Going Out to Eat — 15s Photoreal POV — AI Video from Neta Studio

Going Out to Eat is a 15-second photoreal video shot entirely from the diner's own eyes. One man takes himself out for a proper dinner: he sets his phone and keys on the wood, a server lands a heavy plate of braised oxtail over rice with collard greens and cornbread, he eats the first slow bite, sips his sweet tea, and settles in — and the city drifts past the glass the whole time.

It is documentary-real, hand-held and vertical, trapped in one seat beside one window: warm tungsten pooled on dark wood, a deep-blue summer dusk on the street outside, a phone camera catching the moment like a saved memory. There is no face in frame — only dark-brown hands, black satin sleeves, and one table for one.

Watch it now: Going Out to Eat →

9:16 vertical · 15 seconds · 720×1280 — rendered with Seedance 2.0 Fast (image-to-video) from the world's own reference frame, in a single continuous pass.

Download the MP4 →

15 seconds. Shot on a smartphone like a real saved memory, documentary-real handheld footage, vertical. Interior of a cozy neighborhood restaurant at night, one corner table for one beside a tall plate-glass window; warm tungsten pendant light pooled on a worn dark-wood table, cool deep-blue summer dusk street outside the window with amber streetlights.

The attached reference_image shows this exact table, window, dusk street, folded napkin, and warm lamp light; every shot keeps this same table, this same window, this same warm-on-cool light.

Event: a man alone treats himself to a proper dinner — he sets his phone down, a server brings the plate, he takes the first slow bite, sips his sweet tea, and settles content while the city passes outside the window.

Spatial blocking: the camera is the diner's own eyes, seated at the table, his dark-brown hands and black satin shirt sleeves entering the bottom of the frame; the window is straight ahead; other tables blur at the far left and right frame edges; a server's hands and dark apron enter from the left edge.

Camera: gentle handheld POV with natural micro-motion, and one slow pan toward the window late in the clip.

Beat timeline:
0.0 to 3.0 seconds: the camera settles at the table as his dark-brown hands set a black phone face-down and a set of car keys onto the wood, then lift away; the dusk street beyond the window stays steady. Hard cut.
3.0 to 6.5 seconds: a server's hands set a heavy white plate down in front of the camera — braised oxtail over rice, glossy collard greens, a square of cornbread — the plate lands with a soft thud and steam rises through the warm light; the hands withdraw to the left. Hard cut.
6.5 to 10.5 seconds: his right hand lifts the black fork, loads rice and a piece of falling-apart meat, and carries it up and out of the bottom of the frame; one beat of stillness while he eats; a soft satisfied exhale; the fork lowers back toward the plate. Hard cut.
10.5 to 13.0 seconds: his left hand closes around the cold sweating glass of sweet tea, lifts it out of frame for a sip, then sets it down with a soft clink; the camera pans gently to the window as a city bus glides past and headlights sweep across the glass. Hard cut.
13.0 to 15.0 seconds: his hand lays the fork across the plate and pats the table once; the camera lifts slowly from the plate to the window, where warm streetlights drift past; slow fade to black.

Sound: sound effects only, no background music — low restaurant room tone with distant forks and low voices, the soft thud of the plate, one quiet satisfied exhale, the clink of the glass, a muffled bus passing outside the window. No spoken words, no dialogue, no narration.

Keep stable across every shot: the same table, the same plate of food, the same glass of sweet tea, the same window and dusk street, his dark-brown hands and black satin sleeves. The camera-wearer's face never appears. Realism: slightly missed focus, mild compression, natural motion blur, no HDR, no beauty filter. Avoid: no duplicate person, no extra hands or limbs, no hand clipping, no visible text, no logos.

Prompt as submitted to Seedance 2.0 Fast, together with the world's own reference frame as reference_image.

Professional — The cloud harness supports the whole path from a one-line world to a finished cinematic short, entirely in the cloud with no local setup. Stable access to the latest models: Claude Opus 5 · GPT Image 2 · Seedance 2.0 Fast.

One-click generation — Officially tuned skills drive every step: world building, reference frames, prompt authoring, image-to-video rendering, and instant publishing to a shareable link.

Community & IP-first — Adapt any world or start fresh from a single sentence. Keep your frames, prompts and key art — remix them into new episodes whenever you like.