Langsung ke konten
NowStarring

How to make Tung Tung Tung Sahur with AI or in ChatGPT (prompts included)

Make yourself next to Tung Tung Tung Sahur with AI: a ChatGPT prompt for a still picture, our tested video prompt for the dance, which tools take it, and the mistakes that break the log.

Baca 8 menit · Diperbarui

Jadi bintang Tung Tung Tung Sahur dengan satu foto

Buat punyamu

A picture or a video?

“Make Tung Tung Tung Sahur in ChatGPT” usually means one of two things: a still of you standing with the log, which any image model with photo upload can do, or the dance video, which needs an image-to-video model. Both prompts are below. Neither tool knows the character well by name, so both describe it.

A still picture in ChatGPT

  1. Open ChatGPT and attach one clear selfie of you, face to the camera.
  2. Paste the prompt below and send it.
  3. If the log comes out too tall or with a human face, reply “make the log shorter, its top level with my eyes, with its own carved wooden face”.

You beside the log, a still

Using my photo, make a photoreal vertical picture of me standing in a bright, empty living room with white walls and a light wooden floor, dancing with my knees bent and my arms out. On my left stands Tung Tung Tung Sahur: a short, thick, upright log of smooth golden-brown wood with a flat top, a big carved cartoon face with wide round eyes and a wide grin, thin wooden stick arms and legs, holding a plain pale-wood toy baseball bat up at its side. The top of the log is level with my eyes. Keep my face, hair and clothes exactly as in my photo. Both of us are seen head to toe. No text.

Use your own photo, or one of another adult who agreed to it. Don’t upload photos of children.

Our video prompt for the dance

This is the timed prompt we wrote for the dance, as text only: the log on the left, you on the right, ten seconds of bounces, jumps, sways and hands up. On our own page we also guide the movement with a reference clip, so expect this text-only version to need a couple of runs.

Tung Tung Tung Sahur duet, 10 seconds, vertical

A vertical 9:16 phone video, exactly 10 seconds, photoreal live action in full natural colour, one continuous shot from a phone standing on a tripod, like a TikTok duet dance video. No text, no subtitles, no labels, no logos, no watermarks. DIVISION OF SOURCES — follow this strictly: - From @Image1 take who the dancer is: face, hair, skin tone, build and the exact clothes and accessories in the photo. - @Image1 is NOT the first frame: the video opens on the wide shot below. - The partner is a cartoon wooden log character, a short, thick, flat-topped log on short bent stick legs, a little shorter than the dancer, the flat top of the log level with the dancer's eyes, never taller. It looks like this: one thick, round, upright log of smooth golden-brown wood with fine vertical grain and a flat top. On the upper half, a big carved cartoon face: wide round white eyes with dark pupils, thick carved eyebrows, a long straight nose and a wide grinning mouth. Thin wooden stick arms with simple wooden hands, short thin wooden stick legs with bare wooden feet. A plain pale-wood toy baseball bat with no writing on it. It is a cartoon wooden character, never a person in a costume, and it never takes the dancer's face. It is a lively, bouncy dancer: it bends its knees, hops and twists its whole log body from side to side, never standing stiff. - The room comes from this text: a bright, almost empty living room with plain white walls, a light grey wooden floor, a tall window on one side and soft daylight. No furniture in front of the dancers, no posters, no readable text. WHO: exactly two dancers, the dancer from @Image1 and the wooden log character, nobody else in the room. Each appears exactly once in every frame. PLACES: the log stands on the LEFT of the frame and the dancer on the RIGHT, side by side about an arm’s length apart, and they never swap places. FRAMING: the phone stays still at chest height. Both are seen head to toe the whole time, feet in frame, faces large enough to read. Both face the camera; nobody turns their back. ACTION, both on the same beat in every move, big, bouncy and goofy, full energy from the first frame: [00:00-01.5] both already dancing: deep knee bounces on every beat, shoulders and log body rocking side to side, the log's bat resting on its shoulder, the dancer grinning at the camera. [01.5-04.0] the jump: both jump straight up on each beat with both feet clearly off the floor, landing with bent knees; at the top of each jump both cross the forearms over the chest, on landing they fling the arms open wide. Four jumps. [04.0-06.0] the sway: knees bent, both swing the hips side to side in big arcs; the log tilts its whole body like a metronome and swings its toy bat out towards the LEFT edge of the frame, away from the dancer. [06.0-08.0] hands up: both spring onto their toes and throw both hands high above the head, then pump them twice on the beat, the log holding the bat straight up. [08.0-10.0] the finish: one last small hop together, a glance at each other, then a big cheeky grin into the camera. The bat is only a light toy prop: it rests on the shoulder, swings out to the left edge of the frame or points up. It stays below the dancer's chest whenever it swings towards the dancer's side, never reaches anyone's head, never touches anyone and never passes in front of anyone's face. Both faces stay fully visible in every frame. IDENTITY: the dancer is exactly the person from @Image1 in every frame: the same face, age, skin tone, freckles, hairstyle, hair length and hair colour, the exact body build, and the exact top they wear in their photo. If the photo shows only the top, the dancer wears plain light-blue straight jeans and white sneakers below it, the same from first frame to last. The hair stays the same through every jump: no ponytail, hat, cap, hood or headphones unless the photo shows them. The log keeps its own carved face, wood colour and shape in every frame; its face never becomes human and never looks like the dancer. Avoid: a log taller than the dancer, a giant log, long stilt legs, a round head on the log, a second log, a second person, a crowd, a person in a log costume, a human face on the log, the dancer’s face on the log, the log hitting or threatening anyone, a metal bat, a real weapon, the bat at head height, the bat in front of a face, swapping places, turning away from the camera, cropped feet, close-ups, camera cuts, green screen, cartoon or animated look for the dancer, a ponytail, hats, caps, headphones, rings and bead bracelets, logos on the bat or clothes, captions, text, watermarks, extra arms, extra legs, extra fingers, distorted hands, morphing faces. Sound: none needed; the soundtrack is added later.

  1. Open the image-to-video or reference mode of a tool like Kling, Seedance, Veo or Sora.
  2. Attach one sharp selfie. If the tool doesn’t use “@Image1”, replace it with “the person in my photo”.
  3. Set 9:16 and 10 seconds, or the closest option, and turn off any music the tool adds.
  4. Generate, compare with the frames below and fix what’s off with the table further down.
  5. Add a Tung Tung Tung Sahur dance sound when you post.
A laptop on a desk at night showing an AI video tool, with a reference selfie on the left, a long prompt in the middle and a preview on the right
One selfie attached, prompt pasted, vertical

Why the prompt is written this way

  • The size line. Left alone, models make the log a head or two taller than you, on long stilt legs. Tying its flat top to your eye line keeps it the squat, comic size from the meme.
  • The bat lines. In our first runs the bat swung across the frame at head height. Now it rests on the shoulder, swings out to the far edge or points up, and never crosses a face.
  • Fixed places. Saying who stands where, and that they never swap, stops the two from trading sides mid-dance.
  • Head to toe, phone on a tripod. Without it, models cut to close-ups and crop the jumps.
  • The identity block. Jumping makes hair and faces drift. Repeating that the hair, face and top stay the same from first frame to last holds you together.
  • A silent render. Models invent their own music; it’s cleaner to add the trend sound yourself.
The wooden log with a carved grin holds its toy bat up on the left; a red-haired woman in a mustard sweater and light jeans dances on the right, in a bright living room
0:01: side by side, bat up
The woman and the log bounce with bent knees and fists up, the bat down at the log’s side
0:03: the bounce
The log swings its toy bat low and out to its own side while the woman sways beside it
0:04: the bat swing
Both groove on the same beat, the woman’s arms out, the log’s stick knees bent
0:07: the groove
The woman springs onto her toes with her arms flung out while the log holds its bat
0:08: the jump

Fixes for common results

What you getChange
The log is a giantRepeat the eye-line rule and add “a little shorter than the dancer” to the first line
The log has a human face, or yoursAdd “a carved wooden cartoon face, never human” and attach only your photo
The log barely movesAdd “a lively, bouncy dancer that bends its knees and twists its whole body”
You barely leave the floorWrite “both feet clearly off the floor” into the jump line
A cap or ponytail appearsAdd “no ponytail, hat or cap unless the photo shows them”

Rather skip the retries? The Tung Tung Tung Sahur page makes it from a selfie with the moves and the sound already in. The step-by-step guide shows what you get.

NowStarring is an independent fan tool, not linked to the creators of the Tung Tung Tung Sahur character, the makers of any green-screen clip or video of it, or the owners of any sound used with the trend.

Pertanyaan

Can ChatGPT make Tung Tung Tung Sahur?

ChatGPT’s image tool can make a still picture of you beside the log from your photo; the prompt is in this guide. It doesn’t give you the dance itself, which needs a video model.

What is the best AI for a Tung Tung Tung Sahur video?

Any image-to-video model that takes a reference photo, such as Kling, Seedance, Veo or Sora. What matters more is the prompt: the log’s size and the bat need spelling out, or most models get them wrong.

Why does the AI make the log huge?

Models picture “a tall log” as taller than a person. In our tests even “the same height as the dancer” came out a head taller; tying the top of the log to the dancer’s eye line is what held.

Why does the bat swing at my head?

Because nothing told the model where it goes. Our prompt keeps the bat below chest height on the dancer’s side and swings it out to the far edge of the frame.

Can I use a reference picture of the log?

You can, but many tools then mix the two references and put a carved face on you, or your face on the log. Describing the log in words and attaching only your photo was more reliable for us.

Ada kursi
dengan namamu di atasnya

Buat videoku