Culture & Arts
Following the lyrics from myth through the dynasties to the modern city, Chinese AI models generate grand scenes and close-ups of people, strung together into one continuous music video.
01 / Background
The lyrics of “Five Thousand Years” span mythology and many historical periods. The visuals had to change quickly to keep up with the song, while still letting viewers see how one section leads into the next. The project went with a Chinese-style fantasy look, using the grand scenes, lighting and fantasy spaces that AI handles well to carry such a long stretch of time, and broke the lyrics down shot by shot into characters, settings and actions.
In production, every still image was finished and reviewed together before any video was generated. Only Chinese models were used: SenseTime SenseMirage for stills and mainly Kling for video, with faces fixed, black borders removed and images adjusted shot by shot. The look and composition of each character were locked at the still stage, and the motion was then built around those locked subjects.
The film opens with mythic images such as Pangu and Nüwa, moves into palace cities, armies and historical figures, and in the final section turns to red flags, the Forbidden City, modern buildings and warships at sea, before ending in a fantasy space. The images unfold an imagined history along with the song; the priority was making the mood, the scale of the scenes and the rhythm flow from one section to the next.
02 / Creative Highlights
Where the myth section ends, the script called for golden curtains, brush writing and the opening of the dynasties. In the film, curtains first drop from the clouds, then the shot cuts to a brush writing and a figure at a palace gate, turning “a new era has begun” into a visible action, so viewers know at once that a new chapter has started.
There were already plenty of court scenes before and after, so the Three Kingdoms line uses a 3D map. In the film, wide views of the terrain alternate with the cavalry and banner formations that follow: the map first shows who holds which territory, then the people and armies clash, avoiding a monotonous run of palace shots.
City gates, armies and brightly lit palace cities convey the grandeur of each era, while faces, figures seen from behind and hands writing give the emotion somewhere to land. Alternating wide and close shots gives the images the sweep of a fantasy epic while always bringing viewers back to real people.
The chrysanthemum section moves from a figure in a sea of flowers to a single giant orange-gold bloom, with scattering petals and particles swirling around it. The contrast between the huge flower and the detail in the foreground gives this lyric a striking visual impression.
03 / Scene Breakdown
A golden sun and a great mountain open the film, Pangu swings his axe, Nüwa holds up clay, the Yan and Yellow Emperors fight by the Yellow River, and Cangjie, Hou Yi, Jingwei and Gonggong appear one after another: one myth for each lyric line.
The sun above a sea of clouds pushes in to a golden mountain, then Pangu swings his axe in a fiery vortex, with a close-up of the giant axe sending rocks flying.
Cangjie stands before a boulder carved with characters; Hou Yi draws his bow against the light, followed by a burning golden bird; then Jingwei in a blue feathered robe and a blue bird spreading its wings.
Red-haired Gonggong bursts out of a huge wave, a sky-high tree towers over the sea of clouds, then falls and becomes a golden road that someone walks along.
Golden curtains fall and part again, a brush writes history, and the story sings on from Yao and Shun through the Shang and Zhou, the Spring and Autumn period and the Hundred Schools of Thought.
Two golden curtains hang from the clouds, a brush writes on bamboo and silk, and when the curtains part again, a dynastic palace city is already behind them.
A ruined city at sunset, the back of the last king, King Zhou kneeling on one knee propped on his sword amid the flames of war, then high walls in a sea of fire.
An old man in white fishes in the snow, morning light falls on a sprawling royal city, then figures among clouds and mountains.
A beauty on the city wall glances back, an army raises banners under the beacon fires, and finally a close-up of the king and the beauty leaning on each other.
A teacher lectures with a book in hand in a schoolroom, a giant statue of Confucius stands in a temple, then Laozi rides through purple clouds.
Shang Yang's pole of trust, returning the jade intact to Zhao, standardized axle widths, the Great Wall and the First Emperor of Qin, then the Chu–Han contention, Emperor Wu of Han and the Champion Marquis, with close-ups of rulers alternating with grand scenes.
A strong man carries a log toward the city gate, two hands present a glowing jade disc, and Qu Yuan, his back to the camera, gazes at the river at sunset.
Carriages race down a straight road, then the Great Wall in autumn colors, then the First Emperor of Qin in his great hall and a giant figure in green lightning.
Lightning strikes beside a figure, a tower ship breaks through the waves in a thunderstorm, then a crowd raises their arms and shouts.
Army camps and rivers in firelight, a figure with a sword seen from behind by the Wu River, Consort Yu and Xiang Yu bowing their heads in farewell, then Liu Bang firmly seated in Xianyang.
A great hall before a statue of Confucius, a camel caravan in the desert, a young general charging into the enemy ranks, and a brush writing the history books.
A court in opera face paint and a 3D map tell the Three Kingdoms, the Grand Canal and the imperial exams tell the Sui, the empress and the Wordless Stele tell the Tang, and finally Huang Chao's chrysanthemum poem.
The whole court wears Peking Opera face paint, commoners gossip in a teahouse, a boy emperor sits on the dragon throne, and finally a 3D map marking the three territories.
Yang Guang stands at the bow of a dragon boat, scholars bend over their exam papers, and trackers on the riverbank haul a great boat.
A brightly lit street packed with people, the empress in red walking into the great hall and taking the golden throne, and finally the Wordless Stele on a mountaintop.
A figure stands in a sea of golden chrysanthemums as the camera slowly pushes in.
A giant orange-gold flower opens in the middle of the sea of blooms, petals and sparks flying out without end.
Armies in golden armor hold up their banners, filling the square before the palace city, and the firelight turns the whole frame gold.
The yellow robe placed on the founder, Yue Fei, Genghis Khan, Zhu Yuanzhang, the Tumu Crisis, the emperor guarding the gates of the realm, then the Qing army entering the pass and the closing of the country, with warm and cool colors switching back and forth by dynasty.
Generals raise golden cups, a figure in a red robe looks back at the city against the setting sun, then an old general with a full head of white hair.
A kneeling statue bound in iron chains, a general on horseback, a red-robed figure seen from behind walking through an army in the snow, and finally a row of golden command tablets.
Cavalry charge across a plain, a close-up of a Mongol general's face, a shaven-headed monk in a run-down lane, then mountains and rivers at dusk.
A lone figure beneath an ancient tree, a dead tree and palace in firelight, a stele of ancestral instructions in gold characters, and a palace gate in a sea of fire with troops closing in from every side.
A 3D map spread across a chess table, smoke billowing, a general in red charging out on horseback, and a beauty glancing back.
Warships set out to sea in the fog, a stone stele among distant mountains, and finally a firmly shut gate.
Pangu reappears in flames, red silk sweeps over the city, the Forbidden City and the Great Wall give way to modern towers, warships and rockets, and the film comes to rest on a figure beneath a yin-yang ring.
Pangu in flames, cavalry carrying red flags, great sheets of red silk drifting over the city, then a city sparkling with lights.
Farmers in terraced fields, the square of the Forbidden City, the Great Wall in sunshine, then a modern city skyline and the sunrise.
Warships sail toward the sunset, a rocket rises through the clouds, the city at night, and finally a space station in orbit.
Firelight reflects on the water, warriors roar, and at the end a lone figure stands in the water beneath a glowing yin-yang ring.
04 / Storyboards & Final Frames
The film has more than 160 shots. Here, 42 lyric lines are picked in the film's order: for each shot we first describe what it shows, then show the final image used in the film.
Shot 1.1
Before the singing starts, two establishing images: a golden sun above a sea of clouds, then a push in to a glittering golden mountain that fills the frame.
Shot 1.2
First a full-length shot of Pangu swinging his axe in a chaotic vortex, then a cut to a close-up of the giant axe, with rocks and fire bursting outward.
Shot 1.3
Nüwa looks down at the golden clay in her hands, her headdress and earrings swaying gently, with mist and green mountains behind her.
Shot 1.4
Cangjie leans on a wooden staff before a boulder carved with characters. With the man and the carvings in one frame, “creating writing” is clear at a glance.
Shot 1.5
Hou Yi draws his bow fully against the sun, and the next shot shows a bird burning with golden light, standing for the sun he shot down.
Shot 1.6
Jingwei in a blue feathered robe shares the frame with a white bird, and in the next shot becomes a great bird of blue light flying over a village at night.
Shot 1.7
Red-haired Gonggong appears full length, bursting out of a towering wave with fierce movement.
Shot 1.8
The ancient sky-high tree falls, and a golden road unrolls from its hollow, leading into the next line, “step by step, we…”
Shot 2.1
Two golden curtains hang from the clouds with the sun between them. The script said “the myths are over, now history begins,” and here it is staged as a curtain falling on a stage.
Shot 2.2
A brush writes on bamboo and silk covered in ancient script, with golden specks of light drifting upward.
Shot 2.3
The curtains part again to reveal a dynastic palace city, with a figure standing at the gate. The script specified an “opening” transition here.
Shot 2.4
King Zhou kneels on one knee, propped on his sword amid the flames of war, staring into the camera, his gaze and the firelight both heavy and dark.
Shot 2.5
An old man with white hair and white robes sits on a stone in a snowy stream, fishing, with snowy mountains behind him.
Shot 2.6
The beauty on the city wall turns to look at the camera as red petals drift in the wind.
Shot 2.7
Sunlight streams into the hall of a Confucian temple, where a scholar stands beneath a giant standing statue of Confucius in a hall of great depth.
Shot 3.1
A strong man carries a huge log toward the city gate, with the act of carrying placed in the center of the frame.
Shot 3.2
A straight road runs to the horizon, carriages racing down the middle, with neat rows of cavalry on both sides.
Shot 3.3
The Great Wall on autumn ridges, with clouds pressing on the mountainsides; a wide shot conveys the “ten thousand li.”
Shot 3.4
A giant figure rises in green lightning above a tiny crowd below.
Shot 3.5
In a thunderstorm, a huge tower ship pushes forward against the waves, leading into the next line, “asking what eternal life could mean.”
Shot 3.6
Consort Yu and Xiang Yu bow their heads in farewell, foreheads almost touching.
Shot 4.1
The ministers at court all wear Peking Opera face paint. A white face already means a treacherous official, so no expressions are needed.
Shot 4.2
City towers stand on a 3D terrain map marking the three territories, setting it apart from the court scenes before and after.
Shot 4.3
Yang Guang stands at the bow of a dragon boat, in the foreground, with water-town houses lining both banks.
Shot 4.4
Wu Zetian sits on a golden throne, facing the camera head on, with commanding presence.
Shot 4.5
A blank stone stele stands on a mountaintop, with the setting sun and a sea of clouds behind it, all in warm colors.
Shot 4.6
A figure stands among mountains of golden chrysanthemums; from this line, the film enters Huang Chao's chrysanthemum poem.
Shot 4.7
A giant orange-gold flower opens in the middle of the sea of blooms, petals and sparks flying everywhere, its warm colors linking with the sections before and after.
Shot 4.8
Armies in golden armor hold up their banners, filling the square before the palace city.
Shot 5.1
A figure in a red robe, back to the camera, looks at the city under the setting sun.
Shot 5.2
A kneeling statue bound in iron chains, seen from behind with no face shown, expresses “spat upon.”
Shot 5.3
A Mongol general in a fur helmet, his face in a close-up that fills the frame.
Shot 5.4
In heavy snow, several people are chained together, with red plum blossoms and a palace gate behind them, telling of an injustice.
Shot 5.5
A stele carved with ancestral instructions in gold characters stands before a palace hall as the main subject of the frame.
Shot 5.6
A 3D map lies spread on a chess table, surrounded by chess pieces and weapons; the map itself is the chessboard.
Shot 5.7
Qing warships set out to sea in the fog, several ships lined up one behind another.
Shot 6.1
Pangu reappears in flames, axe in hand, joining the five thousand years end to end.
Shot 6.2
Great sheets of red silk sweep over the city, and the whole frame is dominated by Chinese red.
Shot 6.3
Neat ranks of an honor guard line up on the square of the Forbidden City.
Shot 6.4
An ancient pagoda hides within a modern city skyline, past and present in one frame.
Shot 6.5
A rocket rises through the clouds, with the moon hanging beside it.
Shot 6.6
In the sunset, a general on horseback and the army behind him become silhouettes, banners fluttering in the wind.
05 / Results
The final film runs about 4 minutes 39 seconds at 1920×1080. With permission from the song's creator Xiao Shaoyan, it was released on Kuaishou credited to “Zhongzhong × Xiao Shaoyan,” reaching 600K+ views, with 10K likes shown at the time of the screenshot.
Case Study
From Pangu splitting heaven and earth to a rocket lifting off.
A song just over 4 minutes long, with five thousand years of myths and dynasties packed into its lyrics.
In summer 2024, with permission from the song's creator Xiao Shaoyan, we gave it images made entirely with Chinese models: SenseTime SenseMirage made the stills, and Kling turned the stills into video.
The film went out on Kuaishou and drew 600K+ views.
01
“Five Thousand Years” moves fast, with a new line roughly every second and a half, and each line is about a different person or event. The final film cuts to more than 160 shots.
We broke the whole song down by lyric line into a shared document: each line got its own subheading, with what the shot shows and which prompt to use written underneath, followed by the generated images and their revised versions. More than ten creators on the team each claimed a section and owned it.
We locked the style first: mythic fantasy, in a 16:9 frame. We leaned on the grand scenes, lighting and fantasy spaces that AI does well, and steered clear of the fiddly movements it does poorly.
A golden sun and a golden mountain, then Pangu swinging his axe and a close-up of the giant axe: the first four seconds establish the grandeur of myth.
02
Everyone made only images at first, no video. All the stills went into the document for one shared review, and only those that passed were used to generate video.
Problems surfaced at the still stage: a broken face got fixed first, black borders were filled in, and a weak composition was redrawn. Once the images were locked, the video only had to work around the characters and compositions already set.
Take “First Cangjie created writing”: the first draft was a close-up of an old man writing at a desk, which didn't read as “creating writing.” The final version shows Cangjie leaning on a staff before a boulder carved with characters, the man and the writing in one frame, so viewers get it at a glance.
Shots of characters were chosen by the same standard: Gonggong's first draft was a half-length portrait in front of snowy mountains, so the final was redrawn to the brief “full length, water, red hair,” with him bursting out of a huge wave; Wu Zetian, first shown from behind surveying the land from a mountaintop, was changed to face the camera on a golden throne; “All for one backward glance from a beauty” went from a figure seen from behind on the city wall to a face turning back toward us.
03
After Qin and Han, the song covers a dynasty or a historical figure in every line. If all of them were drawn as palaces and ministers, they would soon look identical.
For the Three Kingdoms line, “Then the realm was split in three,” the first draft was a cavalry melee. In review we saw there were already too many court scenes before and after, so it became a 3D map: mountains, rivers, city towers and red lines lay out the three territories on a table, and “split in three” is clear at a glance.
The same approach was used in many places: “This realm is like a game of chess” puts the map on a chess table, “On the Wordless Stele, not a single line” simply shows a blank stele, and “So he honored the words of his ancestors' teachings” swaps the palace for a stele of ancestral instructions in gold characters.
Some lines took several approaches before they were settled. “This realm is like a game of chess” tried Go, Western chess and a glowing grid before landing on “the map as the chessboard”; “To this day, still spat upon” went through more than twenty images, and the final choice was a kneeling figure in chains with its back to the camera, without drawing a specific face.
04
Some images are hard for AI to get right directly. “These thieves who stole the realm, filling the whole court” needed the sinister expressions of treacherous ministers, and after several attempts the faces still weren't right.
So we changed tack: every minister at court wears Peking Opera face paint. On stage, a white face already means a treacherous official, so viewers understand without needing to read expressions.
“A pity he was never made a god” was handled the same way. The line is about the First Emperor of Qin; revisions tried the Terracotta Warriors and a golden statue, and the final version is a giant figure rising in green lightning above a tiny crowd, turning “almost becoming a god” into a single scene.
05
The song has a few turning points, and we designed images specifically for them.
Where the myths end and history begins, the lyrics go “When myth takes its final bow, real history only begins, raise the curtain.” So we actually built a curtain: golden curtains drop from the clouds, a brush writes on bamboo and silk, and when the curtains part again, a dynastic palace city is already behind them.
The curtain falls, the brush writes, the curtain opens: three shots complete a change of chapter.
When the song reaches Huang Chao's chrysanthemum poem, the image pushes from mountains of golden chrysanthemums to a giant orange-gold flower, with petals and sparks scattering all around, then cuts to armies in golden armor filling the palace city. These few shots are the most richly colored stretch of the whole film. For the giant flower we also tried translucent blooms in blue and red, but settled on warm colors to link with the sections before and after.
One giant flower fills the frame, petals and sparks flying outward.
The ending sings back to today: red silk sweeps over the city, the Forbidden City and the Great Wall give way to modern towers, warships set out to sea, a rocket lifts off, and finally a figure stands on the water beneath a yin-yang ring as the song ends on “Watch me remake the world with one stroke.”
After the warriors' roar, the image comes to rest on a figure beneath the yin-yang ring.
06
This film used only Chinese models from start to finish: every still came from SenseTime SenseMirage, and the video was mainly generated with Kling. In 4 minutes 39 seconds and more than 160 shots, it covers myths, ancient wars, palaces and modern cities.
It shows that for AI films with a long time span and many shots, the key lies not only in the model but in the pre-production: breaking the lyrics into shots, having several people work on sections in parallel, reviewing the images before making video, and switching to a different approach wherever something can't be drawn well.
The film went out on Kuaishou, earning 600K+ views, with 10K likes at the time of the screenshot. AI films on history, culture or tourism can all follow this same process.
FAQ
Yes. All the stills for “Five Thousand Years” were generated with SenseTime SenseMirage and the video mainly with Kling. The 4-minute-39-second, 1920×1080 film has more than 160 shots covering myths, ancient wars, palaces and modern cities, and after release on Kuaishou it drew 600K+ views and 10K+ likes.
We broke the lyrics into shots line by line in a shared document, spelling out the image and prompt for each line, and team members claimed sections. Everyone made stills first, which were reviewed together before any video was generated; the style was set in advance as mythic fantasy, 16:9.
Wherever possible, change the approach: a 3D map for the Three Kingdoms, opera face paint for the treacherous ministers, the map on a chess table for “This realm is like a game of chess,” and a blank stele for “the Wordless Stele.” Palaces and courts appear only when needed, interspersed with close-ups of people and props.
This project took about a week from forming the group and dividing the work to release on Kuaishou: first the lyrics were broken into shots and several people made images in parallel, then after review and retouching, the video was generated and edited all together.
Approach
We first listed the shots by lyric line, writing out each shot's characters, costumes, setting and action in detail, then reviewed the stills together and revised them shot by shot, including fixing faces and handling black borders.
Shots of golden curtains, writing and maps divide the chapters, while wide views of palace cities, close-ups of people and fluttering banner formations carry the rhythm; warm firelight, golden light and weighty set designs recur throughout, giving every era the same fantasy feel.
More Work
Ideal for brand launches · public-sector communications · cultural-tourism campaigns · IP transformation · event openers · earnings visualization · social campaigns · AI workflow development
