از روی هر عکسی که دوست داری، یک فرمول تصویری حرفه‌ای بساز

Now use the extracted "transferable_style" and "final_generation_prompt" on my second uploaded image. IDENTITY LOCK: Preserve the exact identity of the person in my second image. Do not change: - facial identity - face shape - eyes - eyebrows - nose - lips - jawline - skin tone - age - gender - recognizable facial features Transfer ONLY: - pose - composition - camera angle - lighting - environment - color grading - photographic style - mood - visual atmosphere The final result must clearly look like the SAME PERSON from my second uploaded image, photographed in the visual style of the first reference image.

Analyze the uploaded reference video and reverse-engineer it into a highly detailed, production-ready video style recipe. IMPORTANT: Do NOT claim to recover the original script, prompt, or editing software used to create the video. Instead, reconstruct the visual and editing recipe as accurately as possible from what is visible and audible in the video. Analyze everything that materially affects the final look and feel, including: - subject(s) and main focus - setting / locations - shot types (wide, medium, close-up, etc.) - camera movement (static, pan, tilt, handheld, tracking, drone, etc.) - framing and composition per scene - pacing and editing rhythm - cut frequency and transition style - lighting direction and quality - color palette - color grading - texture and image quality - on-screen text or caption style, if present - sound design and music mood - voiceover or dialogue tone, if present - overall visual style - visual mood - aspect ratio - frame rate / motion feel (smooth, cinematic, snappy, etc.) Then return ONLY valid JSON in exactly this structure: { "video_analysis": { "subject": "", "setting_and_locations": "", "shot_types": "", "camera_movement": "", "framing_and_composition": "", "pacing_and_editing_rhythm": "", "transitions": "", "lighting": "", "color_palette": [], "color_grading": "", "texture_and_detail": "", "on_screen_text_style": "", "sound_and_music_mood": "", "voiceover_or_dialogue_tone": "", "overall_visual_style": "", "mood": "", "aspect_ratio": "", "motion_feel": "" }, "transferable_style": { "visual_style": "", "camera_recipe": "", "editing_recipe": "", "color_recipe": "", "sound_recipe": "", "details_to_preserve": [] }, "final_generation_prompt": "", "negative_prompt": [ "low quality", "shaky unintended motion", "inconsistent color grading", "jarring cuts", "distorted faces", "audio out of sync", "unnatural lighting", "wrong aspect ratio" ] } RULES: 1. Be extremely specific and visual. 2. Never use vague phrases like "dynamic editing" without explaining the actual cut rhythm and shot lengths. 3. Describe camera movement and shot progression scene by scene where possible. 4. The final_generation_prompt must combine all important information into one polished English prompt ready for video style generation or editing. 5. Keep identity-specific and content-specific details (people, dialogue, storyline) separate from the transferable visual/editing style whenever possible. 6. Do not invent details that are not visible or audible in the video. 7. If something cannot be determined reliably, write "unknown". 8. Output JSON only. 9. Do not add explanations before or after the JSON. 10. Make sure the JSON is syntactically valid and easy to copy.