{"observation":{"id":"e20b4e01-3872-493d-893a-03a97c39e2a4","tool":"leonardo-ai","tool_name":"Leonardo AI","criterion":"visual-consistency-no-distortion","criterion_name":"Visual Consistency (No Distortion)","criterion_definition":"How well the output preserves faces, objects, and scene structure without warping or other visual breakage.","criterion_evidence_type":"transformation","criterion_rank_role":"decisive","criterion_rank_role_reason":"A single-image video generator must preserve the source subject and scene structure; distortion means the tool is failing at the main job. (3 of 3 judges)","scenario":"3d-rendered-street-scene-image-with-forward-dolly-prompt","scenario_name":"3D rendered street scene image with forward dolly prompt","group_tag":"image-to-cinematic-video","scenario_description":"A rendered 3D street scene with multiple characters and environment detail, tested using a forward-dolly cinematic prompt with crowd motion, cart movement, atmospheric haze, and sunset lighting.","modality":"image","input_text":null,"input_artifact_refs":[{"alt":null,"url":"https://d3epheqghktydj.cloudfront.net/luma-ai-dream-machine-3d-image-input-2-8bec799b14bc.png","role":"input"}],"stresses":["Complex multi-subject scene animation","Character consistency in crowd motion","Camera movement stability during forward dolly","Environmental motion such as clouds, trees, and birds","Prompt adherence across foreground and background elements"],"verdict":"struggled","score":null,"score_total":null,"note":"The tool introduces noticeable facial distortion in characters, showing that it does not reliably preserve faces and human structure in 3D scenes.","evidence_state":"verified","source":null,"artifacts":[{"url":"https://d3epheqghktydj.cloudfront.net/leonardo-ai-3d-output-b445b62c2610.mp4","role":"output","alt":null}],"run_id":"b8778c3c-4fb1-48b2-bf35-4e08a3b82d8e","study_title":"Generate a cinematic AI video from a single image","study_kind":"generation","research_task":"86b94urgr","tested_at":null,"completeness":"input-and-output","input":{"state":"files","text":null,"files":[{"url":"https://d3epheqghktydj.cloudfront.net/luma-ai-dream-machine-3d-image-input-2-8bec799b14bc.png","filename":"luma-ai-dream-machine-3d-image-input-2-8bec799b14bc.png","alt":"3D rendered street scene image with forward dolly prompt","role":"input"}],"modality":"image","stresses":["Complex multi-subject scene animation","Character consistency in crowd motion","Camera movement stability during forward dolly","Environmental motion such as clouds, trees, and birds","Prompt adherence across foreground and background elements"]},"tool_page_slug":"leonardo-ai","tool_url":"https://aidemos.com/tools/leonardo-ai","permalink":"https://aidemos.com/evidence/e20b4e01-3872-493d-893a-03a97c39e2a4","api_url":"https://ai.aidemos.com/v1/observations/e20b4e01-3872-493d-893a-03a97c39e2a4"},"peers":[{"id":"04a73b4c-336e-4b43-9962-371922367429","tool":"google-flow","tool_name":"Google Flow","verdict":"worked","score":null,"score_total":null,"note":"The 3D output preserved character structure cleanly enough that the report says there was no visible distortion in characters.","artifact_count":1,"thumbnail":"https://d3epheqghktydj.cloudfront.net/google-flow-3d-output-478f5388b5c0.mp4","evidence_url":"https://aidemos.com/evidence/04a73b4c-336e-4b43-9962-371922367429"},{"id":"f5e96a23-d773-442e-82aa-0798dab8a10e","tool":"luma-ai-dream-machine","tool_name":"Luma AI Dream Machine","verdict":"worked","score":null,"score_total":null,"note":"The 3D run shows no visible distortion in either version, according to the report.","artifact_count":2,"thumbnail":"https://d3epheqghktydj.cloudfront.net/luma-ai-dream-machine-3d-output-version-1-691346969900.mp4","evidence_url":"https://aidemos.com/evidence/f5e96a23-d773-442e-82aa-0798dab8a10e"},{"id":"14777a46-a664-4ce7-802e-e88b84c57566","tool":"pika-labs","tool_name":"Pika Labs","verdict":"struggled","score":null,"score_total":null,"note":"The 3D output shows minor character distortion and slight inconsistency in complex areas, with the report calling out that the distortion runs throughout and the cart position fails to settle cleanly in the final frame.","artifact_count":1,"thumbnail":"https://d3epheqghktydj.cloudfront.net/pika-labs-3d-output-eb0cade0b853.mp4","evidence_url":"https://aidemos.com/evidence/14777a46-a664-4ce7-802e-e88b84c57566"},{"id":"0eebc77a-94e0-4e97-96e7-44fd48cbd60c","tool":"pixverse-ai","tool_name":"PixVerse AI","verdict":"struggled","score":null,"score_total":null,"note":"The 3D output shows noticeable facial distortion in characters and reduced clarity compared with the input, so structural fidelity is only partial.","artifact_count":1,"thumbnail":"https://d3epheqghktydj.cloudfront.net/pixverse-ai-pixverse-ai-3d-image-output-1fc102a5d3b2.mp4","evidence_url":"https://aidemos.com/evidence/0eebc77a-94e0-4e97-96e7-44fd48cbd60c"}],"other_criteria":[{"id":"8c90b8ba-23cf-4e77-bb7f-a8dd8bce8ec6","criterion":"motion-quality-realism","criterion_name":"Motion Quality & Realism","rank_role":"decisive","verdict":"worked","score":null,"score_total":null,"note":"The tool maintains smooth animation and consistent scene motion across multiple moving subjects.","artifact_count":1,"evidence_url":"https://aidemos.com/evidence/8c90b8ba-23cf-4e77-bb7f-a8dd8bce8ec6"},{"id":"293512d4-0fe0-47d3-a867-42336caaf9a4","criterion":"prompt-accuracy","criterion_name":"Prompt Accuracy","rank_role":"decisive","verdict":"struggled","score":null,"score_total":null,"note":"The scene animates, but the report says the requested prompt details are not properly reflected.","artifact_count":1,"evidence_url":"https://aidemos.com/evidence/293512d4-0fe0-47d3-a867-42336caaf9a4"}],"appears_in":[{"page_type":"ranking","slug":"image-to-video-generators","title":"Best AI Tools for Generating Cinematic Video from a Single Image","url":"https://aidemos.com/best/image-to-video-generators","binding":"run"},{"page_type":"use-case","slug":"generate-ai-video-from-image","title":"Generate a Cinematic AI Video from a Single Image","url":"https://aidemos.com/use-cases/generate-ai-video-from-image","binding":"run"}],"same_scenario":[{"id":"04a73b4c-336e-4b43-9962-371922367429","tool":"google-flow","tool_name":"Google Flow","verdict":"worked","score":null,"score_total":null,"note":"The 3D output preserved character structure cleanly enough that the report says there was no visible distortion in characters."},{"id":"f5e96a23-d773-442e-82aa-0798dab8a10e","tool":"luma-ai-dream-machine","tool_name":"Luma AI Dream Machine","verdict":"worked","score":null,"score_total":null,"note":"The 3D run shows no visible distortion in either version, according to the report."},{"id":"14777a46-a664-4ce7-802e-e88b84c57566","tool":"pika-labs","tool_name":"Pika Labs","verdict":"struggled","score":null,"score_total":null,"note":"The 3D output shows minor character distortion and slight inconsistency in complex areas, with the report calling out that the distortion runs throughout and the cart position fails to settle cleanly in the final frame."},{"id":"0eebc77a-94e0-4e97-96e7-44fd48cbd60c","tool":"pixverse-ai","tool_name":"PixVerse AI","verdict":"struggled","score":null,"score_total":null,"note":"The 3D output shows noticeable facial distortion in characters and reduced clarity compared with the input, so structural fidelity is only partial."}]}