LTX Text-to-Video Prompt Guide

TL;DR

4 ready-to-use text-to-video scenes: a food close-up, a vertical sports tracking shot, a product-ad punch, and an animation previz — all real ltx-2.5-fast/text-to-video clips; a 6-second 720p clip is $0.258.
Each prompt packs scene, motion, camera, and sound into one line; the links pre-fill it. Code lives in the API guide.
New accounts get 200 credits; to animate stills, use the image-to-video model page.

Scene 1: Food close-up (with ambience)

A barista pours steamed milk into a latte, creating a leaf pattern, close-up, soft morning light through a café window, gentle clinking cups and café ambience

Try this prompt →

ltx-2.5-fast/text-to-video, 720p, 6 s, 16:9, 1280×720, audio, $0.258.

How to write it: action (the pour, the leaf) + camera (close-up) + light + sound (clinking cups, ambience) — four things in one line, and the audio follows the picture.

Scene 2: Vertical sports tracking (short-video feeds)

Poster frame · Watch the full clip →

A skateboarder carves down an empty coastal road at sunset, tracking shot from behind, wind noise and wheels rolling on asphalt

Try this prompt →

ltx-2.5-fast/text-to-video, 1080P, 4 s, 9:16, 1080×1920, audio, $0.344.

How to write it: name the rig ("tracking shot from behind") for sports; the sounds the motion carries (wind, wheels) give the strongest sense of presence.

Scene 3: Product-ad punch (a 5-second opener)

Poster frame · Watch the full clip →

Product commercial: a white running sneaker drops onto wet asphalt in slow motion, water splashes outward, dramatic rim light, bass hit sound

Try this prompt →

ltx-2.5-fast/text-to-video, 720p, 5 s, 16:9, 1280×720, audio, $0.215.

How to write it: open with "Product commercial" to set the register; slow motion + rim light + one bass hit is the ad-look trio.

Scene 4: Animation previz

Poster frame · Watch the full clip →

A small robot explorer walks through a misty bioluminescent forest at night, glowing mushrooms pulse with light, camera follows at low angle, soft ambient hum and crickets

Try this prompt →

ltx-2.5-fast/text-to-video, 720p, 5 s, 16:9, 1280×720, audio, $0.215.

How to write it: world (bioluminescent forest) + character action + rig (low-angle follow); previz at 720p and upgrade the keeper to 1080p.

How to write these prompts (3 lines)

One line, four things: scene, motion, camera, sound — leave sound out and the track won't land.
Name the rig concretely: close-up, tracking shot from behind, low angle beats "cinematic."
Match duration to content: one action fits 4–6 seconds; save 20 for continuous narrative.

FAQ

Yes, audio generates by default — describing the sound (wind, clinking) keeps the track on picture; turning generate_audio off doesn't cut the price.
Yes, 9:16 gives vertical output (720×1280 or 1080×1920), ready for short-video feeds.
Yes, $10 buys about 38 clips (720p, 6 seconds each).

Next: pick the scene closest to your work, follow the link to pre-fill the prompt, and run one 720p take first.