this AI character stole another AI character's viral video.

3 days later. same clip.
the copy got 25.7M views. the original got 20M.

meet dahab daddy. 9 days old. 222,000 followers. 106M views.

he has never been filmed. not once.

and he's not alone. 53 characters like him ws in 3 weeks.

here's where the money is:

- brand deals: his bio already has a brand contact
- AI Invasion: $1 per 1,000 views, up to $2,000 per video, from a $1,000,000 pool
- Higgsfield Earn: up to $2,500 per video
- affiliate: up to 25% of every subscription you refer
- merch or your own guide

one 10-sec video costs $0.50–$9.68 to make.

here's the exact step by step 👇

STEP 1: the concept
one exaggerated feature that reads as a silhouette. one outfit in one color. a prim look doing street stuff.
dahab = bald top + long curls on the sides + horseshoe mustache + purple 70s leather jacket.

paste this into chatgpt or claude:

"pitch me 10 original characters for short-form video. each one: a name, a one-line vibe, ONE exaggerated hairstyle readable at thumbnail size, ONE standout facial feature, an old-fashioned outfit in one dominant color, a signature move or catchphrase. deadpan, dorky, never mean. no real people, no existing characters."

STEP 2: the base photo
paste this into nano banana or gpt-image:

"ultra-photorealistic full-body studio portrait of [character], facing the camera, deadpan. plain light-grey seamless backdrop, soft even light. 85mm, f/8, head to toe in frame. natural skin texture, individual hair strands, real fabric. no text."

want to change something? never regenerate. edit:

"keep this exact person identical. same face, hair, body, outfit, lighting. change only: [one thing]."

STEP 3: the character sheet

"character reference sheet of the exact same person from the image. one wide image, light-grey background in every panel. top row: 5 full-body views (front, 3/4 front, side, 3/4 back, back). bottom row: 3 close-ups (front deadpan, 3/4, signature expression). face, hair and outfit identical in every panel. thin white dividers, no text."

STEP 4: pick the source video
this step decides the views. the source needs:
→ 1 person, 0–1 cuts, under 30 sec
→ an everyday place + real people reacting (banquet, bar, metro)
→ the person big in frame, facing the camera
→ posted in the last week

skip: pro sports, studios, cinematic shots, kids, politics

STEP 5: the swap
upload the clip + your sheet to Genjutsu, Seedance 2.5 or Kling:

"character swap edit of @ video1. replace the person in @ video1 with the character from @ image1. use @ image1 for identity only. keep everything else the same: motion, camera movement, framing, background, lighting, color grading, audio and length."

face blending with the original? blur the original face before you upload. the model can't copy what it can't see.
draft in 480p. render 1080p only for the take you keep.

STEP 6: posting (the first 72 hours decide everything)
→ character on screen from frame 0
→ 10–15 sec, no text on the video
→ keep the ORIGINAL sound
→ caption under 25 characters, first person, zero hashtags
→ post on tiktok AND instagram. the same clip did 15.3M on one and 179k on the other
→ the best video usually lands on day 2. after that growth halves every ~3 days

STEP 7: the universe
→ repeat the winning situation, not the location
→ crossovers with other characters. that's where smaller ones hit their best numbers
→ milestone gags. "100K" posts become hits on their own

the internet is about to be full of people who don't exist.
the ones who post first get paid.