Gallery
50 creations
POSE REFERENCE: use the input image strictly as a reference for body pose, limb positions, head tilt, framing, and camera angle. The subject stays in the same position in frame, same stance, same gesture, same depth. Description: Patrick Bateman from American Psycho, sharp navy 90s pinstripe suit, slicked-back dark hair, cold psychopathic stare, NYC apartment. Follow the description above: - If it names a different character, person, or entity to become, FULLY transform the subject into them — replace the original face, hair, body, skin, and clothing with the canonical look of that character. Do not bleed the original face through. - If it asks to keep, preserve, or retain the original subject, honor that. - If it names a specific scene, location, or setting, use that as the background. - If it doesn't name a scene AND the change is a full character transformation, replace the background with this default setting: inside a sleek modern luxury Manhattan penthouse apartment with floor-to-ceiling windows showing the NYC skyline at dusk, warm interior lighting, cinematic depth of field. - If it asks to keep the original background, kitchen, room, or scene, honor that. Photorealistic, cinematic detail, no text or watermarks.
May 25, 10:06 PM
POSE REFERENCE: use the input image strictly as a reference for body pose, limb positions, head tilt, framing, and camera angle. The subject stays in the same position in frame, same stance, same gesture, same depth. Description: Keep. Follow the description above: - If it names a different character, person, or entity to become, FULLY transform the subject into them — replace the original face, hair, body, skin, and clothing with the canonical look of that character. Do not bleed the original face through. - If it asks to keep, preserve, or retain the original subject, honor that. - If it names a specific scene, location, or setting, use that as the background. - If it doesn't name a scene AND the change is a full character transformation, replace the background with this default setting: inside a sleek modern luxury Manhattan penthouse apartment with floor-to-ceiling windows showing the NYC skyline at dusk, warm interior lighting, cinematic depth of field. - If it asks to keep the original background, kitchen, room, or scene, honor that. Photorealistic, cinematic detail, no text or watermarks. Use the second image as the visual reference for the character's appearance, identity, and signature features — match its colors, costume details, and facial structure.
May 25, 09:54 PM
POSE REFERENCE ONLY: use the input image strictly as a reference for body pose, limb positions, head tilt, framing, and camera angle. Do NOT preserve the original subject's identity, face, hair, body type, skin, or clothing. Do NOT preserve any of the original background, walls, furniture, or props. The subject IS Keep the older lady in her kitchen — the canonical, recognizable look of this character / person, not someone dressed up as them. Render their canonical face, body, hairstyle, costume, and identity faithfully. Lock the body pose, limb positions, head tilt, camera angle, and framing EXACTLY as shown in the input. The new character occupies the same position in frame, same stance, same gesture, same depth. Background: completely replace the original photo's background and lighting. If the description above names a specific scene or location, use that. Otherwise, use this default setting: inside a sleek modern luxury Manhattan penthouse apartment with floor-to-ceiling windows showing the NYC skyline at dusk, warm interior lighting, cinematic depth of field. Photorealistic where the character allows; cinematic detail; no text or watermarks.
May 25, 09:49 PM
POSE REFERENCE ONLY: use the input image strictly as a reference for body pose, limb positions, head tilt, framing, and camera angle. Do NOT preserve the original subject's identity, face, hair, body type, skin, or clothing. Do NOT preserve any of the original background, walls, furniture, or props. The subject IS Keep the older lady in her kitchen — the canonical, recognizable look of this character / person, not someone dressed up as them. Render their canonical face, body, hairstyle, costume, and identity faithfully. Lock the body pose, limb positions, head tilt, camera angle, and framing EXACTLY as shown in the input. The new character occupies the same position in frame, same stance, same gesture, same depth. Background: completely replace the original photo's background and lighting. If the description above names a specific scene or location, use that. Otherwise, use this default setting: inside a sleek modern luxury Manhattan penthouse apartment with floor-to-ceiling windows showing the NYC skyline at dusk, warm interior lighting, cinematic depth of field. Photorealistic where the character allows; cinematic detail; no text or watermarks.
May 25, 09:48 PM
POSE REFERENCE ONLY: use the input image strictly as a reference for body pose, limb positions, head tilt, framing, and camera angle. Do NOT preserve the original subject's identity, face, hair, body type, skin, or clothing. Do NOT preserve any of the original background, walls, furniture, or props. The subject IS Keep the older lady and the scene — the canonical, recognizable look of this character / person, not someone dressed up as them. Render their canonical face, body, hairstyle, costume, and identity faithfully. Lock the body pose, limb positions, head tilt, camera angle, and framing EXACTLY as shown in the input. The new character occupies the same position in frame, same stance, same gesture, same depth. Background: completely replace the original photo's background and lighting. If the description above names a specific scene or location, use that. Otherwise, use this default setting: inside a sleek modern luxury Manhattan penthouse apartment with floor-to-ceiling windows showing the NYC skyline at dusk, warm interior lighting, cinematic depth of field. Photorealistic where the character allows; cinematic detail; no text or watermarks. Use the second image as the visual reference for the character's appearance, identity, and signature features — match its colors, costume details, and facial structure.
May 25, 09:47 PM
POSE REFERENCE ONLY: use the input image strictly as a reference for body pose, limb positions, head tilt, framing, and camera angle. Do NOT preserve the original subject's identity, face, hair, body type, skin, or clothing. The subject IS a fearsome Sith Lord in a glossy black helmet with breathing-apparatus vents, all-black armored suit, long flowing black cape, gloved hands — the canonical, recognizable look of this character / person, not someone dressed up as them. Render their canonical face, body, hairstyle, costume, and identity faithfully. Lock the body pose, limb positions, head tilt, camera angle, and framing EXACTLY as shown in the input. The new character occupies the same position in frame, same stance, same gesture, same depth. If the description names a scene or location, replace the background to match it; otherwise keep the original background. Subject position in frame must not move. Photorealistic where the character allows; cinematic detail; no text or watermarks.
May 25, 07:52 PM
Keep the EXACT same composition, framing, camera angle, body pose, limb positions, and depth of the subject in frame as the input image. Transform the character into: Darth Vader. Replace appearance, costume, skin/fur/material, and facial features to match the new character — but the silhouette, stance, and action must be identical to the input. If the description names a scene or location, transform the background to match it; if no scene is named, keep the original background. The camera angle and subject position in frame must not move. Photorealistic where the new character allows; cinematic detail; no text or watermarks.
May 24, 09:23 PM
Keep the EXACT same composition, framing, camera angle, body pose, limb positions, and depth of the subject in frame as the input image. Transform the character into: Darth Vader. Replace appearance, costume, skin/fur/material, and facial features to match the new character — but the silhouette, stance, and action must be identical to the input. If the description names a scene or location, transform the background to match it; if no scene is named, keep the original background. The camera angle and subject position in frame must not move. Photorealistic where the new character allows; cinematic detail; no text or watermarks.
May 24, 09:23 PM
compose intro plus motion
May 24, 09:20 PM
Hot girl in luxury apartment with floor to ceiling windows natural lighting
May 24, 09:12 PM
Keep the EXACT same composition, framing, camera angle, body pose, limb positions, and depth of the subject in frame as the input image. Transform the character into: 24-year-old blonde woman with long wavy honey-blonde hair past her shoulders, smooth tan skin, blue eyes, in a luxury Manhattan penthouse with floor-to-ceiling windows showing a sunset cityscape, soft natural golden-hour lighting. Replace appearance, costume, skin/fur/material, and facial features to match the new character — but the silhouette, stance, and action must be identical to the input. If the description names a scene or location, transform the background to match it; if no scene is named, keep the original background. The camera angle and subject position in frame must not move. Photorealistic where the new character allows; cinematic detail; no text or watermarks.
May 24, 09:12 PM
Keep the EXACT same composition, framing, camera angle, body pose, limb positions, and depth of the subject in frame as the input image. Transform the character into: 24-year-old blonde woman with long wavy honey-blonde hair past her shoulders, smooth tan skin, blue eyes, in a luxury Manhattan penthouse with floor-to-ceiling windows showing a sunset cityscape, soft natural golden-hour lighting. Replace appearance, costume, skin/fur/material, and facial features to match the new character — but the silhouette, stance, and action must be identical to the input. If the description names a scene or location, transform the background to match it; if no scene is named, keep the original background. The camera angle and subject position in frame must not move. Photorealistic where the new character allows; cinematic detail; no text or watermarks.
May 24, 09:11 PM
Keep the EXACT same composition, framing, camera angle, body pose, limb positions, and depth of the subject in frame as the input image. Transform the character into: Hot blonde girl, luxury apartment with floor to ceiling windows natural lighting. Replace appearance, costume, skin/fur/material, and facial features to match the new character — but the silhouette, stance, and action must be identical to the input. If the description names a scene or location, transform the background to match it; if no scene is named, keep the original background. The camera angle and subject position in frame must not move. Photorealistic where the new character allows; cinematic detail; no text or watermarks.
May 24, 09:10 PM
Keep the EXACT same composition, framing, camera angle, body pose, limb positions, and depth of the subject in frame as the input image. Transform the character into: Hot blonde girl, luxury apartment with floor to ceiling windows natural lighting. Replace appearance, costume, skin/fur/material, and facial features to match the new character — but the silhouette, stance, and action must be identical to the input. If the description names a scene or location, transform the background to match it; if no scene is named, keep the original background. The camera angle and subject position in frame must not move. Photorealistic where the new character allows; cinematic detail; no text or watermarks.
May 24, 09:07 PM
Keep the EXACT same composition, framing, camera angle, body pose, limb positions, and depth of the subject in frame as the input image. Transform the character into: Hot girl in luxury apartment with floor to ceiling windows natural lighting. Replace appearance, costume, skin/fur/material, and facial features to match the new character — but the silhouette, stance, and action must be identical to the input. If the description names a scene or location, transform the background to match it; if no scene is named, keep the original background. The camera angle and subject position in frame must not move. Photorealistic where the new character allows; cinematic detail; no text or watermarks.
May 24, 09:06 PM
Keep the EXACT same composition, framing, camera angle, body pose, limb positions, and depth of the subject in frame as the input image. Transform the character into: Keanu reeves. Replace appearance, costume, skin/fur/material, and facial features to match the new character — but the silhouette, stance, and action must be identical to the input. If the description names a scene or location, transform the background to match it; if no scene is named, keep the original background. The camera angle and subject position in frame must not move. Photorealistic where the new character allows; cinematic detail; no text or watermarks. Use the second image as the visual reference for the character's appearance, identity, and signature features — match its colors, costume details, and facial structure.
May 23, 11:15 PM
Keep the EXACT same composition, framing, camera angle, body pose, limb positions, and depth of the subject in frame as the input image. Transform the character into: Keanu reeves. Replace appearance, costume, skin/fur/material, and facial features to match the new character — but the silhouette, stance, and action must be identical to the input. If the description names a scene or location, transform the background to match it; if no scene is named, keep the original background. The camera angle and subject position in frame must not move. Photorealistic where the new character allows; cinematic detail; no text or watermarks. Use the second image as the visual reference for the character's appearance, identity, and signature features — match its colors, costume details, and facial structure.
May 23, 11:08 PM
Keep the EXACT same composition, framing, camera angle, body pose, limb positions, and depth of the subject in frame as the input image. Transform the character into: a grimy fantasy goblin with mottled green skin and sharp yellow teeth. Replace appearance entirely.
May 23, 09:57 PM
Keep the EXACT same composition, framing, camera angle, body pose, limb positions, and depth of the subject in frame as the input image. Transform the character into: a grimy fantasy goblin with mottled green skin and sharp yellow teeth. Replace appearance entirely.
May 23, 09:30 PM
Storm trooper from Star Wars
May 23, 08:43 PM
Keep the EXACT same composition, framing, camera angle, body pose, limb positions, and depth of the subject in frame as the input image. Transform the character into: Storm trooper from Star Wars. Replace appearance, costume, skin/fur/material, and facial features to match the new character — but the silhouette, stance, and action must be identical to the input. If the description names a scene or location, transform the background to match it; if no scene is named, keep the original background. The camera angle and subject position in frame must not move. Photorealistic where the new character allows; cinematic detail; no text or watermarks.
May 23, 08:40 PM
Keep the EXACT same composition, framing, camera angle, body pose, limb positions, and depth of the subject in frame as the input image. Transform the character into: Darth Vader. Replace appearance, costume, skin/fur/material, and facial features to match the new character — but the silhouette, stance, and action must be identical to the input. If the description names a scene or location, transform the background to match it; if no scene is named, keep the original background. The camera angle and subject position in frame must not move. Photorealistic where the new character allows; cinematic detail; no text or watermarks.
May 23, 10:01 AM
compose intro plus motion
May 23, 09:55 AM
compose intro plus motion
May 23, 09:44 AM
Keep the EXACT same composition, framing, camera angle, body pose, limb positions, and depth of the subject in frame as the input image. Transform the character into: Darth Vader. Replace appearance, costume, skin/fur/material, and facial features to match the new character — but the silhouette, stance, and action must be identical to the input. If the description names a scene or location, transform the background to match it; if no scene is named, keep the original background. The camera angle and subject position in frame must not move. Photorealistic where the new character allows; cinematic detail; no text or watermarks.
May 23, 09:33 AM
Keep the EXACT same composition, framing, camera angle, body pose, limb positions, and depth of the subject in frame as the input image. Transform the character into: Darth Vader. Replace appearance, costume, skin/fur/material, and facial features to match the new character — but the silhouette, stance, and action must be identical to the input. If the description names a scene or location, transform the background to match it; if no scene is named, keep the original background. The camera angle and subject position in frame must not move. Photorealistic where the new character allows; cinematic detail; no text or watermarks.
May 22, 11:03 PM
Keep the EXACT same composition, framing, camera angle, body pose, limb positions, and depth of the subject in frame as the input image. Transform the character into: Darth Vader. Replace appearance, costume, skin/fur/material, and facial features to match the new character — but the silhouette, stance, and action must be identical to the input. If the description names a scene or location, transform the background to match it; if no scene is named, keep the original background. The camera angle and subject position in frame must not move. Photorealistic where the new character allows; cinematic detail; no text or watermarks.
May 22, 10:35 PM
Keep the EXACT same composition, framing, camera angle, body pose, limb positions, and depth of the subject in frame as the input image. Transform the character into: Darth Vader. Replace appearance, costume, skin/fur/material, and facial features to match the new character — but the silhouette, stance, and action must be identical to the input. If the description names a scene or location, transform the background to match it; if no scene is named, keep the original background. The camera angle and subject position in frame must not move. Photorealistic where the new character allows; cinematic detail; no text or watermarks.
May 22, 10:12 PM
Natural blink. Subtle breath. Surrounding crowd shifts and reacts subtly in the background. Slow push in.
May 14, 06:36 AM
![Generate a 16:9 horizontal landscape image. The output must be wider than tall.
Use the JSON below as the structured constraints for the image. Each field is a directive, not a suggestion.
```json
{
"task": "Generate a candid MLB live TV broadcast cutaway frame in 16:9 horizontal landscape format. The reference photo's subject is the fan in the stands. Aspect: 16:9 landscape only.",
"subject_anchor": {
"source": "uploaded reference image",
"preserve": [
"face shape",
"eyes (shape, spacing, color)",
"nose bridge",
"jawline",
"lips",
"ears",
"ethnicity",
"age",
"hair color and style",
"facial hair",
"glasses if present",
"distinctive marks (scars, moles, freckles)",
"skin texture and pores"
],
"do_not": [
"beautify",
"slim face",
"enlarge eyes",
"sharpen jawline",
"westernize",
"age",
"smooth skin"
]
},
"composition": {
"aspect_ratio": "16:9 landscape",
"shot_type": "medium-wide telephoto broadcast cutaway",
"face_size_in_frame": "10-18% of frame height — subject is one of many fans, NOT a portrait",
"subject_position": "right-third or center-third, off-center",
"visible_in_frame": "5-10 fans total including subject"
},
"camera": {
"position": "dugout-level robotic camera on the home side",
"lens": "200mm telephoto",
"direction": "pointing UP and ACROSS into the lower-bowl seating",
"playing_surface_visible": false,
"reasoning": "the camera is on the field side; the field is BEHIND the camera and is not in the frame"
},
"matchup": {
"home": "San Diego Padres (SD)",
"away": "Los Angeles Dodgers (LAD)",
"league": "MLB"
},
"subject_state": {
"seated_in": "lower-bowl seating",
"wearing": "authentic San Diego Padres jersey or fitted cap",
"team_colors": {
"primary_hex": "#2F241D",
"accent_hex": "#FFC425"
},
"props": "stadium drink (large soda or beer cup), foam finger, cheering stick, snack tray on lap",
"pose_mood": "engaged fan watching their team in a close game — mid-reaction to the play, mid-sip, or mid-chat with the friend beside them. Relaxed shoulders, natural posture",
"eyeline": "tracking the play below the camera lens — eyes look down and forward, NOT at the lens",
"expression": "natural facial tension and unforced emotion. Could be a slight genuine smile after a good play, lips parted in mild reaction, or relaxed engaged neutral. NOT a posed photo-smile, NOT dour or blank",
"camera_awareness": "unaware — caught accidentally on the broadcast cutaway"
},
"surrounding_fans": {
"orientation": "every visible fan faces the same direction as the subject (toward the playing surface)",
"density": "packed row, plus rows in front and behind at telephoto-compressed depth",
"scale_match": "subject's head and shoulders the same apparent size as the fans flanking them"
},
"background_behind_subject": [
"more rows of seated fans rising up the bowl",
"sponsor signage and structural railings",
"upper-deck overhang slightly out of focus",
"LED ribbon boards softly out of focus"
],
"forbidden_in_background": "field, grass, dirt, bases, players, pitcher's mound",
"scoreboard_overlay": {
"style": "ESPN / MLB Network broadcast graphic, composited at top of frame",
"position": "upper-left or upper-right corner",
"content": "\"SD\" with plausible score above \"LAD\" with score, plus inning, outs/strikes, base-runner diamond (e.g. \"TOP 6\", 2 outs, runner on 1st)"
},
"lighting": {
"scene_lighting": "warm late-afternoon to golden-hour stadium light — soft, flattering, gentle falloff across faces",
"subject_lighting": "matches the scene exactly — same key direction, color temperature, intensity as the fan beside them. Catches the subject favorably with natural catchlight in the eyes from the field lights",
"shadows": "single scene-consistent direction across all faces (no double shadows on the subject)",
"anti": "do NOT carry studio / portrait / reference-photo lighting onto the subject"
},
"render": {
"look": "live broadcast video frame, grounded photographic, flattering but believable",
"quality": "slightly blurry live-stream, mild compression noise, broadcast video softness",
"skin_quality": "preserve real skin texture and fine details — magazine-ready realism while maintaining natural human texture. NOT airbrushed, NOT smoothed, but ALSO NOT exaggerating blemishes or under-eye shadows",
"color_grade": "broadcast-camera color, NOT cinematic, NOT movie-poster, NOT dramatic",
"grain_match": "subject and surrounding fans share the same white balance and grain — no compositing artifacts"
},
"negative": [
"selfie composition",
"portrait headshot",
"face fills the frame",
"eye contact with camera",
"vertical 9:16 framing",
"pasted-in or floating subject",
"cinematic grading",
"beautified influencer face"
]
}
```
Final reminder: 16:9 landscape orientation, MLB broadcast cutaway look, subject is one of many forward-facing fans in the row.](https://cdn.gluely.ai/outputs/1778740558756-y2qul7bclc.png)
Generate a 16:9 horizontal landscape image. The output must be wider than tall. Use the JSON below as the structured constraints for the image. Each field is a directive, not a suggestion. ```json { "task": "Generate a candid MLB live TV broadcast cutaway frame in 16:9 horizontal landscape format. The reference photo's subject is the fan in the stands. Aspect: 16:9 landscape only.", "subject_anchor": { "source": "uploaded reference image", "preserve": [ "face shape", "eyes (shape, spacing, color)", "nose bridge", "jawline", "lips", "ears", "ethnicity", "age", "hair color and style", "facial hair", "glasses if present", "distinctive marks (scars, moles, freckles)", "skin texture and pores" ], "do_not": [ "beautify", "slim face", "enlarge eyes", "sharpen jawline", "westernize", "age", "smooth skin" ] }, "composition": { "aspect_ratio": "16:9 landscape", "shot_type": "medium-wide telephoto broadcast cutaway", "face_size_in_frame": "10-18% of frame height — subject is one of many fans, NOT a portrait", "subject_position": "right-third or center-third, off-center", "visible_in_frame": "5-10 fans total including subject" }, "camera": { "position": "dugout-level robotic camera on the home side", "lens": "200mm telephoto", "direction": "pointing UP and ACROSS into the lower-bowl seating", "playing_surface_visible": false, "reasoning": "the camera is on the field side; the field is BEHIND the camera and is not in the frame" }, "matchup": { "home": "San Diego Padres (SD)", "away": "Los Angeles Dodgers (LAD)", "league": "MLB" }, "subject_state": { "seated_in": "lower-bowl seating", "wearing": "authentic San Diego Padres jersey or fitted cap", "team_colors": { "primary_hex": "#2F241D", "accent_hex": "#FFC425" }, "props": "stadium drink (large soda or beer cup), foam finger, cheering stick, snack tray on lap", "pose_mood": "engaged fan watching their team in a close game — mid-reaction to the play, mid-sip, or mid-chat with the friend beside them. Relaxed shoulders, natural posture", "eyeline": "tracking the play below the camera lens — eyes look down and forward, NOT at the lens", "expression": "natural facial tension and unforced emotion. Could be a slight genuine smile after a good play, lips parted in mild reaction, or relaxed engaged neutral. NOT a posed photo-smile, NOT dour or blank", "camera_awareness": "unaware — caught accidentally on the broadcast cutaway" }, "surrounding_fans": { "orientation": "every visible fan faces the same direction as the subject (toward the playing surface)", "density": "packed row, plus rows in front and behind at telephoto-compressed depth", "scale_match": "subject's head and shoulders the same apparent size as the fans flanking them" }, "background_behind_subject": [ "more rows of seated fans rising up the bowl", "sponsor signage and structural railings", "upper-deck overhang slightly out of focus", "LED ribbon boards softly out of focus" ], "forbidden_in_background": "field, grass, dirt, bases, players, pitcher's mound", "scoreboard_overlay": { "style": "ESPN / MLB Network broadcast graphic, composited at top of frame", "position": "upper-left or upper-right corner", "content": "\"SD\" with plausible score above \"LAD\" with score, plus inning, outs/strikes, base-runner diamond (e.g. \"TOP 6\", 2 outs, runner on 1st)" }, "lighting": { "scene_lighting": "warm late-afternoon to golden-hour stadium light — soft, flattering, gentle falloff across faces", "subject_lighting": "matches the scene exactly — same key direction, color temperature, intensity as the fan beside them. Catches the subject favorably with natural catchlight in the eyes from the field lights", "shadows": "single scene-consistent direction across all faces (no double shadows on the subject)", "anti": "do NOT carry studio / portrait / reference-photo lighting onto the subject" }, "render": { "look": "live broadcast video frame, grounded photographic, flattering but believable", "quality": "slightly blurry live-stream, mild compression noise, broadcast video softness", "skin_quality": "preserve real skin texture and fine details — magazine-ready realism while maintaining natural human texture. NOT airbrushed, NOT smoothed, but ALSO NOT exaggerating blemishes or under-eye shadows", "color_grade": "broadcast-camera color, NOT cinematic, NOT movie-poster, NOT dramatic", "grain_match": "subject and surrounding fans share the same white balance and grain — no compositing artifacts" }, "negative": [ "selfie composition", "portrait headshot", "face fills the frame", "eye contact with camera", "vertical 9:16 framing", "pasted-in or floating subject", "cinematic grading", "beautified influencer face" ] } ``` Final reminder: 16:9 landscape orientation, MLB broadcast cutaway look, subject is one of many forward-facing fans in the row.
May 14, 06:35 AM
Natural blink. Subtle breath. Surrounding crowd shifts and reacts subtly in the background. Slow push in.
May 14, 01:29 AM
Generate a HORIZONTAL 16:9 LANDSCAPE image. THE OUTPUT MUST BE 16:9 LANDSCAPE FORMAT — wider than it is tall. ABSOLUTELY DO NOT produce a vertical, portrait, square, or 9:16 image. The final canvas dimensions must be wider than tall. This is a non-negotiable geometric requirement. Aspect ratio: 16:9 landscape. The uploaded reference image is provided for IDENTITY ONLY. Use it ONLY for facial features — identical eye shape, eye spacing, nose bridge contour, jawline angle, lip proportions, ear shape, ethnicity, age, body type, hair color, hair length, and distinctive features (scars, moles, freckles, facial hair, glasses if present). Preserve facial geometry exactly; do NOT redesign or beautify the face. DISCARD the lighting from the reference photo entirely. DISCARD the white balance from the reference photo. DISCARD the color cast from the reference photo. The reference photo's lighting environment does NOT exist in the output — only the stadium scene's lighting exists. The face is re-lit by the stadium scene's golden-hour light from the same direction as every other fan in the frame. Skin tone is rendered under the stadium's warm color temperature, not the reference photo's color temperature. Generate the image as a real ESPN / Sportsnet NHL live broadcast cutaway frame — the "fan behind the glass" shot a director cuts to between whistles. Game: Anaheim Ducks vs Vegas Golden Knights. Bright overhead rink lighting. The camera is the bench-side or goal-cam broadcast camera, telephoto pointed across the rink and up into the glass-side lower-bowl seating to find a fan reaction. CAMERA POSITION (mandatory): the TV broadcast camera is positioned at FIELD LEVEL — on the field, in the dugout, behind home plate, or at the baseline — looking UP and ACROSS into the lower-bowl spectator seating. The 200mm telephoto lens is pointed AT the stands FROM the field. The field of play is BEHIND the camera and is NOT visible in the frame. Do NOT show the field, the grass, the dirt, the bases, or any players in the shot. WHAT IS BEHIND THE SUBJECT: more spectator seating, more fans in their seats, the next tier of stands rising up, structural elements (railings, sponsor signage, support beams). NOT the field. NOT the grass. The background is the BACK of the stadium bowl from the camera's POV. SUBJECT ORIENTATION: the subject is facing the camera direction — because the field is in front of them and the camera is between them and the field. They are NOT turned sideways, NOT looking over their shoulder, NOT facing away. Their body is oriented toward the camera. EYELINE: even though the subject is facing the camera direction, their eyes are looking DOWN AND FORWARD AT THE PLAY ON THE FIELD (which is below and behind the camera lens). Eyes do NOT meet the lens. Their gaze passes over or under the camera. Think: someone watching the game intently, lens just happens to be in their general line of sight. ADJACENT FANS: every other fan visible in the frame is also facing the camera direction (because they're all facing the field). A whole row of forward-facing spectators, not a circle around the subject. (Hockey-specific camera note: the camera is on the OPPOSITE side of the rink from this fan section, looking ACROSS the ice into the stands. The plexiglass surrounding the rink is between the camera and the fans, so a faint reflection or sponsor-decal ghosting may appear in the lower portion of the frame — but the ice itself is BELOW the camera's sightline and is NOT visible in the broadcast crop.) WHO IS IN THE FRAME: - The reference person is seated in the glass-side lower bowl, in the middle of a row of fans. Multiple other fans flank them on both sides; partial heads/shoulders of rows in front and behind. Telephoto compression flattens depth. - Wearing an authentic Ducks jersey or knit beanie (primary #F47A38, accent #B9975B). - Holding a beer cup or hot dog. Maybe a hand-painted sign visible. - Natural relaxed pose — watching the game, mid-chat, or settled into the seat. WHAT IS BEHIND THE SUBJECT: - More rows of seated fans rising up the bowl. - Section signage, aisle railings, suite-level windows. - Far arena structure slightly out of focus. - ABSOLUTELY NO open ice, skaters, goal nets, or referees in frame — the ice surface is below the camera's sightline. SCOREBOARD OVERLAY (composited broadcast graphic, locked to top of 16:9 frame): - ESPN / Sportsnet NHL banner pinned upper-left or upper-right. Clearly shows "ANA" score above "VGK" score, period (e.g. "P2"), game clock (e.g. "12:08"). This is a TELEPHOTO BROADCAST CAPTURE — wide shot showing 5-10 people in the row, NOT a portrait, NOT a selfie. The subject's face occupies 10-18% of the frame at most. Medium-wide shot scale: heads + shoulders + chest of the subject AND multiple flanking fans clearly visible. The subject is positioned off-center (right-third or center-third). Imperfect candid framing slightly off-center. 16:9 horizontal broadcast composition. Avoid: face-fill portrait, centered headshot, vertical/9:16 framing, anyone holding a phone-camera-distance shot of the subject. Slightly blurry live-stream quality, subtle compression noise, mild broadcast video softness, slight motion blur in the surrounding crowd, natural skin texture with visible pores, imperfect lighting, real broadcast-camera color grade. Match the subject scale and shadow pattern to the existing scene. Correct scale and perspective with believable grounding: the subject is seated in the row at the same camera distance as the fans flanking them; head and shoulders the same apparent size as the neighboring fans' heads and shoulders. Use the same style of lighting, composition, white balance, and film grain across the entire frame — the subject and every surrounding fan are lit by the SAME stadium lighting from the SAME direction with the SAME color temperature. Single, scene-consistent shadow direction across all faces (no double shadows, no contradictory shadows on the inserted subject vs. the rest of the scene). Anchor realism by specifying a grounded photographic look — natural broadcast-camera lighting, believable detail, NO cinematic grading, NO movie-poster styling, NO dramatic color treatment on the subject's face. Subjects must not look pasted or floating. Avoid cutout / halo / overlay edges around the subject, white-balance/grain mismatch between subject and scene, and any "pasted-in" or "photoshopped" appearance. Lighting on the subject is warm late-afternoon to golden-hour stadium light — soft, flattering, with gentle falloff across the face and dimensional highlights from the stadium lights. Natural catchlight in the eyes from the field lights. The camera's slightly-below-eye angle (field-up POV) gives a naturally flattering perspective on the subject's jawline and posture. Preserve fine facial detail and authentic skin texture — magazine-ready realism while maintaining human imperfections. The subject should look like the best version of themselves caught on broadcast, not a stylized version and not an unflattering candid. The subject is UNAWARE of the camera. They are watching the play, mid-reaction. Natural facial tension and unforced emotion — could be a slight genuine smile after a good play, lips parted in mild reaction, or relaxed neutral while taking in the game. NOT a posed photo-smile, NOT "say cheese", but also NOT a dour blank stare — match whatever a real engaged fan would look like watching their team in a close game. Eyes track the play below the camera lens — not eye contact. NO turning toward the camera. Do NOT reshape the subject's face — no enlarged eyes, no slimmed or sharpened jaw, no altered nose, no idol-like restyling. Do NOT smooth or airbrush the skin — preserve real skin texture, pores, fine details, faint stubble or baby hairs that are naturally present. Do NOT apply makeup that wasn't in the reference photo. The goal is the same person looking like a real fan in the stands, photographed in flattering golden-hour stadium light — believable and recognizable as themselves, but not stylized into an AI-influencer or beauty-filter version. FINAL REMINDER — output dimensions MUST be 16:9 LANDSCAPE (e.g. 1920×1080, wider than tall). Do not produce vertical 9:16. Do not produce square 1:1. Crop and compose the scene so a wide horizontal frame is filled with the stadium and surrounding fans, not a vertical portrait of the subject.
May 14, 01:27 AM