{"observation":{"id":"7d21adab-f4e6-4f0a-9975-37fabe9dc1c2","tool":"gemini","tool_name":"Gemini","criterion":"expression-accuracy","criterion_name":"Expression accuracy","criterion_definition":"Whether the emotional tone and facial expression match what was explicitly prompted.","criterion_evidence_type":"transformation","criterion_rank_role":"decisive","criterion_rank_role_reason":"If the character’s intended emotion or facial expression does not match the prompt, the generated character is not being controlled reliably across scenes. (3 of 3 judges)","scenario":"three-quarter-face-portrait","scenario_name":"Three-quarter face portrait","group_tag":null,"scenario_description":"Three-quarter face reference image with medium-dark skin, tight curly hair in an updo, bindi, and a floral dress. The partial angle and softer lighting make it a harder reference than the frontal portrait and are meant to stress identity consistency.","modality":"image","input_text":null,"input_artifact_refs":[],"stresses":["Identity preservation with partial face angle","Hair texture and updo retention","Skin tone fidelity","Reference-image difficulty under softer lighting"],"verdict":"failed","score":null,"score_total":null,"note":"The tool can miss the requested emotion entirely, producing a neutral and emotionless face instead of the prompted angry and guarded look.","evidence_state":"verified","source":null,"artifacts":[{"url":"https://d3epheqghktydj.cloudfront.net/gemini-input-2-1c4e066a03c3.png","role":"input","alt":"Input 2.png"},{"url":"https://d3epheqghktydj.cloudfront.net/gemini-gemini-input2-interrogation-62c75d7431c4.png","role":"output","alt":null}],"run_id":"1dfb8fa4-f7f0-47dc-911b-7db3af467e9a","study_title":"Generate Consistent AI Characters Across Different Scenes and Poses","study_kind":"generation","research_task":"86b96df11","tested_at":null,"completeness":"input-and-output","input":{"state":"not-captured","text":null,"files":[],"modality":"image","stresses":["Identity preservation with partial face angle","Hair texture and updo retention","Skin tone fidelity","Reference-image difficulty under softer lighting"]},"tool_page_slug":"gemini","tool_url":"https://aidemos.com/tools/gemini","permalink":"https://aidemos.com/evidence/7d21adab-f4e6-4f0a-9975-37fabe9dc1c2","api_url":"https://ai.aidemos.com/v1/observations/7d21adab-f4e6-4f0a-9975-37fabe9dc1c2"},"peers":[{"id":"8b496d76-afd9-4658-b0a0-2b73533010ff","tool":"chatgpt","tool_name":"ChatGPT","verdict":"worked","score":null,"score_total":null,"note":"Delivers the requested cold, guarded, unsmiling expression with the exact vibe the prompt asked for.","artifact_count":1,"thumbnail":"https://d3epheqghktydj.cloudfront.net/chatgpt-chatgpt-input2-interrogation-c78a75b73445.png","evidence_url":"https://aidemos.com/evidence/8b496d76-afd9-4658-b0a0-2b73533010ff"},{"id":"9f6dd6f3-855d-4f0a-9535-a5219634595e","tool":"imagineart","tool_name":"ImagineArt","verdict":"worked","score":null,"score_total":null,"note":"The interrogation-room output preserves the reference expression well.","artifact_count":1,"thumbnail":"https://d3epheqghktydj.cloudfront.net/imagineart-imagineart-input2-interrogation-ce155eab95e8.jpg","evidence_url":"https://aidemos.com/evidence/9f6dd6f3-855d-4f0a-9535-a5219634595e"},{"id":"ca959663-ea8a-4b8b-9cee-60d17e3305d5","tool":"leonardo-ai","tool_name":"Leonardo AI","verdict":"failed","score":null,"score_total":null,"note":"The same neutral-expression failure repeats on a second reference, so the tool still misses the prompt's angry, guarded tone.","artifact_count":2,"thumbnail":"https://d3epheqghktydj.cloudfront.net/leonardo-ai-image-95c2a8e685d7.png","evidence_url":"https://aidemos.com/evidence/ca959663-ea8a-4b8b-9cee-60d17e3305d5"}],"other_criteria":[{"id":"84e483a7-bbbe-4f0b-8b0a-242be173f910","criterion":"accessory-and-detail-retention","criterion_name":"Accessory & detail retention","rank_role":"decisive","verdict":"worked","score":null,"score_total":null,"note":"A busy market scene can still preserve fine clothing and hair details, keeping the sari, blouse, hairstyle, and curly hair texture recognizable.","artifact_count":2,"evidence_url":"https://aidemos.com/evidence/84e483a7-bbbe-4f0b-8b0a-242be173f910"},{"id":"1a0bf38a-fcde-4dc7-977f-cf458af631b5","criterion":"identity-preservation","criterion_name":"Identity preservation","rank_role":"decisive","verdict":"worked","score":null,"score_total":null,"note":"In the crowded market scene, the character stays highly recognisable, preserving face shape, smile, eyebrows, and overall facial structure.","artifact_count":2,"evidence_url":"https://aidemos.com/evidence/1a0bf38a-fcde-4dc7-977f-cf458af631b5"},{"id":"384e897c-c4a6-4bc4-a19d-1afddd298650","criterion":"identity-preservation","criterion_name":"Identity preservation","rank_role":"decisive","verdict":"worked","score":null,"score_total":null,"note":"With a 3/4 reference, the interrogation-room output still preserves identity well, keeping face shape, skin tone, curly hair texture, eyebrows, and overall structure close to the input.","artifact_count":2,"evidence_url":"https://aidemos.com/evidence/384e897c-c4a6-4bc4-a19d-1afddd298650"},{"id":"835c9105-4bf9-4973-bb87-e95bbb06de11","criterion":"scene-compliance","criterion_name":"Scene compliance","rank_role":"decisive","verdict":"worked","score":null,"score_total":null,"note":"The crowded market prompt is followed strongly, with a busy outdoor background, realistic crowd density, market stalls, and a natural walking pose with tote bag.","artifact_count":2,"evidence_url":"https://aidemos.com/evidence/835c9105-4bf9-4973-bb87-e95bbb06de11"},{"id":"1d7eac2d-40a7-4ad1-b7d3-07add2186115","criterion":"scene-compliance","criterion_name":"Scene compliance","rank_role":"decisive","verdict":"worked","score":null,"score_total":null,"note":"The interrogation-room prompt is followed well, including the table, plain wall, overhead light, formal clothing, and clinical lighting.","artifact_count":2,"evidence_url":"https://aidemos.com/evidence/1d7eac2d-40a7-4ad1-b7d3-07add2186115"}],"appears_in":[{"page_type":"ranking","slug":"consistent-ai-characters","title":"Best AI Tools for Consistent AI Characters Across Scenes and Poses","url":"https://aidemos.com/best/consistent-ai-characters","binding":"run"}],"same_scenario":[{"id":"8b496d76-afd9-4658-b0a0-2b73533010ff","tool":"chatgpt","tool_name":"ChatGPT","verdict":"worked","score":null,"score_total":null,"note":"Delivers the requested cold, guarded, unsmiling expression with the exact vibe the prompt asked for."},{"id":"9f6dd6f3-855d-4f0a-9535-a5219634595e","tool":"imagineart","tool_name":"ImagineArt","verdict":"worked","score":null,"score_total":null,"note":"The interrogation-room output preserves the reference expression well."},{"id":"ca959663-ea8a-4b8b-9cee-60d17e3305d5","tool":"leonardo-ai","tool_name":"Leonardo AI","verdict":"failed","score":null,"score_total":null,"note":"The same neutral-expression failure repeats on a second reference, so the tool still misses the prompt's angry, guarded tone."}]}