Culture & ArtsMusic Video
A question-and-answer exchange in a tavern peels back an old story layer by layer; the characters' attitudes, warm and cold settings and the red sleeves themselves move the period tale forward.
01 / Background
“Red Sleeves” tells its story through a song built on dialogue between characters: a visitor asks after a woman's whereabouts, the waiter recounts the past, and the visitor's reply then gives the audience a new understanding of the whole story. The film had to turn these accounts into character relationships viewers could follow, rather than just period scenery and scattered fragments of mood.
The look was set as photorealistic, with the texture of a period feature film. The characters' faces, how they look young and in middle age, their costumes, the tavern and the streets were all fixed in the storyboard first; dance, objects and settings then lead into the past. The tavern is the fixed place where the story is told, so however the scenes change, they always return to the same conversation.
This section focuses on the AI visuals and story structure of the song's MV. The nearly four-minute film cuts between character close-ups, dance, streets and wedding scenes, and also had to line up with the vocal recording and calligraphy lyrics so that the dense story turns keep pace with the music.
02 / Creative Highlights
The story jumps from the conversation at hand into memory, then back to a character's self-justification. The waiter, the visitor, the wine cups and the warm lamps keep reappearing, so every transition lands back in the tavern. Once viewers know who is speaking, they can follow the different attitudes toward the same past in the images that follow.
Golden lanterns and warm interior light alternate with blue rainy nights and moonlit cliffs. The woman in red stands out sharply in the cold-toned streets, and character close-ups pull attention back to the eyes. Light and color separate the locations and also make waiting, closeness and loss visible without any words.
The first half builds the mood with the woman waiting, dancing and being helped up after she is hurt; in the second half, the visitor speaks up, raises his cup and lets slip a faint smile, and the way the story is told changes. The tavern looks much the same throughout, so the characters' reactions matter more than new scenes, and close-ups are needed to read their attitudes.
Dance, red robes and falling petals keep the woman in viewers' minds; later, the broken bowl, the swordsman in the tavern and close-ups of the sword push the characters' argument to the end. The images move from soft sleeves to objects and action, giving the ending the force of a jianghu tale.
03 / Scene Breakdown
A visitor walks into the warmly lit tavern and asks the waiter if he has seen a certain woman; the waiter recalls the dancer in red who came every night and sat in the corner waiting for someone.
The camera looks into the tavern from outside. The room glows warm yellow as a figure in black walks in, and the first sung line appears: “Waiter, was there ever a woman here?”
A close-up of the waiter, his gaze sinking; a line of small text at the bottom reads “From the waiter's view, lost in memory,” telling viewers that what follows is his account of the past.
In a cold blue night alley, the woman in red turns and her skirt sweeps out wide. Throughout the film, her red remains the most striking color.
The woman sits in a corner of the tavern with a black dog lying at her feet; the camera then pushes in to her profile as she keeps gazing out the door.
A close-up of the carved wooden door cuts to a close-up of the woman turning her head: every time someone comes in, she looks up.
A few drinkers sit around a low table drinking and talking; the camera pulls back as her story grows more convincing with every retelling.
The waiter tells of the woman's past: sold off by her family at a young age, she kept to singing and dancing in the house; when someone offered a thousand gold for one night with her, she fought back with a knife, and a man stepped in to shield her.
In a cold-toned room, her parents and a matchmaker laugh around a table of gold while the woman sits in the shadows to one side; a single line tells her whole backstory.
A girl in yellow walks past a doorway full of blossoms, cutting to the woman in red spinning in a beam of light as the lyric sings “Yet she stayed true to herself.”
Several wealthy merchants clap and laugh; the next shot shows the woman holding a knife in front of her, her eyes not giving an inch.
A man in black shields the woman and looks back at the crowd; this is followed by a wide shot of a brawl in a night alley, matching the line “Blocked three strikes and sixteen punches.”
After a sky full of burning clouds, the film cuts to a blood-soaked white cloth. Rather than showing the fight itself, it uses color alone to tell viewers what happened that night.
The two vow to grow old together, but the man leaves before dawn, marries into a wealthy family and kneels before another clan's ancestors; at this point the waiter sighs. The woman dances her last dance, and all that remains is a gravestone.
A couple in red stands on high ground looking at the city gate tower; the next shot is an empty room in blue moonlight, with the man standing with his back to the camera.
A newlywed couple seen from behind in a red wedding hall, followed by the man kneeling before the ancestral tablets of another family's shrine, kneeling “without tiring.”
Back in the tavern, the waiter pours wine for his guest and sings “Sir, there's something you don't know”; he turns to look at the visitor with regret on his face.
The woman dances her last dance in an empty hall as red petals cover the floor; after she is gone, only the petals drift in the light.
A stone tablet stands in a weedy field in the mist as the lyric reaches “It was the madam who gathered her remains,” and the tale of the past is told.
With a cold laugh, the visitor reveals that he is the man from back then; that heroic rescue was a setup arranged by him and a friend.
A hand pushes a black wine cup out of frame, followed by a close-up of the visitor with a smile at the corner of his mouth; small text reads “The visitor reveals who he is: he is that man.”
An extreme close-up of the woman's face as a tear rolls down her cheek, matching his line “As for those few tears.”
After a close-up of a red candle, two men stand in a night alley talking in low voices: “It was a setup by me and a friend.”
The visitor laughs as he waves a folding fan, and the film flashes back to him embracing the woman in front of the tavern, her red skirt swirling.
He walks with the woman down a lantern-lit street; the next shot shows someone knocked to the ground: the fight back then was staged for her benefit.
The woman was driven to her death, yet the visitor calls it merely a mistake; he tells of once loving someone himself, but he was half a tael short of the bride price, she married a 69-year-old man, and he turned around and married into a wealthy family.
Under the moon, the woman in red walks alone down stone steps into the dark; the camera doesn't follow, only watching her get farther and farther away.
The visitor sits under a lamp toying with his folding fan, then we see his hand straightening his sleeve as he sings “Let's just call it a mistake,” looking entirely unconcerned.
An empty shot with no one in it: a lake at night and a small house by the water, giving the dense lyrics a pause.
He walks with a woman in green along a street after the rain; this is his other love. Then comes a close-up of a few copper coins on a table: “Just half a tael short of the bride price.”
A bride in full wedding dress bows to a white-haired old man as guests fill the courtyard, then the film cuts to a wedding hall hung with red lanterns.
Behind a vermilion gate guarded by stone lions, the visitor stands at the door with his arm around a wealthy woman, gesturing and smiling smugly.
Having heard his excuses, the waiter rebuts him point by point; the visitor smashes a bowl and gets up, a swordsman in a bamboo hat nearby finally speaks, and a close-up of his sword leads the story into the next chapter.
Under red lanterns, he and the woman stand face to face as the lyric sings “You won't love the one who's only good to you, you only want the bad one to stay.”
The waiter stands with his back to the camera in front of a steaming stove, with small text reading “Having heard his sophistry, the waiter begins to push back”; the visitor sits behind a table, listening.
A figure walks away down a night alley; the next shot is the black dog lying at the tavern door, the very one that lay at the woman's feet at the start.
The stone tablet in the weeds appears once more, then the waiter stands before the big drum outside the magistrate's office: “Let me find someone and report it to the magistrate.”
A broken bowl on the floor, followed by the swordsman sitting before the firelight in a bamboo hat with his back to the camera, singing the last line: “Waiter, why waste words on him?”
A hand grips a sword in the firelight, the image fades to dark, and a caption explains that this line will open “The Swordsman,” the next chapter in the “One Tavern” series.
04 / Storyboards & Final Frames
The storyboard document sets out the image, shot size and prompt for each lyric line. Below, in the order of the final film, are the final stills that match the finished cut.
Shot 1.1
Late at night, someone walks in from outside into the warmly lit tavern. The camera stays outside the door: cold outside, warm inside.
Shot 1.2
The waiter looks down to remember, then stands by the table with the flower vase and points it out to the visitor.
Shot 1.3
The first glimpse in memory: the woman in red turns in a night alley, her skirt sweeping out.
Shot 1.4
The woman sits in a corner of the tavern with a black dog lying at her feet; the dog will appear again at the end.
Shot 1.5
A close-up of the carved wooden door, cutting to a close-up of the woman looking back at the door.
Shot 1.6
The woman leans against the door frame looking out, with a courtyard full of blossoms behind her.
Shot 1.7
Under the moon, a figure in red stands on a path at night, seen from behind.
Shot 2.1
In a cold-toned room, her parents and a matchmaker gather around a table of gold while the woman sits in the shadows.
Shot 2.2
A girl in yellow walks past a doorway full of blossoms.
Shot 2.3
The woman in red spins and dances in a beam of light.
Shot 2.4
The woman holds a knife in front of her, a ring of men watching behind her; then the man appears and shields her.
Shot 2.5
A wide shot of a brawl in a night alley, showing only silhouettes, not wounds.
Shot 2.6
A sky full of burning clouds, then a cut to a blood-stained white cloth.
Shot 3.1
An empty room in blue moonlight, the man standing with his back to the camera; after dawn, the woman sleeps alone in the bed.
Shot 3.2
The man kneels in another family's ancestral shrine as incense smoke rises through the light.
Shot 3.3
Back in the tavern: the waiter pours wine for a guest, then turns to look at the visitor.
Shot 3.4
In the empty hall, the woman dances her last dance as red petals cover the floor.
Shot 4.1
A hand pushes a black wine cup forward, and the visitor begins to tell his own version.
Shot 4.2
An extreme close-up of the woman's face as a tear runs down.
Shot 4.3
In a wealthy household's side chamber, the woman turns to look at the man at the door.
Shot 4.4
A burning red candle.
Shot 4.5
Two men stand in a night alley talking in secret.
Shot 4.6
In front of the tavern, he and the woman embrace and say goodbye.
Shot 4.7
On the lantern-lit street, the man walks away, seen from behind as he grows smaller.
Shot 4.8
He smiles as he walks with the woman down the street, looking back at her.
Shot 4.9
Someone is knocked to the ground, surrounded by onlookers: the fight back then was staged.
Shot 5.1
Under the moon, the woman in red walks down stone steps into the dark.
Shot 5.2
A hand straightening a sleeve, looking entirely unconcerned.
Shot 5.3
A lake at night and a small house by the water, in an empty shot with no one in it.
Shot 5.4
A close-up of a few copper coins on a table.
Shot 5.5
A bride in full wedding dress bows to a white-haired old man as guests fill the courtyard.
Shot 5.6
A wedding hall hung with red lanterns.
Shot 5.7
A vermilion gate guarded by stone lions, with the visitor standing in front, his arm around a wealthy woman.
Shot 6.1
The visitor sits at a table holding a wine cup, speaking as if it were nothing.
Shot 6.2
The woman sits alone at a table scattered with love tokens; then a cut to the two of them standing face to face on the lantern street.
Shot 6.3
A smashed bowl on the floor.
05 / Results
A nearly four-minute, 16:9 period-style narrative MV, complete with the full character story, music and calligraphy lyrics.
The film was released on Bilibili and Douyin. At the time of the screenshot it had 769K views, 27K likes and 21K favorites on Bilibili, and 234K likes, 40K favorites and 68K shares on Douyin. Released by Xiao Shaoyan on October 25, 2024.
Case Study
In the song “Red Sleeves,” three people are talking.
A visitor asks after someone, the waiter tells of the past, and halfway through, the visitor speaks up: the man back then was me.
We made a nearly four-minute period-style narrative MV for this song, using AI visuals to turn this question-and-answer exchange into a story with a twist.
After its release on Bilibili and Douyin on October 25, 2024, it had 769K views on Bilibili and 234K likes on Douyin at the time of the screenshot.
01
Most period-style MVs let the images follow the mood: beautiful women, lanterns, petals, and as long as it looks good, that's enough.
“Red Sleeves” is different. The lyrics themselves are a conversation: the visitor asks, the waiter answers, and in the second half the visitor speaks for himself, turning the story told so far completely on its head.
Viewers hear it only once and have to keep up with who is speaking, and whether they're talking about now or the past.
We set the tavern as “the present”: warm yellow lamps, wooden tables, wine cups, with the waiter and the visitor both here. The past went into “then”: cold blue night alleys, moonlight, an empty room. Each time the story reaches the end of a section, the images return to the tavern.
The warmly lit tavern and the cold night outside the door establish “the present” from the very first shot.
Red is the other thread. Nearly every shot the woman appears in has a touch of red: a skirt sweeping as she turns, sleeves flying as she dances, petals covering the floor. In the cold-toned night, viewers recognize her at a glance.
02
Before generating any images, the storyboard document first set out character profiles: the visitor as a middle-aged and a young man, the dancer, the waiter, the swordsman, plus supporting characters like the madam and the wealthy merchants. Each one's looks, clothing and personality were written out, and a look-test image made for each.
The style terms were the same across the film: photorealistic, period drama, period feature film. Whether a close-up in the tavern or a wide shot on a moonlit night, everything looks like it belongs to the same film.
Then we broke the whole song into 59 lines, one shot per lyric line: each line spells out what's in the frame, the shot size and what needs special attention. Stills were upscaled and reviewed together, the best ones picked and then turned into motion shots. For shots like the woman leaning on the door frame looking out, where the figure, hand and door frame have to line up, we did separate inpainting.
The singer wanted the characters not to look straight into the camera. So we had each character look at the person they were speaking to: the waiter looks at the visitor, the woman at the door, the visitor at the wine cup in his hand. The relationships between characters are set up first through their eyelines.
03
The storyboard has one small touch: as the woman waits in the corner of the tavern, a black dog lies at her feet.
The dog barely registers at the time. Then at the end, the waiter scolds the visitor, “You're not even as good as that dog back then,” and the camera cuts to the tavern door, where the same black dog is still lying.
On both lines, the storyboard document carries the same note: keep the dog consistent with earlier and later shots. Viewers may not put it into words, but the moment the two match up makes them feel someone really thought this story through.
04
The hardest moment in the film to follow is when the visitor suddenly speaks: “Heh, how could I not know?”
For more than a minute before this, viewers assume he's just someone asking for news. If this line relied on the singing alone, many would miss it.
So at a few key turning points, a line of small text was added at the bottom of the frame: at the start, “From the waiter's view, lost in memory”; here, “The visitor reveals who he is: he is that man”; and later, “Having heard his sophistry, the waiter begins to push back.”
The wine cup is pushed out of frame, the visitor smiles briefly under the warm lamp, and the small text at the bottom tells viewers: he is the man from back then.
In these lines of confession, the visitor is a young man dressed in black, matching the man in the memories who embraces the woman in front of the tavern and walks with her down the lantern street, so viewers know without thinking twice that it's the same person.
05
The lyrics are set in white calligraphy, with the one or two most important characters in each line marked in red: “lovely,” “corner,” “setup,” “wholehearted.” Even if viewers can't make out the lyrics, a glance at the red characters tells them what the line is about.
In the edit, the images were checked line by line against the singer's recording, and the subtitles follow the words actually sung on the recording, not the first draft in the document. Where the lyrics are dense, empty shots like the night lake and the empty room were cut in to give viewers a breather.
At the end of the film comes the sound of a sword being drawn. In the edit we added a close-up of a hand holding a sword so the sound and image match, which also hands the story over to the swordsman of the next chapter.
After the broken bowl, the swordsman in a bamboo hat speaks with his back to the camera, and the last sung line is left to the next character.
06
“Red Sleeves” is one chapter in the “One Tavern” series. The closing caption says it plainly: the next chapter, “The Swordsman,” begins with the line “Waiter, why waste words on him?”
For the singer, this MV gives the song a storyline people can follow: listeners can track the three accounts layer by layer through the images, then wait for the next chapter.
For us, it shows that AI visuals can do more than “look good”: lock the characters first, lay out the images one shot per line, use warm and cold tones to separate time, red characters to highlight key points and small text to explain the turns, and a nearly four-minute song can clearly tell a story with a twist.
FAQ
Before generating any images, “Red Sleeves” wrote character profiles: the visitor, the dancer, the waiter, the swordsman and several supporting characters each had their looks, clothing and look-test image fixed, and the whole film used the same set of style terms (photorealistic, period feature film). Every lyric line was then generated according to the profiles, and even small props like the black dog at the woman's feet were marked in the storyboard to stay consistent from start to end.
The song was broken into 59 lyric lines, with one shot per line written up with its image and shot size; empty shots and inserts were added in the final film, which ended up with roughly seventy-odd cuts.
The storyboard and image-generation documents were written on October 8–9, 2024, editing and lyric subtitles were finished in mid-October, and the film was released on Bilibili and Douyin on October 25: about two and a half weeks in all.
“Red Sleeves” was released in 1920×1080, 16:9 landscape. At the time of the screenshot it had 769K views, 27K likes and 21K favorites on Bilibili, and 234K likes, 40K favorites and 68K shares on Douyin.
Approach
The storyboard lists the subject, setting and shot size for each section of the song, with separate references for the characters' looks, ages and costumes. Characters look at whoever they're speaking to wherever possible, and multi-character frames are kept balanced, so expressions and eyelines establish the relationships first.
Stills were first generated with MJ, upscaled and reviewed together, then turned into motion. Shots like the dancer holding the door frame were partly inpainted so the figure, hand and door frame all fit the scene; every shot size shares the same period-film color grade.
The images were checked line by line against the singer's recording, and the lyric text was aligned with the motion footage in the edit. The lyrics are white with key words in red, marking the story's sections while keeping a period flavor; character close-ups, wide action shots and empty shots alternate, so even a dense story has room to breathe.
More Work
Ideal for brand launches · public-sector communications · cultural-tourism campaigns · IP transformation · event openers · earnings visualization · social campaigns · AI workflow development
