Diffusion Video Editing

subject_definitions:
<Video 1> is the source video to be edited. Every visual element of <Video 1> --
the wood-panelled walls and their exact surfaces, the LED cove light in the ceiling, the low wooden coffee table with its tray and dark bowls, the dark armchairs, the beige sofa, the potted tree, the floor-to-ceiling glass doors and the camera's slow rightward pan -- defines the target video and must be reproduced unchanged.

summary:
[video editing] The target video is an edited version of <Video 1>. Every element
of the original shot is preserved: the same camera movement at the same speed,
the same geometry, the same objects in the same places. The only change is:
golden hour, a low warm sun through the glass.

retention_analysis:
<Video 1> (source video editing): fully_preserved - every surface, object and
material in <Video 1> stays exactly as it appears there, in the same place, with
the same shape and the same construction. Flat wall panels stay flat wall panels and every piece of furniture keeps its place and shape. The camera path and
framing are unchanged. Only the illumination, weather and atmosphere change.

detailed_description:
The target video is <Video 1> itself, relit. Every frame matches the
corresponding frame of <Video 1> in composition, geometry and camera position:
the same surfaces in the same places, the same objects at the same coordinates,
the same camera path at the same speed.
[Shot 1] Low golden sunlight enters through the glass doors and rakes across the wood-panelled wall in long warm bars, throwing long soft shadows from the furniture across the floor. Amber highlights sit on the table edge and the tray, and the sky outside is warm orange. Nothing is redesigned, restyled, added, removed or rearranged.
Existing surfaces keep their own materials and colours and simply respond to the
new light and conditions.

overall_soundscape:
The quiet interior ambience of the room.

non_diegetic_music:
None.

subject_definitions:
<Video 1> is the source video to be edited. Every visual element of <Video 1> --
the wood-panelled walls and their exact surfaces, the LED cove light in the ceiling, the low wooden coffee table with its tray and dark bowls, the dark armchairs, the beige sofa, the potted tree, the floor-to-ceiling glass doors and the camera's slow rightward pan -- defines the target video and must be reproduced unchanged.

summary:
[video editing] The target video is an edited version of <Video 1>. Every element
of the original shot is preserved: the same camera movement at the same speed,
the same geometry, the same objects in the same places. The only change is:
night, warm lamps glowing against a black sky outside.

retention_analysis:
<Video 1> (source video editing): fully_preserved - every surface, object and
material in <Video 1> stays exactly as it appears there, in the same place, with
the same shape and the same construction. Flat wall panels stay flat wall panels and every piece of furniture keeps its place and shape. The camera path and
framing are unchanged. Only the illumination, weather and atmosphere change.

detailed_description:
The target video is <Video 1> itself, relit. Every frame matches the
corresponding frame of <Video 1> in composition, geometry and camera position:
the same surfaces in the same places, the same objects at the same coordinates,
the same camera path at the same speed.
[Shot 1] It is night. Warm amber lamplight glows from within the room, pooling on the coffee table and the tray and falling away into deep shadow in the corners. The LED cove reads as a warm line against the ceiling. Beyond the glass doors the sky and terrace are black, with a few distant warm points of light. Nothing is redesigned, restyled, added, removed or rearranged.
Existing surfaces keep their own materials and colours and simply respond to the
new light and conditions.

overall_soundscape:
The quiet interior ambience of the room.

non_diegetic_music:
None.

subject_definitions:
<Video 1> is the source video to be edited. Every visual element of <Video 1> --
the wood-panelled walls and their exact surfaces, the LED cove light in the ceiling, the low wooden coffee table with its tray and dark bowls, the dark armchairs, the beige sofa, the potted tree, the floor-to-ceiling glass doors and the camera's slow rightward pan -- defines the target video and must be reproduced unchanged.

summary:
[video editing] The target video is an edited version of <Video 1>. Every element
of the original shot is preserved: the same camera movement at the same speed,
the same geometry, the same objects in the same places. The only change is:
heavy rain on the glass, dark grey overcast outside.

retention_analysis:
<Video 1> (source video editing): fully_preserved - every surface, object and
material in <Video 1> stays exactly as it appears there, in the same place, with
the same shape and the same construction. Flat wall panels stay flat wall panels and every piece of furniture keeps its place and shape. The camera path and
framing are unchanged. Only the illumination, weather and atmosphere change.

detailed_description:
The target video is <Video 1> itself, relit. Every frame matches the
corresponding frame of <Video 1> in composition, geometry and camera position:
the same surfaces in the same places, the same objects at the same coordinates,
the same camera path at the same speed.
[Shot 1] Heavy rain runs down the outside of the glass doors in broad streaks and beads, and the terrace beyond is wet and dark under a grey overcast sky. The interior light turns cool and flat, and wet reflected light moves faintly across the floor near the glass. Nothing is redesigned, restyled, added, removed or rearranged.
Existing surfaces keep their own materials and colours and simply respond to the
new light and conditions.

overall_soundscape:
Rain against glass and the muffled hiss of downpour.

non_diegetic_music:
None.

subject_definitions:
<Video 1> is the source video to be edited. Every visual element of <Video 1> --
the wood-panelled walls and their exact surfaces, the LED cove light in the ceiling, the low wooden coffee table with its tray and dark bowls, the dark armchairs, the beige sofa, the potted tree, the floor-to-ceiling glass doors and the camera's slow rightward pan -- defines the target video and must be reproduced unchanged.

summary:
[video editing] The target video is an edited version of <Video 1>. Every element
of the original shot is preserved: the same camera movement at the same speed,
the same geometry, the same objects in the same places. The only change is:
bright neutral morning daylight, crisp shadows.

retention_analysis:
<Video 1> (source video editing): fully_preserved - every surface, object and
material in <Video 1> stays exactly as it appears there, in the same place, with
the same shape and the same construction. Flat wall panels stay flat wall panels and every piece of furniture keeps its place and shape. The camera path and
framing are unchanged. Only the illumination, weather and atmosphere change.

detailed_description:
The target video is <Video 1> itself, relit. Every frame matches the
corresponding frame of <Video 1> in composition, geometry and camera position:
the same surfaces in the same places, the same objects at the same coordinates,
the same camera path at the same speed.
[Shot 1] Bright clear midday daylight floods in through the glass doors under a blue sky. The light is neutral and white, the wood panels read warm and clean, and the furniture casts crisp defined shadows across the floor. Nothing is redesigned, restyled, added, removed or rearranged.
Existing surfaces keep their own materials and colours and simply respond to the
new light and conditions.

overall_soundscape:
The quiet interior ambience of the room.

non_diegetic_music:
None.

subject_definitions:
<Subject 1> is the young man seated on the patterned couch in <Video 1>, in a grey sweatshirt over a white long-sleeved top, holding a phone to his ear. Their appearance, clothing and place in the scene are
taken from <Video 1> and stay exactly as they appear there.
<Video 1> is the source video to be edited. Its environment, lighting, framing
and camera movement define the target video. What the edit changes is the
movement and behaviour of <Subject 1>.

summary:
[video editing + motion animation] The target video is an edited version of
<Video 1>. <Subject 1> keeps identity, clothing and position, and the
environment, lighting and camera movement of <Video 1> are unchanged. What
changes is the movement and behaviour of <Subject 1>: he stands and walks toward the camera.

retention_analysis:
<Video 1> (source video editing): partially_preserved - the environment,
lighting, framing and camera motion are preserved: the patterned couch, the teal patterned wallpaper and the framed pictures behind him stay exactly as they are, and the camera does not move.
<Subject 1> (appears in [Shot 1]): partially_preserved - appearance is retained
exactly; the body movement, head orientation and gaze are re-animated as
described below.

detailed_description:
The target video opens matching <Video 1> in environment, lighting, framing and
the appearance of <Subject 1>.
[Shot 1] He puts the phone down, plants both feet, rises from the couch and walks steadily straight towards the camera, growing larger in the frame until his upper body fills much of it. The movement is continuous and physically natural across the
full five seconds, held to the end rather than snapping back. The environment,
lighting and camera movement of <Video 1> continue unchanged behind the action.

overall_soundscape:
The quiet interior ambience of the room.

non_diegetic_music:
None.

subject_definitions:
<Subject 1> is the young man seated on the patterned couch in <Video 1>, in a grey sweatshirt over a white long-sleeved top, holding a phone to his ear. Their appearance, clothing and place in the scene are
taken from <Video 1> and stay exactly as they appear there.
<Video 1> is the source video to be edited. Its environment, lighting, framing
and camera movement define the target video. What the edit changes is the
movement and behaviour of <Subject 1>.

summary:
[video editing + motion animation] The target video is an edited version of
<Video 1>. <Subject 1> keeps identity, clothing and position, and the
environment, lighting and camera movement of <Video 1> are unchanged. What
changes is the movement and behaviour of <Subject 1>: he glances sharply over his shoulder.

retention_analysis:
<Video 1> (source video editing): partially_preserved - the environment,
lighting, framing and camera motion are preserved: the patterned couch, the teal patterned wallpaper and the framed pictures behind him stay exactly as they are, and the camera does not move.
<Subject 1> (appears in [Shot 1]): partially_preserved - appearance is retained
exactly; the body movement, head orientation and gaze are re-animated as
described below.

detailed_description:
The target video opens matching <Video 1> in environment, lighting, framing and
the appearance of <Subject 1>.
[Shot 1] He stops mid-conversation and turns his head sharply to look back over his own shoulder into the space behind him, holding that look with his body tense, then slowly turning back to face forward. The movement is continuous and physically natural across the
full five seconds, held to the end rather than snapping back. The environment,
lighting and camera movement of <Video 1> continue unchanged behind the action.

overall_soundscape:
The quiet interior ambience of the room.

non_diegetic_music:
None.

subject_definitions:
<Subject 1> is the young man seated on the patterned couch in <Video 1>, in a grey sweatshirt over a white long-sleeved top, holding a phone to his ear. Their appearance, clothing and place in the scene are
taken from <Video 1> and stay exactly as they appear there.
<Video 1> is the source video to be edited. Its environment, lighting, framing
and camera movement define the target video. What the edit changes is the
movement and behaviour of <Subject 1>.

summary:
[video editing + motion animation] The target video is an edited version of
<Video 1>. <Subject 1> keeps identity, clothing and position, and the
environment, lighting and camera movement of <Video 1> are unchanged. What
changes is the movement and behaviour of <Subject 1>: he leans forward, elbows on knees, talking to camera.

retention_analysis:
<Video 1> (source video editing): partially_preserved - the environment,
lighting, framing and camera motion are preserved: the patterned couch, the teal patterned wallpaper and the framed pictures behind him stay exactly as they are, and the camera does not move.
<Subject 1> (appears in [Shot 1]): partially_preserved - appearance is retained
exactly; the body movement, head orientation and gaze are re-animated as
described below.

detailed_description:
The target video opens matching <Video 1> in environment, lighting, framing and
the appearance of <Subject 1>.
[Shot 1] He lowers the phone, leans forward until his elbows rest on his knees, brings his face closer to the camera and begins speaking directly towards the lens, gesturing lightly with one hand as he talks. The movement is continuous and physically natural across the
full five seconds, held to the end rather than snapping back. The environment,
lighting and camera movement of <Video 1> continue unchanged behind the action.

overall_soundscape:
A man speaking directly and conversationally.

non_diegetic_music:
None.

subject_definitions:
<Subject 1> is every person walking in the pedestrian plaza in <Video 1>, with their own clothing and their own positions in the crowd. Their appearance, clothing and place in the scene are
taken from <Video 1> and stay exactly as they appear there.
<Video 1> is the source video to be edited. Its environment, lighting, framing
and camera movement define the target video. What the edit changes is the
movement and behaviour of <Subject 1>.

summary:
[video editing + motion animation] The target video is an edited version of
<Video 1>. <Subject 1> keeps identity, clothing and position, and the
environment, lighting and camera movement of <Video 1> are unchanged. What
changes is the movement and behaviour of <Subject 1>: everyone stops, turns to stare into the lens, then sprints straight at the camera.

retention_analysis:
<Video 1> (source video editing): partially_preserved - the environment,
lighting, framing and camera motion are preserved: the plaza paving, the kiosk, the autumn trees, the arched building in the distance and the daylight all stay exactly as they are, and the camera does not move.
<Subject 1> (appears in [Shot 1]): partially_preserved - appearance is retained
exactly; the body movement, head orientation and gaze are re-animated as
described below.

detailed_description:
The target video opens matching <Video 1> in environment, lighting, framing and
the appearance of <Subject 1>.
[Shot 1] Every person in the plaza suddenly halts mid-stride and freezes in place. In unison every head turns to look straight into the camera lens, expressions fixed and unblinking. Then, together, they break into a full-speed sprint directly towards the camera, filling the foreground as they close the distance fast. The movement is continuous and physically natural across the
full five seconds, held to the end rather than snapping back. The environment,
lighting and camera movement of <Video 1> continue unchanged behind the action.

overall_soundscape:
Outdoor city ambience with footsteps and distant voices.

non_diegetic_music:
None.

subject_definitions:
<Subject 1> is every person walking in the pedestrian plaza in <Video 1>, with their own clothing and their own positions in the crowd. Their appearance, clothing and place in the scene are
taken from <Video 1> and stay exactly as they appear there.
<Video 1> is the source video to be edited. Its environment, lighting, framing
and camera movement define the target video. What the edit changes is the
movement and behaviour of <Subject 1>.

summary:
[video editing + motion animation] The target video is an edited version of
<Video 1>. <Subject 1> keeps identity, clothing and position, and the
environment, lighting and camera movement of <Video 1> are unchanged. What
changes is the movement and behaviour of <Subject 1>: the crowd parts, opening a clear path down the centre.

retention_analysis:
<Video 1> (source video editing): partially_preserved - the environment,
lighting, framing and camera motion are preserved: the plaza paving, the kiosk, the autumn trees, the arched building in the distance and the daylight all stay exactly as they are, and the camera does not move.
<Subject 1> (appears in [Shot 1]): partially_preserved - appearance is retained
exactly; the body movement, head orientation and gaze are re-animated as
described below.

detailed_description:
The target video opens matching <Video 1> in environment, lighting, framing and
the appearance of <Subject 1>.
[Shot 1] The people in the middle of the frame move aside to left and right, stepping deliberately out of the centre of the plaza until a clear open corridor of paving runs straight away from the camera towards the arched building, with the crowd standing in two banks along its edges. The movement is continuous and physically natural across the
full five seconds, held to the end rather than snapping back. The environment,
lighting and camera movement of <Video 1> continue unchanged behind the action.

overall_soundscape:
Outdoor city ambience with footsteps and distant voices.

non_diegetic_music:
None.

subject_definitions:
<Subject 1> is every person walking in the pedestrian plaza in <Video 1>, with their own clothing and their own positions in the crowd. Their appearance, clothing and place in the scene are
taken from <Video 1> and stay exactly as they appear there.
<Video 1> is the source video to be edited. Its environment, lighting, framing
and camera movement define the target video. What the edit changes is the
movement and behaviour of <Subject 1>.

summary:
[video editing + motion animation] The target video is an edited version of
<Video 1>. <Subject 1> keeps identity, clothing and position, and the
environment, lighting and camera movement of <Video 1> are unchanged. What
changes is the movement and behaviour of <Subject 1>: rain begins, umbrellas open and the pace quickens.

retention_analysis:
<Video 1> (source video editing): partially_preserved - the environment,
lighting, framing and camera motion are preserved: the plaza paving, the kiosk, the autumn trees, the arched building in the distance and the daylight all stay exactly as they are, and the camera does not move.
<Subject 1> (appears in [Shot 1]): partially_preserved - appearance is retained
exactly; the body movement, head orientation and gaze are re-animated as
described below.

detailed_description:
The target video opens matching <Video 1> in environment, lighting, framing and
the appearance of <Subject 1>.
[Shot 1] Rain begins to fall across the plaza, darkening the paving and speckling it with wet. People raise umbrellas that open above them, pull hoods up, and quicken their pace, hurrying across the frame through the falling rain. The movement is continuous and physically natural across the
full five seconds, held to the end rather than snapping back. The environment,
lighting and camera movement of <Video 1> continue unchanged behind the action.

overall_soundscape:
Rain on paving, hurried footsteps.

non_diegetic_music:
None.

subject_definitions:
<Subject 1> is every person walking in the pedestrian plaza in <Video 1>, with their own clothing and their own positions in the crowd. Their appearance, clothing and place in the scene are
taken from <Video 1> and stay exactly as they appear there.
<Video 1> is the source video to be edited. Its environment, lighting, framing
and camera movement define the target video. What the edit changes is the
movement and behaviour of <Subject 1>.

summary:
[video editing + motion animation] The target video is an edited version of
<Video 1>. <Subject 1> keeps identity, clothing and position, and the
environment, lighting and camera movement of <Video 1> are unchanged. What
changes is the movement and behaviour of <Subject 1>: the same plaza at night, shopfronts and lamps lit.

retention_analysis:
<Video 1> (source video editing): partially_preserved - the environment,
lighting, framing and camera motion are preserved: the plaza paving, the kiosk, the autumn trees, the arched building in the distance and the daylight all stay exactly as they are, and the camera does not move.
<Subject 1> (appears in [Shot 1]): partially_preserved - appearance is retained
exactly; the body movement, head orientation and gaze are re-animated as
described below.

detailed_description:
The target video opens matching <Video 1> in environment, lighting, framing and
the appearance of <Subject 1>.
[Shot 1] The same plaza at night. Warm shopfront windows and street lamps light the paving in pools, the kiosk glows from within, and the sky above the distant arch is dark. The same people continue walking through the frame, lit from the side by the shop windows. The movement is continuous and physically natural across the
full five seconds, held to the end rather than snapping back. The environment,
lighting and camera movement of <Video 1> continue unchanged behind the action.

overall_soundscape:
Outdoor city ambience with footsteps and distant voices.

non_diegetic_music:
None.

subject_definitions:
<Video 1> is the source video to be edited. Every visual element of <Video 1> --
the road surface and its double yellow centre line and dashed white lane markings, the low stone wall on the right, the dense trees on both sides, the car ahead in the distance, and the camera's forward motion and its speed -- defines the target video and must be reproduced unchanged.

summary:
[video editing] The target video is an edited version of <Video 1>. Every element
of the original shot is preserved: the same camera movement at the same speed,
the same geometry, the same objects in the same places. The only change is:
night driving, headlights on wet reflective asphalt.

retention_analysis:
<Video 1> (source video editing): fully_preserved - every surface, object and
material in <Video 1> stays exactly as it appears there, in the same place, with
the same shape and the same construction. The road keeps its exact markings and width, the stone wall stays where it is, and the forward motion and its speed are unchanged. The camera path and
framing are unchanged. Only the illumination, weather and atmosphere change.

detailed_description:
The target video is <Video 1> itself, relit. Every frame matches the
corresponding frame of <Video 1> in composition, geometry and camera position:
the same surfaces in the same places, the same objects at the same coordinates,
the same camera path at the same speed.
[Shot 1] It is night on the same road. The headlights throw a bright cone onto the asphalt ahead, the yellow centre line and white dashes catching the light and reflecting back. The trees on both sides fall away into darkness beyond the beam, and the tail lights of the car ahead glow red. Nothing is redesigned, restyled, added, removed or rearranged.
Existing surfaces keep their own materials and colours and simply respond to the
new light and conditions.

overall_soundscape:
Steady road noise from a moving car, tyres on asphalt.

non_diegetic_music:
None.

subject_definitions:
<Video 1> is the source video to be edited. Every visual element of <Video 1> --
the road surface and its double yellow centre line and dashed white lane markings, the low stone wall on the right, the dense trees on both sides, the car ahead in the distance, and the camera's forward motion and its speed -- defines the target video and must be reproduced unchanged.

summary:
[video editing] The target video is an edited version of <Video 1>. Every element
of the original shot is preserved: the same camera movement at the same speed,
the same geometry, the same objects in the same places. The only change is:
heavy snow, the road and verges white.

retention_analysis:
<Video 1> (source video editing): fully_preserved - every surface, object and
material in <Video 1> stays exactly as it appears there, in the same place, with
the same shape and the same construction. The road keeps its exact markings and width, the stone wall stays where it is, and the forward motion and its speed are unchanged. The camera path and
framing are unchanged. Only the illumination, weather and atmosphere change.

detailed_description:
The target video is <Video 1> itself, relit. Every frame matches the
corresponding frame of <Video 1> in composition, geometry and camera position:
the same surfaces in the same places, the same objects at the same coordinates,
the same camera path at the same speed.
[Shot 1] Heavy snow falls across the road and lies thick on the verges, the stone wall and the trees, which are weighed down white. The asphalt is covered in packed snow with dark tyre tracks, and the light is cold, flat and pale blue-white. Nothing is redesigned, restyled, added, removed or rearranged.
Existing surfaces keep their own materials and colours and simply respond to the
new light and conditions.

overall_soundscape:
Steady road noise from a moving car, tyres on asphalt.

non_diegetic_music:
None.

subject_definitions:
<Video 1> is the source video to be edited. Every visual element of <Video 1> --
the road surface and its double yellow centre line and dashed white lane markings, the low stone wall on the right, the dense trees on both sides, the car ahead in the distance, and the camera's forward motion and its speed -- defines the target video and must be reproduced unchanged.

summary:
[video editing] The target video is an edited version of <Video 1>. Every element
of the original shot is preserved: the same camera movement at the same speed,
the same geometry, the same objects in the same places. The only change is:
autumn, the roadside trees turned orange and gold.

retention_analysis:
<Video 1> (source video editing): fully_preserved - every surface, object and
material in <Video 1> stays exactly as it appears there, in the same place, with
the same shape and the same construction. The road keeps its exact markings and width, the stone wall stays where it is, and the forward motion and its speed are unchanged. The camera path and
framing are unchanged. Only the illumination, weather and atmosphere change.

detailed_description:
The target video is <Video 1> itself, relit. Every frame matches the
corresponding frame of <Video 1> in composition, geometry and camera position:
the same surfaces in the same places, the same objects at the same coordinates,
the same camera path at the same speed.
[Shot 1] The trees lining both sides of the road have turned deep orange, gold and red, and fallen leaves lie along the verges and drift across the asphalt. Warm autumn light comes through the canopy in patches onto the road surface. Nothing is redesigned, restyled, added, removed or rearranged.
Existing surfaces keep their own materials and colours and simply respond to the
new light and conditions.

overall_soundscape:
Steady road noise from a moving car, tyres on asphalt.

non_diegetic_music:
None.

subject_definitions:
<Video 1> is the source video to be edited. Every visual element of <Video 1> --
the road surface and its double yellow centre line and dashed white lane markings, the low stone wall on the right, the dense trees on both sides, the car ahead in the distance, and the camera's forward motion and its speed -- defines the target video and must be reproduced unchanged.

summary:
[video editing] The target video is an edited version of <Video 1>. Every element
of the original shot is preserved: the same camera movement at the same speed,
the same geometry, the same objects in the same places. The only change is:
dense fog, visibility closing in ahead.

retention_analysis:
<Video 1> (source video editing): fully_preserved - every surface, object and
material in <Video 1> stays exactly as it appears there, in the same place, with
the same shape and the same construction. The road keeps its exact markings and width, the stone wall stays where it is, and the forward motion and its speed are unchanged. The camera path and
framing are unchanged. Only the illumination, weather and atmosphere change.

detailed_description:
The target video is <Video 1> itself, relit. Every frame matches the
corresponding frame of <Video 1> in composition, geometry and camera position:
the same surfaces in the same places, the same objects at the same coordinates,
the same camera path at the same speed.
[Shot 1] Dense fog lies over the road so that the trees on both sides become soft grey shapes and the road ahead dissolves into white within a short distance. The car ahead is a dim silhouette with glowing red tail lights, and the light is diffuse and low-contrast. Nothing is redesigned, restyled, added, removed or rearranged.
Existing surfaces keep their own materials and colours and simply respond to the
new light and conditions.

overall_soundscape:
Steady road noise from a moving car, tyres on asphalt.

non_diegetic_music:
None.

subject_definitions:
<Video 1> is the source video to be edited. Every visual element of <Video 1> --
the road surface and its double yellow centre line and dashed white lane markings, the low stone wall on the right, the dense trees on both sides, the car ahead in the distance, and the camera's forward motion and its speed -- defines the target video and must be reproduced unchanged.

summary:
[video editing] The target video is an edited version of <Video 1>. Every element
of the original shot is preserved: the same camera movement at the same speed,
the same geometry, the same objects in the same places. The only change is:
a low golden sun ahead, long shadows across the road.

retention_analysis:
<Video 1> (source video editing): fully_preserved - every surface, object and
material in <Video 1> stays exactly as it appears there, in the same place, with
the same shape and the same construction. The road keeps its exact markings and width, the stone wall stays where it is, and the forward motion and its speed are unchanged. The camera path and
framing are unchanged. Only the illumination, weather and atmosphere change.

detailed_description:
The target video is <Video 1> itself, relit. Every frame matches the
corresponding frame of <Video 1> in composition, geometry and camera position:
the same surfaces in the same places, the same objects at the same coordinates,
the same camera path at the same speed.
[Shot 1] A low golden sun sits ahead down the road, flaring warm across the windscreen line and throwing long tree shadows in stripes across the asphalt. The yellow centre line glows warm and the stone wall catches amber light along its top. Nothing is redesigned, restyled, added, removed or rearranged.
Existing surfaces keep their own materials and colours and simply respond to the
new light and conditions.

overall_soundscape:
Steady road noise from a moving car, tyres on asphalt.

non_diegetic_music:
None.

subject_definitions:
<Video 1> is the source video to be edited. Every visual element of <Video 1> --
the road surface and its double yellow centre line and dashed white lane markings, the low stone wall on the right, the dense trees on both sides, the car ahead in the distance, and the camera's forward motion and its speed -- defines the target video and must be reproduced unchanged.

summary:
[video editing] The target video is an edited version of <Video 1>. Every element
of the original shot is preserved: the same camera movement at the same speed,
the same geometry, the same objects in the same places. The only change is:
a thunderstorm, rain sheeting across the road.

retention_analysis:
<Video 1> (source video editing): fully_preserved - every surface, object and
material in <Video 1> stays exactly as it appears there, in the same place, with
the same shape and the same construction. The road keeps its exact markings and width, the stone wall stays where it is, and the forward motion and its speed are unchanged. The camera path and
framing are unchanged. Only the illumination, weather and atmosphere change.

detailed_description:
The target video is <Video 1> itself, relit. Every frame matches the
corresponding frame of <Video 1> in composition, geometry and camera position:
the same surfaces in the same places, the same objects at the same coordinates,
the same camera path at the same speed.
[Shot 1] Rain sheets down across the road, standing in films on the asphalt that throw back the light, and spray rises from the road behind the car ahead. The sky above the trees is dark storm grey and lightning flashes illuminate the scene in brief bright bursts. Nothing is redesigned, restyled, added, removed or rearranged.
Existing surfaces keep their own materials and colours and simply respond to the
new light and conditions.

overall_soundscape:
Heavy rain and rolling thunder over road noise.

non_diegetic_music:
None.

subject_definitions:
<Video 1> is the source aerial video to be edited. Every visual element of
<Video 1> -- the autumn woodland and scrub with its rust, olive and gold foliage, the band of trees across the middle distance, the low hills along the horizon, the pale sky with thin cloud and the warm low sunlight -- defines the target video and must be
reproduced unchanged. What the edit changes is the flight path of the camera.

summary:
[video editing + camera movement] The target video is an edited version of
<Video 1>. The landscape, its colours, the light and the time of day are
preserved exactly as in <Video 1>. What changes is how the camera moves through
the scene: the drone descends until it is skimming just above the treetops.

retention_analysis:
<Video 1> (source video editing): partially_preserved - the terrain, vegetation,
horizon, sky and lighting are fully preserved: The vegetation, the horizon line and the warm low-sun lighting stay exactly as they are in <Video 1>. The camera
trajectory is re-animated as described below; nothing in the landscape itself is
redesigned, added or removed.

detailed_description:
The target video shows the same landscape as <Video 1>, in the same light, from
a camera that now moves differently through it.
[Shot 1] The camera descends steadily from its high vantage until it is flying low and fast just above the tops of the trees, the foliage rushing past close beneath it while the distant hills and horizon rise into the upper part of the frame. The motion is smooth and continuous across the full five
seconds, in the manner of a drone flight, with the horizon staying level and the
terrain remaining exactly the terrain of <Video 1>.

overall_soundscape:
Open air, a faint wind across an empty landscape.

non_diegetic_music:
None.

subject_definitions:
<Video 1> is the source aerial video to be edited. Every visual element of
<Video 1> -- the autumn woodland and scrub with its rust, olive and gold foliage, the band of trees across the middle distance, the low hills along the horizon, the pale sky with thin cloud and the warm low sunlight -- defines the target video and must be
reproduced unchanged. What the edit changes is the flight path of the camera.

summary:
[video editing + camera movement] The target video is an edited version of
<Video 1>. The landscape, its colours, the light and the time of day are
preserved exactly as in <Video 1>. What changes is how the camera moves through
the scene: the drone accelerates hard forward across the landscape.

retention_analysis:
<Video 1> (source video editing): partially_preserved - the terrain, vegetation,
horizon, sky and lighting are fully preserved: The vegetation, the horizon line and the warm low-sun lighting stay exactly as they are in <Video 1>. The camera
trajectory is re-animated as described below; nothing in the landscape itself is
redesigned, added or removed.

detailed_description:
The target video shows the same landscape as <Video 1>, in the same light, from
a camera that now moves differently through it.
[Shot 1] The camera accelerates sharply forward, the woodland below streaming past faster and faster beneath it as it races out across the landscape towards the distant hills, the ground blurring with speed at the bottom of the frame. The motion is smooth and continuous across the full five
seconds, in the manner of a drone flight, with the horizon staying level and the
terrain remaining exactly the terrain of <Video 1>.

overall_soundscape:
Open air, a faint wind across an empty landscape.

non_diegetic_music:
None.

subject_definitions:
<Video 1> is the source aerial video to be edited. Every visual element of
<Video 1> -- the autumn woodland and scrub with its rust, olive and gold foliage, the band of trees across the middle distance, the low hills along the horizon, the pale sky with thin cloud and the warm low sunlight -- defines the target video and must be
reproduced unchanged. What the edit changes is the flight path of the camera.

summary:
[video editing + camera movement] The target video is an edited version of
<Video 1>. The landscape, its colours, the light and the time of day are
preserved exactly as in <Video 1>. What changes is how the camera moves through
the scene: the drone slows to an almost stationary hover.

retention_analysis:
<Video 1> (source video editing): partially_preserved - the terrain, vegetation,
horizon, sky and lighting are fully preserved: The vegetation, the horizon line and the warm low-sun lighting stay exactly as they are in <Video 1>. The camera
trajectory is re-animated as described below; nothing in the landscape itself is
redesigned, added or removed.

detailed_description:
The target video shows the same landscape as <Video 1>, in the same light, from
a camera that now moves differently through it.
[Shot 1] The forward motion of the camera slows and settles until it is hanging almost stationary above the woodland, drifting only very slightly, the landscape below nearly still beneath it. The motion is smooth and continuous across the full five
seconds, in the manner of a drone flight, with the horizon staying level and the
terrain remaining exactly the terrain of <Video 1>.

overall_soundscape:
Open air, a faint wind across an empty landscape.

non_diegetic_music:
None.

subject_definitions:
<Video 1> is the source aerial video to be edited. Every visual element of
<Video 1> -- the autumn woodland and scrub with its rust, olive and gold foliage, the band of trees across the middle distance, the low hills along the horizon, the pale sky with thin cloud and the warm low sunlight -- defines the target video and must be
reproduced unchanged. What the edit changes is the flight path of the camera.

summary:
[video editing + camera movement] The target video is an edited version of
<Video 1>. The landscape, its colours, the light and the time of day are
preserved exactly as in <Video 1>. What changes is how the camera moves through
the scene: the drone banks into a wide orbit over the woodland.

retention_analysis:
<Video 1> (source video editing): partially_preserved - the terrain, vegetation,
horizon, sky and lighting are fully preserved: The vegetation, the horizon line and the warm low-sun lighting stay exactly as they are in <Video 1>. The camera
trajectory is re-animated as described below; nothing in the landscape itself is
redesigned, added or removed.

detailed_description:
The target video shows the same landscape as <Video 1>, in the same light, from
a camera that now moves differently through it.
[Shot 1] The camera banks into a smooth wide arc, curving sideways over the woodland so that the landscape below rotates steadily through the frame and the line of distant hills sweeps across the horizon from one side to the other. The motion is smooth and continuous across the full five
seconds, in the manner of a drone flight, with the horizon staying level and the
terrain remaining exactly the terrain of <Video 1>.

overall_soundscape:
Open air, a faint wind across an empty landscape.

non_diegetic_music:
None.

subject_definitions:
<Video 1> is the source aerial video to be edited. Every visual element of
<Video 1> -- the autumn woodland and scrub with its rust, olive and gold foliage, the band of trees across the middle distance, the low hills along the horizon, the pale sky with thin cloud and the warm low sunlight -- defines the target video and must be
reproduced unchanged. What the edit changes is the flight path of the camera.

summary:
[video editing + camera movement] The target video is an edited version of
<Video 1>. The landscape, its colours, the light and the time of day are
preserved exactly as in <Video 1>. What changes is how the camera moves through
the scene: the drone climbs and pulls back to reveal the whole landscape.

retention_analysis:
<Video 1> (source video editing): partially_preserved - the terrain, vegetation,
horizon, sky and lighting are fully preserved: The vegetation, the horizon line and the warm low-sun lighting stay exactly as they are in <Video 1>. The camera
trajectory is re-animated as described below; nothing in the landscape itself is
redesigned, added or removed.

detailed_description:
The target video shows the same landscape as <Video 1>, in the same light, from
a camera that now moves differently through it.
[Shot 1] The camera rises steadily and pulls backwards at the same time, the woodland falling away below and more and more of the landscape opening up in the frame, until a wide expanse of autumn terrain stretches to the distant hills. The motion is smooth and continuous across the full five
seconds, in the manner of a drone flight, with the horizon staying level and the
terrain remaining exactly the terrain of <Video 1>.

overall_soundscape:
Open air, a faint wind across an empty landscape.

non_diegetic_music:
None.

subject_definitions:
<Video 1> is the source aerial video to be edited. Every visual element of
<Video 1> -- the autumn woodland and scrub with its rust, olive and gold foliage, the band of trees across the middle distance, the low hills along the horizon, the pale sky with thin cloud and the warm low sunlight -- defines the target video and must be
reproduced unchanged. What the edit changes is the flight path of the camera.

summary:
[video editing + camera movement] The target video is an edited version of
<Video 1>. The landscape, its colours, the light and the time of day are
preserved exactly as in <Video 1>. What changes is how the camera moves through
the scene: the drone dives down toward the trees, then levels out.

retention_analysis:
<Video 1> (source video editing): partially_preserved - the terrain, vegetation,
horizon, sky and lighting are fully preserved: The vegetation, the horizon line and the warm low-sun lighting stay exactly as they are in <Video 1>. The camera
trajectory is re-animated as described below; nothing in the landscape itself is
redesigned, added or removed.

detailed_description:
The target video shows the same landscape as <Video 1>, in the same light, from
a camera that now moves differently through it.
[Shot 1] The camera pitches down and dives towards the woodland below, the trees rushing up towards it and filling more of the frame, then levels out smoothly into low forward flight just above the canopy. The motion is smooth and continuous across the full five
seconds, in the manner of a drone flight, with the horizon staying level and the
terrain remaining exactly the terrain of <Video 1>.

overall_soundscape:
Open air, a faint wind across an empty landscape.

non_diegetic_music:
None.