← all free guides
guide
Loading the guide…
← all free guides
guide
Loading the guide…
Tools
Turn one photo of yourself into the viral 17-second café video: you sitting on the wrong platform, a woman walks in, and you quietly realize your mistake. One prompt, Seedance 2.5, 1080p, 17 seconds.
Loading the guide…
You can now make the viral café video — the one where you're sitting in the wrong spot, a woman walks in, and you quietly realize your mistake — from one photo of yourself and one prompt.
The whole thing is a 17-second, single static shot: no cuts, no camera movement, no dialogue. The awkwardness does all the work. You upload your photo to Higgsfield, paste the prompt, set the model to Seedance 2.5, and generate.
This walkthrough covers the full setup: the photo, the prompt, the exact settings, and what to check before you post.
Take a clear, front-facing photo of yourself — this is the reference the model builds the whole video around.
Your face, hairstyle, skin tone, and glasses (if you wear them) stay identical throughout the video — that's what the prompt locks in.
Upload the photo from Step 1, then paste this prompt in exactly as-is — don't summarize it or trim the timing beats, they're what make the video work:
VIDEO: 17 seconds, vertical 9:16, photorealistic, one continuous static wide shot, natural smartphone-video realism, silent. The camera never moves, never zooms, no cuts.
REFERENCE: The main character is the person in the reference photo. Keep their exact face, hairstyle, skin tone and any glasses identical throughout. They wear a plain white crew-neck t-shirt, loose-fitting straight black jeans and plain black low-profile leather sneakers. They are extremely understated and deadpan; all emotion is shown through tiny eye and head movements only.
SETTING: A quiet, minimalist architectural café with a warm concrete interior. A straight wall of medium warm-gray cement plaster, split into a left bay and a right bay by one thick, flat, rectangular concrete divider slab (slightly lighter beige-gray) running from above the top of the frame down to the ledge. A continuous raised concrete ledge about 50 cm high runs along the wall; its top is flat and clear. Smooth matte sandy-brown concrete floor. Soft even daylight, warm earthy palette, no wooden furniture, no plants.
Two identical small flat round brushed-metal disc platforms are bolted to the front of the ledge, one in each bay, each on a slim vertical silver support with a curved J-hook bracket. They work like small wall-mounted café tables. The platforms never move or slide. Nothing else is in the room.
CAMERA: Standing eye level, several meters back, on the right side of the room, only about 15–20 degrees off the wall (nearly straight-on, just slightly angled). Both people, both platforms, the divider, the ledge and the floor stay in frame the whole time.
SECOND CHARACTER: One stunningly pretty, glamorous adult woman with long silky dark hair, flawless glowing skin, soft glam makeup, full glossy lips and a delicate gold necklace. She wears an oversized charcoal-gray tailored blazer over a dark fitted top, matching very loose extra-wide-leg charcoal trousers, a small black shoulder bag, and carries a brown takeaway coffee cup with a white lid. She looks calm, polished and self-assured.
ACTION (follow this order exactly):
0:00–0:02 — The main character is already sitting on the LEFT metal platform as if it were a chair (the wrong place to sit), astride it, facing the wall, so we see their back and right side. Knees apart, sneakers flat on the ledge, a brown takeaway cup in one hand, feet still. They glance only slightly to the right, a soft casual glance, not a stare. Around 1 second in, the woman starts walking into frame from the right edge.
0:02–0:04 — The woman walks in from the right toward the right bay, looking straight ahead as if nobody else is there. The main character turns their head and looks directly at her, following her with their eyes. Their body stays facing the wall. Neutral, curious, no dramatic reaction.
0:04–0:06.5 — She sits down on the concrete ledge in the RIGHT bay (on the ledge, not on the platform), settles naturally, places her cup on the right platform in front of her and crosses her legs, one knee over the other. She keeps her legs crossed from now on. The main character keeps watching her.
0:06.5–0:08.5 — She looks straight ahead the whole time, relaxed, sipping her coffee, blinking, breathing naturally. She never turns toward the main character and never makes eye contact. The main character looks up at her with chin slightly down and eyes lifted, like someone not sure they are allowed to be there.
0:08.5–0:10.5 — The main character's eyes drop to the platform under them and stay there for a beat, then slowly come back up to her. A tiny pause: a slight narrowing of the eyes, a very small change in expression. They quietly realize they are in the wrong place. Subtle and deadpan. No shrug, no hand over the mouth, no "oh no” face.
0:10.5–0:13 — They lean slightly forward, then get up from the left platform in a slightly awkward half-crouch, one hand steadying on the ledge, cup held level. The platform gives one small wobble as their weight leaves it and settles back in the same place. Human, slightly uneven timing, not robotic.
0:13–0:14.5 — They turn to face forward and lower themselves onto the concrete LEDGE in the left bay, now sitting correctly, with the platform in front of them like a small café table, a leg on each side of its support with clear space around it. They set their cup on the platform.
0:14.5–0:17 — Both are now sitting correctly on the ledge, side by side with the divider between them, each with a cup on their own platform. The main character faces forward, calm and completely neutral, pretending nothing happened, with one tiny easy exhale. The woman sits with legs crossed, takes a small sip, looks straight ahead. Both keep breathing and making tiny natural movements. Hold the awkward deadpan final shot.
RULES: Exactly two people. The woman always enters from the right and never looks at the main character. The main character never shrugs, never touches their face, and keeps their feet mostly still. The platforms never move (except the one tiny wobble). Camera is fixed. No speaking, laughter, music, captions, logos or subtitles.Before you hit generate, check these three settings — they're what make it come out like the original:
If any of them is on a different value, the timing beats in the prompt won't line up with the video — the realization and the move happen at fixed seconds, so the length has to be exactly 17.
Click generate and watch for the beats in this order:
If it comes out wrong, regenerate — don't rewrite the prompt. The most common misses:
Post the 17-second clip as-is — no captions, no music, no text on screen. The silence is part of the joke.
I make guides, tips, and content about AI and tech — the tools worth using and how to actually set them up.
TikTok — @haejuicyy
If you found this helpful, follow for more! 😛
Free guide
Drop your email and it's yours — plus the newsletter when new guides land.
Want the prompt and the base video sent to you? In the video I say: comment "cafe" and I'll send it over.
Two or three generations is normal before one lands clean.