 8–12% breathing room. Levitate the object freely within the original environment, with visible open air beneath and around it, complete separation from surrounding surfaces and shadowless suspension. Reproject the same identifiable background consistently from the new viewpoint. The image fills the complete rectangular canvas edge to edge with natural background imagery in all four corners; optical hardware, circular image boundaries, black rims and vignetting are entirely absent.

copied

Nano Banana Pro

Use the uploaded image as the sole content source and preserve one complete object with its exact recognizable silhouette, rigid construction, proportions, number and placement of parts, colors, materials, textures, patterns, logos, inscriptions and small identifying details. Create an extreme close-proximity forced-perspective photograph using 6 mm rectilinear ultra-wide optics at minimum focusing distance. Treat the object as an unchanged rigid three-dimensional volume; all visual exaggeration comes entirely from the camera position and viewing direction. Analyze the specific geometry and choose an individual camera axis: follow the length of an elongated form, approach a flat form edge-first, use the strongest asymmetrical contour of a compact form, or look along the functional direction of a projecting part. Rotate the object freely in pitch, yaw and roll until this geometry-specific axis points deeply into the camera, placing one meaningful surface extremely close to the lens while the opposite structure recedes rapidly through space. Create an unmistakable continuous scale gradient across the whole object: nearby surfaces expand dramatically, repeated details shrink quickly with distance, parallel structural lines converge and the far end becomes visibly smaller while every part remains correctly connected. Build this extreme perspective first, then extend the surrounding canvas until the entire already-distorted object fits inside the image with clear breathing room. The image contains one object only, fully visible and structurally intact. Levitate it freely in the original environment with visible empty space beneath it, zero support, zero surface contact and zero cast or contact shadow. Preserve the original background, lighting, colors and atmosphere, reprojected consistently from the new camera position. Use a clean full-bleed rectangular photograph with uninterrupted background in all four corners and the physical lens outside the image. The final viewpoint must arise from this object’s unique construction rather than from a standard frontal or three-quarter product composition.

copied

GPT Image 2.5 Flare

Use the uploaded image as the visual source and preserve the object’s exact recognizable shape, construction, proportions, colors, materials, textures, patterns, logos, inscriptions and placement of every detail. Analyze this specific object before composing the shot: identify its strongest asymmetrical feature, dominant spatial axis, direction of visual weight, most expressive contour and the angle from which its construction looks most distinctive. Build the camera position from these individual characteristics so every new object receives its own viewpoint. A protruding feature creates an axial view along its direction, a flat construction creates a grazing lateral view, a vertically weighted form creates an upper or lower oblique view, and a compact symmetrical form rotates toward its most distinctive side while its identifying front details remain readable. Position the object across several depth planes, with one structurally meaningful area close to the camera and the opposite area receding sharply, creating an obvious difference in scale through real near-to-far perspective. Use a 10–12 mm rectilinear ultra-wide lens with full-sensor edge-to-edge coverage, a clean rectangular photographic frame and uninterrupted background extending naturally into all four corners. The lens, glass, housing, circular border and dark rim remain completely outside the image. Levitate the object freely in the original environment with a clear air gap around and beneath it, zero surface contact, zero support and zero cast or contact shadow below. Give the object an input-specific spatial rotation around its natural construction axis rather than a standard centered product orientation. Keep the complete object and every attached element fully inside the frame with breathing room around the silhouette. Preserve the original background, lighting, colors and atmosphere, transformed consistently by the new camera position. The result feels sculptural, dynamic and editorial, with strong wide-angle perspective created by proximity and depth while the object remains completely recognizable.

copied

GPT Image 2.5 Flare

Use the uploaded image as the authoritative full-scene reference for the object and its environment. Preserve the exact number of objects and each object’s recognizable silhouette, rigid construction, proportions, number and placement of parts, colors, materials, textures, patterns, finish, logos, inscriptions and every identifying detail. Preserve the same recognizable background, location, surrounding objects, architecture or landscape, lighting, colors and atmosphere. Analyze the silhouette, principal three-dimensional axis, two most widely separated structural points and the broadest, heaviest or most distinctive volume, then choose a unique camera side, height, tilt and rotation derived from this specific construction. Treat the object as an unchanged rigid volume and create all exaggeration through camera position and perspective. For an elongated object, point its length almost directly toward the camera; for a flat or compact object, turn its main face 55–75 degrees away from the image plane and direct its strongest edge or corner toward the camera; for an asymmetrical object, align its most distinctive volume with the opposite contour point. Position an extreme rectilinear ultra-wide close-focus virtual camera closer to the foreground surface than the object’s own front-to-back depth. The nearest section occupies 50–65% of the composition and appears four to six times larger than the farthest section. Create a continuous scale cascade across the whole object through dramatically expanded near geometry, rapid reduction into depth, severe foreshortening, peripheral stretching and strongly converging structural lines. Build this perspective transformation first, then extend the surrounding canvas until the complete distorted silhouette fits inside the image with 8–12% breathing room. Levitate the object freely within the original environment, with visible open air beneath and around it, complete separation from surrounding surfaces and shadowless suspension. Reproject the same identifiable background consistently from the new viewpoint. The image fills the complete rectangular canvas edge to edge with natural background imagery in all four corners; optical hardware, circular image boundaries, black rims and vignetting are entirely absent.

copied

GPT Image 2.5 Sunburst

Use the uploaded image as the authoritative reference for both the object and the complete environment. Preserve the object one-to-one: exact silhouette, proportions, construction, number and placement of parts, materials, textures, surface finish, colors, transparency, reflections, branding, graphics, labels and correctly spelled text. Preserve the exact number of objects and keep grouped elements visually connected. Analyze the object’s geometry and identify the viewing axis that creates the strongest silhouette and greatest near-to-far depth. Move the camera around the object and orient the complete unchanged form so one distinctive structural feature points toward the lens while the opposite side recedes deeply into the scene. Photograph from point-blank range with an 8mm rectilinear ultra-wide close-focus lens. The closest part appears three to four times larger than the farthest part, creating an expanded foreground, rapid scale reduction into depth, powerful foreshortening and stretched peripheral geometry. The transformation reads entirely as optical perspective while the original design remains intact. Keep the complete silhouette inside the frame with every edge, tip, handle, component and letter fully visible and a 5–8% margin of the original environment around it. Preserve the object’s original physical relationship with the scene, including its support, placement, contact, suspension or interaction with other elements. Retain the original background as the same identifiable place with its surfaces, architecture, furniture, landscape, colors, textures and lighting. Reconstruct the environment consistently from the new camera position so its perspective responds naturally to the ultra-wide lens. Match the original shadows, reflections and color temperature. The final image contains the exact original object or object group within the same recognizable environment, captured as a bold high-end ultra-wide editorial photograph.

copied

GPT Image 2.5 Flare

Use the uploaded image as the sole content source and preserve one complete object with its exact recognizable silhouette, rigid construction, proportions, number and placement of parts, colors, materials, textures, patterns, logos, inscriptions and small identifying details. Create an extreme close-proximity forced-perspective photograph using 6 mm rectilinear ultra-wide optics at minimum focusing distance. Treat the object as an unchanged rigid three-dimensional volume; all visual exaggeration comes entirely from the camera position and viewing direction. Analyze the specific geometry and choose an individual camera axis: follow the length of an elongated form, approach a flat form edge-first, use the strongest asymmetrical contour of a compact form, or look along the functional direction of a projecting part. Rotate the object freely in pitch, yaw and roll until this geometry-specific axis points deeply into the camera, placing one meaningful surface extremely close to the lens while the opposite structure recedes rapidly through space. Create an unmistakable continuous scale gradient across the whole object: nearby surfaces expand dramatically, repeated details shrink quickly with distance, parallel structural lines converge and the far end becomes visibly smaller while every part remains correctly connected. Build this extreme perspective first, then extend the surrounding canvas until the entire already-distorted object fits inside the image with clear breathing room. The image contains one object only, fully visible and structurally intact. Levitate it freely in the original environment with visible empty space beneath it, zero support, zero surface contact and zero cast or contact shadow. Preserve the original background, lighting, colors and atmosphere, reprojected consistently from the new camera position. Use a clean full-bleed rectangular photograph with uninterrupted background in all four corners and the physical lens outside the image. The final viewpoint must arise from this object’s unique construction rather than from a standard frontal or three-quarter product composition.

copied

WIDE ANGLE

Объекты и люди

Для объектов

Для людей