AI Shanghai Songbird Generator — Free Preview, Then Train Your Model
See a Free Preview in Seconds, Before You Commit
Train Your Model Once, Render Shanghai Songbird in About Ten Seconds
Studio-Quality Results Without a Studio Visit
Narcis generates a free preview of your shanghai songbird from one uploaded photo. The preview streams in seconds — a quick sketch, close but not you yet. If you like it, train your model once — 12 to 20 photos, we email you when it's ready — and the full pack of nine is €4,99, rendered from your model in about ten seconds a portrait. No signup before the preview, and no waiting for an email until you choose to train.
The pack is nine professional shanghai songbird rendered in portrait aspect ratio (2:3) — the shape LinkedIn banners, CVs, and print profiles actually use. Each image shows varied crops, expressions, and light angles so you have real choices, not nine identical files.
Photos are processed on European servers. We use your photo to generate your headshots; we do not share it with third parties.
/shanghai-songbird · AI LIFESTYLE PACK
1930s Shanghai Singer Photoshoot

Your photo is used only to render this preview.
no account · ~15s · the first shot is free
Back to the Bund
627 packs · one model wears them all →all 9 shots — generated from one selfie ↓









How it works
Three steps. Upload, preview, deliver.
1
Upload 1 to 5 selfies
One is enough for the free preview. More angles sharpen the likeness.
2
See a free preview
A quick sketch of your likeness streams in seconds — close, not you yet, before you pay.
3
Train your model, then unlock 9
€4,99 one-time. Train your model once — 12 to 20 photos, we email you when it's ready — and all nine render in about ten seconds each. Every future pack and Studio portrait render the same way, no repeat training.
The deco cabaret register
Shanghai in 1937 ran its nights on a sprung ballroom floor and a lit glass dance-tower, and the woman at the silver deco microphone was that city's own invention — the era when jazz arrangements met Mandarin lyrics and a big band in white dinner jackets played behind a wall of music stands. This pack borrows that world as a visual register: art deco cabaret glamour, the hard overhead spot of period stage photography, and the amber-and-emerald colour that later Hong Kong period film made legible to everyone. It is an aesthetic taken wholesale, not a portrait of any particular performer.
The qipao does the styling work. Floor-length and sequined with a high slit, worn with long gloves and a beaded evening bag, a gardenia pinned at the temple above marcel-waved hair. Its three colourways are really three different qualities of light rather than three costumes: champagne sequin needs the hard white spot, emerald silk needs the warm ring of bulbs around a backstage mirror, black sequin under a single deco lamp needs the smoke to cross the beam.
Three situations carry the nine frames. At the deco microphone on the ballroom stage, the spot falls from directly above, the band sits in white dinner jackets behind, and the sequins pick the light up in hard points. Backstage at the mirror rimmed in bulbs, the gloves are peeled back to the wrist, the fur stole is over the chair, and the light is warm, close and unglamorous in the way that reads as real. At a ringside table after the last set, a white fox stole, a cigarette in a long holder, gramophone records stacked by the stage door, one lamp still burning. Framing runs from full-length on stage to close half-length at the mirror, and the palette holds champagne, emerald and black against warm bulb gold.
What comes back is a photograph. These are photoreal AI portraits of you in a 1930s deco cabaret register — not paintings, not illustrations, and not a styled shoot of a model in period dress. The face on the ballroom stage and in the mirror is the face you uploaded.
How to photograph yourself for a 1930s Shanghai glamour photoshoot
Better source photos make better portraits. The pack works with an ordinary phone selfie — the free preview shows you in seconds — but a few things sharpen the likeness so the singer under the spot is unmistakably you.
- Even, soft lightStand near a window with indirect daylight on your face. Overhead bulbs and phone flash flatten the features, and that is the usual reason an AI portrait misses a likeness.
- Front or three-quarter angleCamera at eye level, facing it or turned slightly to three-quarter. Stage portraiture of this era lives on that angle — no uplook from below, no overhead.
- Face fully clearHair off the face, glasses off. A clear front-facing photo is the best photo for an AI portrait because it gives the model the most of you to work with.
- A composed, level expressionThe register is poised rather than cheerful. A calm, direct gaze carries the hard stage spot far better than a wide smile.
- Plain background is fineYour own backdrop is replaced by a ballroom stage, a backstage mirror and a ringside table, so a plain wall behind you just helps the model concentrate on your face.
Frequently asked questions
- What do the nine photos actually look like?
- Nine photoreal portraits of you across three situations: at the silver deco microphone in a floor-length champagne-sequined qipao with the band behind you, backstage at a mirror rimmed in bulbs in emerald sequin with the gloves peeled to the wrist, and at a ringside table after the last set in black sequin and a white fox stole. Three frames per situation, portrait aspect ratio.
- Is this a real singer or a real nightclub?
- No. The pack is an inspired-by aesthetic — 1930s Shanghai cabaret as a costume and lighting register — and depicts no real named individual, venue, group or film property. The only person in the frame is you.
- Does it look like me, or like a generic model in a qipao?
- It is built to look like you. Whichever face you upload drives every render, so the nine come back as you at the microphone and in the mirror, not a stranger in your dress.
- Can men use this pack?
- The sample and styling are built for women — the sequined qipao, the long gloves, the gardenia — so it reads feminine throughout. A men's version of this pack isn't shipped yet, though the Shanghai gangster pack covers the same city and decade in a chalk-stripe jacket over a changshan.
- How is this different from Midjourney or Lensa?
- Text-to-image tools will generate a 1930s Shanghai singer readily enough; she just won't be you. Avatar apps return stylised images that read plainly synthetic. This pack anchors every frame to your uploaded face and keeps the output photoreal, so the likeness survives all nine.
- What if my only photo is a bad phone selfie?
- One clear selfie is enough to start, and the free preview tells you in seconds whether the likeness lands. When it misses, lighting is nearly always the reason — the prep tips above are the fastest fix.
- Can I use it as a profile picture?
- Yes, and it is one of the strongest uses. A sequined qipao under a hard stage spot makes a far more memorable avatar than another cropped photo from a wedding, and it reads as period costume rather than as a claim about your weekend.
- Can I use it on a dating profile?
- You can, honestly framed. It obviously reads as a themed glamour portrait, so it works best as a second or third photo — the one that shows taste — alongside something that shows how you actually look day to day.
- Can I use it as an album cover or a musician's press photo?
- It suits that better than almost anything else in the catalogue. The stage framing, the band behind, the deco microphone and the backstage mirror are all standard music-portrait grammar, and they arrive already lit.
- Can I print it or give it as a gift?
- Yes. The frames are portrait-format and detailed enough to print or frame, and the register works as a gift for anyone who loves deco design, jazz-era style, or old cinema.
- What happens to the photo I upload?
- Your preview photo is processed on European servers. If you train your model, those photos train only your model and are deleted the moment you delete it. Nothing is sold or shared with third parties.
When the Shanghai songbird pack fits
- For a profile picture with genuine stage presence in it
- For a social post that lands as a whole era rather than a filter
- For anyone who has always wanted the deco glamour look and never had the dress
- For an album cover, a press photo, or a musician's bio image
- For a themed party or Halloween costume planned out in pictures first
- For a portrait gift to someone who loves old cinema and jazz-era style
- For a print or framed portrait with sequins and bulb light in it
- For a phone or desktop wallpaper you actually enjoy looking at
- For a dating profile's second photo, kept honest as a themed shot
- For an author photo that needs glamour rather than a plain backdrop
- For anyone drawn to art deco design who wants to stand inside it
About narcis
Most AI portrait tools ship every pose, profession, era, and aesthetic in one giant catalog. The output looks like every other AI photo on the internet — a glaze over a stock face. Narcis goes the other way. One tool, one job: turn photos of you into portraits worth keeping. Drop a selfie and your free preview streams in seconds — a quick sketch: close, not you yet. For portraits that are truly you, train your own model on 12 to 20 photos; it trains in the background, and we email you when it is ready. Train once, render everywhere — this pack, every other pack, and the Studio all come out of that one model, about ten seconds a portrait, and it stays yours to retrain or delete. Each pack is hand-curated against actual references — paintings, photographers, registers — and tested on real faces before it ships. See a free preview before you pay, not after. Packs are one-time — no subscription to buy one; Studio, the unlimited composer, is optional at €12/month and every pack includes 30 days of it. No template generator. Built and run by one team, on our own infrastructure.
All images are AI-generated. By using this service, you confirm you are 18+, that any face photo you upload is your own or used with explicit consent. Full Terms



