{ "version": 1, "generatedAt": "2026-09-25T13:26:06.749106+00:00", "assets": { "product-source": { "id": "product-source", "workflowStep": "Create shared reference", "kind": "image", "endpoint": "openai/gpt-image-2.5/flare/text-to-image", "prompt": "Create one image only, a single 16:9 editorial product photograph. A forest-green enamel camping lantern with a black arched carry handle, ribbed clear glass chamber, warm-white LED core switched off, three small brass fasteners and a cream rectangular label reading exactly \"FENN\" in large readable dark letters. No other readable text. Low three-quarter camera angle, 70mm lens, the entire lantern fits comfortably in frame, object occupying the central 55 percent. Cool blue-gray seamless studio background and matte blue-gray floor, crisp contact shadow. Large softbox from camera left, narrow white edge reflection on glass, deep but open shadows. Tactile powder-coated metal, tiny enamel speckles and realistic glass refraction. Art direction: museum product still life with industrial color discipline. Composition system: one isolated object, generous negative space. Typography: compact stamped grotesk printed on the real product label. No people, no grid, no collage, no fake interface, no watermark.", "phase": "sources", "aspectRatio": "16:9", "input": { "image_size": { "width": 2048, "height": 1152 }, "quality": "max", "num_images": 1, "output_format": "png" }, "qaCriteria": [ "Reference is clear and physically coherent", "Record actual differences without declaring a universal winner", "Verify framing, labels, decoded output and provenance" ], "failurePolicy": "Do not automatically resubmit. Review failures; preserve successful comparison outcomes without selective enhancement.", "requestId": "01a0d8be-a5ff-7290-8ec4-47a7afbd8d53", "resolvedInput": { "image_size": { "width": 2048, "height": 1152 }, "quality": "max", "num_images": 1, "output_format": "png", "prompt": "Create one image only, a single 16:9 editorial product photograph. A forest-green enamel camping lantern with a black arched carry handle, ribbed clear glass chamber, warm-white LED core switched off, three small brass fasteners and a cream rectangular label reading exactly \"FENN\" in large readable dark letters. No other readable text. Low three-quarter camera angle, 70mm lens, the entire lantern fits comfortably in frame, object occupying the central 55 percent. Cool blue-gray seamless studio background and matte blue-gray floor, crisp contact shadow. Large softbox from camera left, narrow white edge reflection on glass, deep but open shadows. Tactile powder-coated metal, tiny enamel speckles and realistic glass refraction. Art direction: museum product still life with industrial color discipline. Composition system: one isolated object, generous negative space. Typography: compact stamped grotesk printed on the real product label. No people, no grid, no collage, no fake interface, no watermark." }, "submittedAt": "2026-09-25T13:26:08.185329+00:00", "qaStatus": "approved", "url": "https://v3b.fal.media/files/b/0aabd8a4/hAxr4nLBvCeB_JjvwKe9g_Riz2K3RZ.png", "sha256": "e69e7ffd291a26dd8d7ff935600d27b85fb2feb985984d8b6e22e00e5abb4853", "bytes": 2555270, "completedAt": "2026-09-25T13:26:51.934770+00:00", "qaNotes": "Readable FENN label; coherent glass, handle and contact shadow. Single object, clean backdrop. Approved for controlled background edits.", "textQa": "readable", "interactionQa": "not-applicable", "identityQa": "stable", "compositionQa": "strong", "status": "ready", "src": "https://v3b.fal.media/files/b/0aabd8a4/hAxr4nLBvCeB_JjvwKe9g_Riz2K3RZ.png", "source": "generated-fal" }, "portrait-source": { "id": "portrait-source", "workflowStep": "Create shared reference", "kind": "image", "endpoint": "openai/gpt-image-2.5/flare/text-to-image", "prompt": "Create one image only, a single 16:9 photographic portrait of a fictional adult woman in her thirties with short dark curls, warm brown eyes, a few natural cheek blemishes and honest visible skin texture, wearing a cobalt cotton shirt. Tight head-and-shoulders crop, face turned slightly toward camera, hair fully visible, shoulders relaxed. Camera: 85mm portrait lens at eye height, face and curls sharply resolved with a softly receding background. Lighting: warm window light from frame left and cool soft fill from the right, one plausible catchlight per eye, a small natural forehead highlight. Background is a simple deep terracotta plaster wall. Art direction: intimate documentary editorial, warm skin against cobalt fabric and terracotta, no glossy beauty-campaign smoothing. Composition: face centered with breathing room around hair. Materials: pores, cotton weave, distinct curl strands. No text, no jewelry, no hands in frame, no collage, no artificial grain overlay, no watermark.", "phase": "sources", "aspectRatio": "16:9", "input": { "image_size": { "width": 2048, "height": 1152 }, "quality": "max", "num_images": 1, "output_format": "png" }, "qaCriteria": [ "Reference is clear and physically coherent", "Record actual differences without declaring a universal winner", "Verify framing, labels, decoded output and provenance" ], "failurePolicy": "Do not automatically resubmit. Review failures; preserve successful comparison outcomes without selective enhancement.", "requestId": "01a0d8be-a6b0-77f0-9b99-e863f53b9cca", "resolvedInput": { "image_size": { "width": 2048, "height": 1152 }, "quality": "max", "num_images": 1, "output_format": "png", "prompt": "Create one image only, a single 16:9 photographic portrait of a fictional adult woman in her thirties with short dark curls, warm brown eyes, a few natural cheek blemishes and honest visible skin texture, wearing a cobalt cotton shirt. Tight head-and-shoulders crop, face turned slightly toward camera, hair fully visible, shoulders relaxed. Camera: 85mm portrait lens at eye height, face and curls sharply resolved with a softly receding background. Lighting: warm window light from frame left and cool soft fill from the right, one plausible catchlight per eye, a small natural forehead highlight. Background is a simple deep terracotta plaster wall. Art direction: intimate documentary editorial, warm skin against cobalt fabric and terracotta, no glossy beauty-campaign smoothing. Composition: face centered with breathing room around hair. Materials: pores, cotton weave, distinct curl strands. No text, no jewelry, no hands in frame, no collage, no artificial grain overlay, no watermark." }, "submittedAt": "2026-09-25T13:26:08.359403+00:00", "qaStatus": "approved", "url": "https://v3b.fal.media/files/b/0aabd8a5/XVmTVEPPJ4IIpamqicDj4_YmMfEKud.png", "sha256": "5c30e36b264e77c896d26d9ca274a4bbc974a44a4c4e007ecff36714bd6763f4", "bytes": 3027580, "completedAt": "2026-09-25T13:26:53.933040+00:00", "qaNotes": "Natural skin texture, clear curls and consistent directional light. Approved for localized retouch comparisons.", "textQa": "none", "interactionQa": "not-applicable", "identityQa": "stable", "compositionQa": "strong", "status": "ready", "src": "https://v3b.fal.media/files/b/0aabd8a5/XVmTVEPPJ4IIpamqicDj4_YmMfEKud.png", "source": "generated-fal" }, "motion-source": { "id": "motion-source", "workflowStep": "Create shared reference", "kind": "image", "endpoint": "openai/gpt-image-2.5/flare/text-to-image", "prompt": "Create one image only, a single 16:9 film production still. A fictional adult cyclist in a burnt-orange windbreaker and dark trousers rides a plain silver city bicycle from left to right along a sunlit riverside path. Full bicycle and rider visible in side profile, both hands naturally holding handlebars, feet placed on pedals, two coherent circular wheels with fine spokes. Camera at waist height, 50mm tracking-lens perspective, cyclist centered with space ahead. Late-afternoon sun through willow branches, bright small leaf highlights, dappled shadows on the path, distant blue water. Art direction: documentary outdoor campaign with truthful physical detail. Composition: clear rider silhouette against green foliage; foreground branch at far left gives parallax. Materials: woven jacket, brushed metal, rubber tires, textured asphalt. No readable text, no logos, no collage, no grid, no watermark.", "phase": "sources", "aspectRatio": "16:9", "input": { "image_size": { "width": 2048, "height": 1152 }, "quality": "max", "num_images": 1, "output_format": "png" }, "qaCriteria": [ "Reference is clear and physically coherent", "Record actual differences without declaring a universal winner", "Verify framing, labels, decoded output and provenance" ], "failurePolicy": "Do not automatically resubmit. Review failures; preserve successful comparison outcomes without selective enhancement.", "requestId": "01a0d8be-a75a-7033-a85d-8af0e74a513f", "resolvedInput": { "image_size": { "width": 2048, "height": 1152 }, "quality": "max", "num_images": 1, "output_format": "png", "prompt": "Create one image only, a single 16:9 film production still. A fictional adult cyclist in a burnt-orange windbreaker and dark trousers rides a plain silver city bicycle from left to right along a sunlit riverside path. Full bicycle and rider visible in side profile, both hands naturally holding handlebars, feet placed on pedals, two coherent circular wheels with fine spokes. Camera at waist height, 50mm tracking-lens perspective, cyclist centered with space ahead. Late-afternoon sun through willow branches, bright small leaf highlights, dappled shadows on the path, distant blue water. Art direction: documentary outdoor campaign with truthful physical detail. Composition: clear rider silhouette against green foliage; foreground branch at far left gives parallax. Materials: woven jacket, brushed metal, rubber tires, textured asphalt. No readable text, no logos, no collage, no grid, no watermark." }, "submittedAt": "2026-09-25T13:26:08.526957+00:00", "qaStatus": "approved", "url": "https://v3b.fal.media/files/b/0aabd8a5/_Wj5jwjRCgU2Ubabni7To_LIzfHIjs.png", "sha256": "68c7e4bf50bb626fc51541d89304c6c95581226824d628ff0241654c75c41dd2", "bytes": 4420410, "completedAt": "2026-09-25T13:26:55.847110+00:00", "qaNotes": "Full cyclist and bicycle visible, plausible hand and foot contacts, two wheels and trackable foliage. Approved as the shared video starting frame.", "textQa": "none", "interactionQa": "correct", "identityQa": "stable", "compositionQa": "strong", "status": "ready", "src": "https://v3b.fal.media/files/b/0aabd8a5/_Wj5jwjRCgU2Ubabni7To_LIzfHIjs.png", "source": "generated-fal" }, "product-flare": { "id": "product-flare", "workflowStep": "Compare identical edit instructions", "kind": "image", "endpoint": "openai/gpt-image-2.5/flare/edit", "prompt": "Use the supplied photograph as the reference. Create one image only. Replace only the blue-gray studio background and floor with a slate ledge beside a softly defocused pine forest at dusk. Preserve the FENN lantern exactly: its forest-green enamel color, black handle, ribbed glass, fastener count, cream label and readable FENN lettering, size, location and camera perspective. Keep the LED switched off. Match ambient cool dusk light and a physically plausible contact shadow on the ledge, without repainting the product or adding glow. Preserve the original 16:9 framing. Camera remains locked, 70mm product perspective. Keep material texture and fine glass edges sharp. No people, extra props, extra words or logos. No collage. No split panels. No watermark or changes to product geometry.", "phase": "edits", "dependsOn": [ "product-source" ], "aspectRatio": "16:9", "input": { "image_size": { "width": 2048, "height": 1152 }, "num_images": 1, "output_format": "png", "quality": "max" }, "qaCriteria": [ "Reference is clear and physically coherent", "Record actual differences without declaring a universal winner", "Verify framing, labels, decoded output and provenance" ], "failurePolicy": "Do not automatically resubmit. Review failures; preserve successful comparison outcomes without selective enhancement.", "requestId": "01a0d8c6-46d8-7932-b4d2-cea0d4112385", "resolvedInput": { "image_size": { "width": 2048, "height": 1152 }, "num_images": 1, "output_format": "png", "quality": "max", "prompt": "Use the supplied photograph as the reference. Create one image only. Replace only the blue-gray studio background and floor with a slate ledge beside a softly defocused pine forest at dusk. Preserve the FENN lantern exactly: its forest-green enamel color, black handle, ribbed glass, fastener count, cream label and readable FENN lettering, size, location and camera perspective. Keep the LED switched off. Match ambient cool dusk light and a physically plausible contact shadow on the ledge, without repainting the product or adding glow. Preserve the original 16:9 framing. Camera remains locked, 70mm product perspective. Keep material texture and fine glass edges sharp. No people, extra props, extra words or logos. No collage. No split panels. No watermark or changes to product geometry.", "image_urls": [ "https://v3b.fal.media/files/b/0aabd8a4/hAxr4nLBvCeB_JjvwKe9g_Riz2K3RZ.png" ] }, "submittedAt": "2026-09-25T13:34:28.109011+00:00", "qaStatus": "approved", "url": "https://v3b.fal.media/files/b/0aabd8d7/gZlPm5TQx-dMHMwolc_o7_UGcImU9N.png", "sha256": "296e71aba3b73bd8c94c81a7f7ca7e476f66cf932ea152552ad99b5b3103b0be", "bytes": 2797049, "completedAt": "2026-09-25T13:35:24.696282+00:00", "qaNotes": "Approved for comparison: readable FENN label and coherent handle/glass. Dark dusk setting, rough ledge and horizon differ from Seedream; retain these actual output differences.", "status": "ready", "src": "https://v3b.fal.media/files/b/0aabd8d7/gZlPm5TQx-dMHMwolc_o7_UGcImU9N.png", "source": "generated-fal" }, "product-seedream": { "id": "product-seedream", "workflowStep": "Compare identical edit instructions", "kind": "image", "endpoint": "bytedance/seedream/v5/pro/edit", "prompt": "Use the supplied photograph as the reference. Create one image only. Replace only the blue-gray studio background and floor with a slate ledge beside a softly defocused pine forest at dusk. Preserve the FENN lantern exactly: its forest-green enamel color, black handle, ribbed glass, fastener count, cream label and readable FENN lettering, size, location and camera perspective. Keep the LED switched off. Match ambient cool dusk light and a physically plausible contact shadow on the ledge, without repainting the product or adding glow. Preserve the original 16:9 framing. Camera remains locked, 70mm product perspective. Keep material texture and fine glass edges sharp. No people, extra props, extra words or logos. No collage. No split panels. No watermark or changes to product geometry.", "phase": "edits", "dependsOn": [ "product-source" ], "aspectRatio": "16:9", "input": { "image_size": { "width": 2048, "height": 1152 }, "num_images": 1, "output_format": "png" }, "qaCriteria": [ "Reference is clear and physically coherent", "Record actual differences without declaring a universal winner", "Verify framing, labels, decoded output and provenance" ], "failurePolicy": "Do not automatically resubmit. Review failures; preserve successful comparison outcomes without selective enhancement.", "requestId": "01a0d8c6-4784-7412-a4c3-e1bf03460661", "resolvedInput": { "image_size": { "width": 2048, "height": 1152 }, "num_images": 1, "output_format": "png", "prompt": "Use the supplied photograph as the reference. Create one image only. Replace only the blue-gray studio background and floor with a slate ledge beside a softly defocused pine forest at dusk. Preserve the FENN lantern exactly: its forest-green enamel color, black handle, ribbed glass, fastener count, cream label and readable FENN lettering, size, location and camera perspective. Keep the LED switched off. Match ambient cool dusk light and a physically plausible contact shadow on the ledge, without repainting the product or adding glow. Preserve the original 16:9 framing. Camera remains locked, 70mm product perspective. Keep material texture and fine glass edges sharp. No people, extra props, extra words or logos. No collage. No split panels. No watermark or changes to product geometry.", "image_urls": [ "https://v3b.fal.media/files/b/0aabd8a4/hAxr4nLBvCeB_JjvwKe9g_Riz2K3RZ.png" ] }, "submittedAt": "2026-09-25T13:34:28.282452+00:00", "qaStatus": "approved", "url": "https://v3b.fal.media/files/b/0aabd8e0/1obMqUtvzsnMojULjYSSD_a5c9d769492b4ffa876aaa46b0df0176.png", "sha256": "b1b0a085e0050f5b29d69883f5a12f5b152993f84b988822464c9f76f787ad6c", "bytes": 3971973, "completedAt": "2026-09-25T13:36:12.637912+00:00", "qaNotes": "Approved for comparison: readable FENN label and coherent glass/handle. Broad slate ledge and close forest backdrop; brighter product highlights remain. No selective retouching.", "status": "ready", "src": "https://v3b.fal.media/files/b/0aabd8e0/1obMqUtvzsnMojULjYSSD_a5c9d769492b4ffa876aaa46b0df0176.png", "source": "generated-fal" }, "portrait-flare": { "id": "portrait-flare", "workflowStep": "Compare identical edit instructions", "kind": "image", "endpoint": "openai/gpt-image-2.5/flare/edit", "prompt": "Use the supplied portrait as the reference. Create one image only. Make a restrained editorial retouch: soften the small cheek blemishes and slightly reduce forehead shine. Preserve visible pores, natural skin texture, facial anatomy, expression, gaze, eye color, curl shape, hair edges, cobalt shirt, terracotta background, lighting direction and exact camera framing. Preserve the warm window key and cool fill; do not change skin color, smooth all texture, enlarge eyes, reshape features or introduce makeup. Maintain 16:9 composition and the 85mm portrait perspective. This is a localized skin edit, not a new photograph. No text, collage, extra people, artificial grain, watermark or beauty-filter appearance.", "phase": "edits", "dependsOn": [ "portrait-source" ], "aspectRatio": "16:9", "input": { "image_size": { "width": 2048, "height": 1152 }, "num_images": 1, "output_format": "png", "quality": "max" }, "qaCriteria": [ "Reference is clear and physically coherent", "Record actual differences without declaring a universal winner", "Verify framing, labels, decoded output and provenance" ], "failurePolicy": "Do not automatically resubmit. Review failures; preserve successful comparison outcomes without selective enhancement.", "requestId": "01a0d8c6-4832-7061-8840-0f67bf095ce7", "resolvedInput": { "image_size": { "width": 2048, "height": 1152 }, "num_images": 1, "output_format": "png", "quality": "max", "prompt": "Use the supplied portrait as the reference. Create one image only. Make a restrained editorial retouch: soften the small cheek blemishes and slightly reduce forehead shine. Preserve visible pores, natural skin texture, facial anatomy, expression, gaze, eye color, curl shape, hair edges, cobalt shirt, terracotta background, lighting direction and exact camera framing. Preserve the warm window key and cool fill; do not change skin color, smooth all texture, enlarge eyes, reshape features or introduce makeup. Maintain 16:9 composition and the 85mm portrait perspective. This is a localized skin edit, not a new photograph. No text, collage, extra people, artificial grain, watermark or beauty-filter appearance.", "image_urls": [ "https://v3b.fal.media/files/b/0aabd8a5/XVmTVEPPJ4IIpamqicDj4_YmMfEKud.png" ] }, "submittedAt": "2026-09-25T13:34:28.455257+00:00", "qaStatus": "approved", "url": "https://v3b.fal.media/files/b/0aabd8d7/j_27G7uYKatyJARvmgcRJ_PGITqwog.png", "sha256": "41e3535d8d28a357328a702ebc5a873f8ec18a3a93259d7e0cf72c95a5d82e67", "bytes": 2991558, "completedAt": "2026-09-25T13:36:14.367712+00:00", "qaNotes": "Approved for comparison: coherent fictional adult identity and framing, pores and cheek texture visible; localized changes are deliberately subtle.", "status": "ready", "src": "https://v3b.fal.media/files/b/0aabd8d7/j_27G7uYKatyJARvmgcRJ_PGITqwog.png", "source": "generated-fal" }, "portrait-seedream": { "id": "portrait-seedream", "workflowStep": "Compare identical edit instructions", "kind": "image", "endpoint": "bytedance/seedream/v5/pro/edit", "prompt": "Use the supplied portrait as the reference. Create one image only. Make a restrained editorial retouch: soften the small cheek blemishes and slightly reduce forehead shine. Preserve visible pores, natural skin texture, facial anatomy, expression, gaze, eye color, curl shape, hair edges, cobalt shirt, terracotta background, lighting direction and exact camera framing. Preserve the warm window key and cool fill; do not change skin color, smooth all texture, enlarge eyes, reshape features or introduce makeup. Maintain 16:9 composition and the 85mm portrait perspective. This is a localized skin edit, not a new photograph. No text, collage, extra people, artificial grain, watermark or beauty-filter appearance.", "phase": "edits", "dependsOn": [ "portrait-source" ], "aspectRatio": "16:9", "input": { "image_size": { "width": 2048, "height": 1152 }, "num_images": 1, "output_format": "png" }, "qaCriteria": [ "Reference is clear and physically coherent", "Record actual differences without declaring a universal winner", "Verify framing, labels, decoded output and provenance" ], "failurePolicy": "Do not automatically resubmit. Review failures; preserve successful comparison outcomes without selective enhancement.", "requestId": "01a0d8c6-48dd-7471-b8ec-ca8b033ed3fb", "resolvedInput": { "image_size": { "width": 2048, "height": 1152 }, "num_images": 1, "output_format": "png", "prompt": "Use the supplied portrait as the reference. Create one image only. Make a restrained editorial retouch: soften the small cheek blemishes and slightly reduce forehead shine. Preserve visible pores, natural skin texture, facial anatomy, expression, gaze, eye color, curl shape, hair edges, cobalt shirt, terracotta background, lighting direction and exact camera framing. Preserve the warm window key and cool fill; do not change skin color, smooth all texture, enlarge eyes, reshape features or introduce makeup. Maintain 16:9 composition and the 85mm portrait perspective. This is a localized skin edit, not a new photograph. No text, collage, extra people, artificial grain, watermark or beauty-filter appearance.", "image_urls": [ "https://v3b.fal.media/files/b/0aabd8a5/XVmTVEPPJ4IIpamqicDj4_YmMfEKud.png" ] }, "submittedAt": "2026-09-25T13:34:28.626612+00:00", "qaStatus": "approved", "url": "https://v3b.fal.media/files/b/0aabd8db/8_Z5_ZB1aRf9tZ7P74sm4_8692341c3f7d45928f6943c820606f27.png", "sha256": "1fbc39a9d705041f9bf1358f65dcd5fd2133a05e87d5c4555a1224d6562449ce", "bytes": 2988081, "completedAt": "2026-09-25T13:36:16.375895+00:00", "qaNotes": "Approved for comparison: coherent face and curls, softer forehead highlights and visibly smoother cheek texture than the paired output. Preserve the actual difference.", "status": "ready", "src": "https://v3b.fal.media/files/b/0aabd8db/8_Z5_ZB1aRf9tZ7P74sm4_8692341c3f7d45928f6943c820606f27.png", "source": "generated-fal" }, "motion-seedance": { "id": "motion-seedance", "workflowStep": "Compare identical motion instructions", "kind": "video", "endpoint": "bytedance/seedance-2.5/image-to-video", "prompt": "Five seconds, one continuous side-tracking shot from the supplied starting frame. The adult cyclist pedals smoothly from left to right along the riverside path while the camera tracks at the same speed and waist height, keeping the entire rider and bicycle comfortably in frame. Willow leaves pass in the foreground and sunlight moves across the burnt-orange jacket. Both wheels rotate around fixed axles with coherent spokes; the chain and pedals stay mechanically connected, hands remain on the handlebars and feet on the pedals. Preserve the face, clothing, bicycle geometry, setting, camera perspective and warm afternoon light. No cuts, zooms, slow motion, new objects, text or logos. Audio: quiet tire noise, soft wind and distant birds; no music or speech.", "phase": "videos", "dependsOn": [ "motion-source" ], "aspectRatio": "16:9", "input": { "duration": "5", "resolution": "1080p", "generate_audio": true, "bitrate_mode": "high", "codec": "H264" }, "qaCriteria": [ "Reference is clear and physically coherent", "Record actual differences without declaring a universal winner", "Verify framing, labels, decoded output and provenance" ], "failurePolicy": "Do not automatically resubmit. Review failures; preserve successful comparison outcomes without selective enhancement.", "requestId": "01a0d8c7-f04d-7743-b254-bae41ddb2953", "resolvedInput": { "duration": "5", "resolution": "1080p", "generate_audio": true, "bitrate_mode": "high", "codec": "H264", "prompt": "Five seconds, one continuous side-tracking shot from the supplied starting frame. The adult cyclist pedals smoothly from left to right along the riverside path while the camera tracks at the same speed and waist height, keeping the entire rider and bicycle comfortably in frame. Willow leaves pass in the foreground and sunlight moves across the burnt-orange jacket. Both wheels rotate around fixed axles with coherent spokes; the chain and pedals stay mechanically connected, hands remain on the handlebars and feet on the pedals. Preserve the face, clothing, bicycle geometry, setting, camera perspective and warm afternoon light. No cuts, zooms, slow motion, new objects, text or logos. Audio: quiet tire noise, soft wind and distant birds; no music or speech.", "image_url": "https://v3b.fal.media/files/b/0aabd8a5/_Wj5jwjRCgU2Ubabni7To_LIzfHIjs.png" }, "submittedAt": "2026-09-25T13:36:17.026474+00:00", "qaStatus": "approved", "url": "https://v3b.fal.media/files/b/0aabd90b/0L2dGhHLODxxKI2FmA4_m_video.mp4", "sha256": "c66c72ab7aaf1e847054fcecb7cbacca7e7c3caff9c103d4bfb4803ac63947a2", "bytes": 37052536, "completedAt": "2026-09-25T13:43:24.907481+00:00", "qaNotes": "Approved for comparison after five-frame contact-sheet inspection and browser playback: cyclist and bicycle remain coherent; a foreground tree occludes the rider late in the clip. This is an output difference, not corrected.", "status": "ready", "src": "https://v3b.fal.media/files/b/0aabd90b/0L2dGhHLODxxKI2FmA4_m_video.mp4", "source": "generated-fal" }, "motion-h3": { "id": "motion-h3", "workflowStep": "Compare identical motion instructions", "kind": "video", "endpoint": "minimax/h3-max/image-to-video", "prompt": "Five seconds, one continuous side-tracking shot from the supplied starting frame. The adult cyclist pedals smoothly from left to right along the riverside path while the camera tracks at the same speed and waist height, keeping the entire rider and bicycle comfortably in frame. Willow leaves pass in the foreground and sunlight moves across the burnt-orange jacket. Both wheels rotate around fixed axles with coherent spokes; the chain and pedals stay mechanically connected, hands remain on the handlebars and feet on the pedals. Preserve the face, clothing, bicycle geometry, setting, camera perspective and warm afternoon light. No cuts, zooms, slow motion, new objects, text or logos. Audio: quiet tire noise, soft wind and distant birds; no music or speech.", "phase": "videos", "dependsOn": [ "motion-source" ], "aspectRatio": "16:9", "input": { "duration": 5, "resolution": "1080P", "prompt_expansion_mode": "disabled", "seed": 42 }, "qaCriteria": [ "Reference is clear and physically coherent", "Record actual differences without declaring a universal winner", "Verify framing, labels, decoded output and provenance" ], "failurePolicy": "Do not automatically resubmit. Review failures; preserve successful comparison outcomes without selective enhancement.", "requestId": "01a0d8c7-f0f4-7d91-89d1-f9875dd14ec1", "resolvedInput": { "duration": 5, "resolution": "1080P", "prompt_expansion_mode": "disabled", "seed": 42, "prompt": "Five seconds, one continuous side-tracking shot from the supplied starting frame. The adult cyclist pedals smoothly from left to right along the riverside path while the camera tracks at the same speed and waist height, keeping the entire rider and bicycle comfortably in frame. Willow leaves pass in the foreground and sunlight moves across the burnt-orange jacket. Both wheels rotate around fixed axles with coherent spokes; the chain and pedals stay mechanically connected, hands remain on the handlebars and feet on the pedals. Preserve the face, clothing, bicycle geometry, setting, camera perspective and warm afternoon light. No cuts, zooms, slow motion, new objects, text or logos. Audio: quiet tire noise, soft wind and distant birds; no music or speech.", "image_url": "https://v3b.fal.media/files/b/0aabd8a5/_Wj5jwjRCgU2Ubabni7To_LIzfHIjs.png" }, "submittedAt": "2026-09-25T13:36:17.192872+00:00", "qaStatus": "approved", "url": "https://v3b.fal.media/files/b/0aabd8e1/m0TIHSGrBz1lP-xS2QG-a_minimax-h3.mp4", "sha256": "897a841b479f4c06da6df707ec8603cc887f1acd237217a71704ab83bc2d7984", "bytes": 27275037, "completedAt": "2026-09-25T13:43:28.070203+00:00", "qaNotes": "Approved for comparison after five-frame contact-sheet inspection and browser playback: coherent rider/bicycle, different pedaling and foreground timing from Seedance. Not frame-aligned action.", "status": "ready", "src": "https://v3b.fal.media/files/b/0aabd8e1/m0TIHSGrBz1lP-xS2QG-a_minimax-h3.mp4", "source": "generated-fal" } }, "delivery": { "product-source.webp": { "source": "product-source", "encoding": "Lossless WebP; unchanged pixels and dimensions", "bytes": 1541104, "sha256": "47132a393a9b3a495cfd89dd1e181b0d9d98a03943035d9059d80da21bece152" }, "product-flare.webp": { "source": "product-flare", "encoding": "Lossless WebP; unchanged pixels and dimensions", "bytes": 1752152, "sha256": "f037869bd1c5d982e77815d50bf3615072aa8bb8866120da588612eb4b5625ad" }, "product-seedream.webp": { "source": "product-seedream", "encoding": "Lossless WebP; unchanged pixels and dimensions", "bytes": 3458174, "sha256": "7289a2a1c15b641cd5b8f98960820cdd6af15fd01a838462aab064f23c0f1468" }, "portrait-source.webp": { "source": "portrait-source", "encoding": "Lossless WebP; unchanged pixels and dimensions", "bytes": 1974510, "sha256": "c85f66076776a68c1fcd5647ee32a524767932223c47199587d9404bccc112ab" }, "portrait-flare.webp": { "source": "portrait-flare", "encoding": "Lossless WebP; unchanged pixels and dimensions", "bytes": 1945268, "sha256": "84970bb270c544da3927d5930b076f8fa77bb48b82fee27de97b0e0cd8ecc635" }, "portrait-seedream.webp": { "source": "portrait-seedream", "encoding": "Lossless WebP; unchanged pixels and dimensions", "bytes": 2047990, "sha256": "d6ea7f1a3e394c74197e406666ed10f3db7428553ca1c56d3b5aa2572dba3360" }, "motion-source.webp": { "source": "motion-source", "encoding": "Lossless WebP; unchanged pixels and dimensions", "bytes": 2956778, "sha256": "fc156094a3728faa7206844e6d4efe49c4150624329c2f064df0493d8914f2a1" }, "motion-seedance.mp4": { "source": "motion-seedance", "codec": "H.264", "crf": 18, "maxRate": "5 Mbps (VBV buffer 5 Mb)", "preset": "slow", "filter": "none", "audio": "removed in all visual comparisons", "trim": "first 5 seconds; no frame-rate conversion", "width": 1920, "height": 1080, "r_frame_rate": "24/1", "duration": "5.000000", "nb_read_frames": "120", "bytes": 3171811, "sha256": "6f3d3cc577ae688a88b3cce4ea64a6ee706a87b4d1289890b2d126c1969bc133" }, "motion-h3.mp4": { "source": "motion-h3", "codec": "H.264", "crf": 18, "maxRate": "5 Mbps (VBV buffer 5 Mb)", "preset": "slow", "filter": "none", "audio": "removed in all visual comparisons", "trim": "first 5 seconds; no frame-rate conversion", "width": 1920, "height": 1080, "r_frame_rate": "24/1", "duration": "5.000000", "nb_read_frames": "120", "bytes": 3096075, "sha256": "08cf4eda802e14b72186d2b786285bc70f0a3b04b0fab70068baa4d56401570e" }, "motion-120p.mp4": { "source": "motion-seedance", "codec": "H.264", "crf": 18, "maxRate": "5 Mbps (VBV buffer 5 Mb)", "preset": "slow", "filter": "scale=214:120:flags=lanczos,setsar=1", "audio": "removed in all visual comparisons", "trim": "first 5 seconds; no frame-rate conversion", "width": 214, "height": 120, "r_frame_rate": "24/1", "duration": "5.000000", "nb_read_frames": "120", "bytes": 466100, "sha256": "b66d6b0fca6acb9df2d0537fe615d9a211d63ca08ea02ea99966a436d5b4f1e6" }, "motion-compressed.mp4": { "source": "motion-seedance", "codec": "H.264", "crf": 40, "maxRate": "uncapped", "preset": "slow", "filter": "none", "audio": "removed in all visual comparisons", "trim": "first 5 seconds; no frame-rate conversion", "width": 1920, "height": 1080, "r_frame_rate": "24/1", "duration": "5.000000", "nb_read_frames": "120", "bytes": 1097297, "sha256": "32cb806baac914d235ba1810f86d552d4b45563f627f8c6cdbad2bd53278b278" }, "product-thumb.webp": { "source": [ "public/studio/product-flare.webp", "public/studio/product-seedream.webp" ], "encoding": "640 \u00d7 360 split preview, WebP quality 72; thumbnails are not comparison inputs", "bytes": 16840, "sha256": "a599c95435dbe9c7d44807500962374ae3a0ce80dcf98399d2e908391cfce339" }, "portrait-thumb.webp": { "source": [ "public/studio/portrait-flare.webp", "public/studio/portrait-seedream.webp" ], "encoding": "640 \u00d7 360 split preview, WebP quality 72; thumbnails are not comparison inputs", "bytes": 14400, "sha256": "be6e59eb9cbfa17c471f4e9a626334473ef15a8f5c0a682fce45093402e2069b" }, "motion-thumb.webp": { "source": "Matching frames at 2 seconds from the corresponding delivery pair", "encoding": "640 \u00d7 360 split preview, WebP quality 72; thumbnails are not comparison inputs", "bytes": 49094, "sha256": "7d3c1863db37b394a0823d43d59faaba5cdbad5fcd26e15094feb8f5008ce636" }, "resolution-thumb.webp": { "source": "Matching frames at 2 seconds from the corresponding delivery pair", "encoding": "640 \u00d7 360 split preview, WebP quality 72; thumbnails are not comparison inputs", "bytes": 39510, "sha256": "95c90d6e6c62aa22b17c5be31136666386e0aa27b766a7af652438d4df85368d" }, "compression-thumb.webp": { "source": "Matching frames at 2 seconds from the corresponding delivery pair", "encoding": "640 \u00d7 360 split preview, WebP quality 72; thumbnails are not comparison inputs", "bytes": 55438, "sha256": "8220ad6e523900199f86b05f3c950b835142a4f9e59fc2cdf198bcb348a1a111" } }, "methodology": [ "One generation per model per example; same reference and exact prompt within each model comparison.", "Image delivery is lossless WebP. Video delivery applies the same CRF 18 encoding capped at 5 Mbps and trims the first 5 seconds; original generated URLs and hashes are retained below.", "Seedance and H3 outputs are independent actions; they share a playback clock, not identical motion timing.", "H3 Max 1080P is the provider\u2019s latent-refined output from 768P, not a claim of native 1080p generation.", "Resolution and compression examples are derived from the same Seedance original. All variants retain 120 frames at 24 fps and omit audio. 120p is 214 \u00d7 120; it is displayed enlarged by the browser, without AI upscaling.", "The early prompt is an illustrative draft, not a separately generated result. The revised prompt is the exact product edit instruction." ] }