How to prompt PixVerse: three sentences, and the four words to delete

PixVerse reads a prompt in order, and it weighs the earliest words most. That single fact decides how the whole prompt should be shaped. On LUVI the family runs as v6 and c1, each in text-to-video, image-to-video, reference-to-video and start-end, with a video-extend model on v6. Clips run from 1 to 15 seconds at 360p, 540p, 720p or 1080p, and sound is generated with the picture. On September 22, 2026 a five-second v6 clip quoted 350 credits at 360p, 600 at the 720p default and 1,150 at 1080p; c1 quoted 650 at 720p.

How long should a PixVerse prompt be?

Long enough to name everything that matters and no longer. PixVerse's own documentation asks you to "keep prompts between 25-200 words for optimal results", and it illustrates the floor with a pair of examples: the good one is "A serene mountain lake at sunrise, with mist rising from the water. The camera slowly pans across the lake, capturing the golden sunlight reflecting on the surface. Birds fly overhead as the morning unfolds." The weak one is "Mountain lake sunrise".

LUVI's prompting guide for the family narrows the band further, to about 50 to 80 words for text-to-video, because past roughly 200 words the prompt starts to dilute its own control. Treat 25 as the floor, 80 as the working target and 200 as the ceiling you should rarely approach.

What is the three-sentence shape?

It is the structure that matches how PixVerse reads: identity first, technical cues last.

• Sentence one. The subject, the traits that make it this one, a single action, and where it is.
• Sentence two. One camera move, named simply, plus the lighting, lens and composition cues.
• Sentence three. The constraints—written as things that stay intact, not things to avoid.

A brass desk fan turns slowly on a cluttered workbench as dust drifts through the air in front of it. Steady lateral tracking, low afternoon side light, 35mm, shallow depth of field, quiet workshop background. Blade shape stays stable, hands remain natural, audio includes a low motor hum and distant traffic.

One action per clip, and one camera move. Competing verbs produce artifacts, and two moves in one shot break the motion apart. If you need speed, show it as physical evidence—motion blur, streaking lights—rather than repeating the word "fast".

Which words should I delete?

Four of them, and they are the ones people reach for first. LUVI's prompting guide for PixVerse singles out "cinematic", "beautiful", "epic" and "professional" as words that sample too broadly: each one asks the model for an average of everything it has seen under that label. Replace each with a physical cue you can point at—"warm rim light", "cyan-magenta contrast", "anamorphic 2.39:1", "one-point perspective".

"Realistic" is a fifth word to drop, for a different reason: realism is PixVerse's unstyled base, so writing it adds nothing. Name the look you want instead, or leave the slot alone.

How do I exclude something?

By describing the state you want. Neither the v6 nor the c1 models on LUVI expose a negative prompt field, so every exclusion lives in the prose—and in prose, naming a defect can summon it. "No bent fingers" puts bent fingers in the prompt; "hands remain natural" does not. The same conversion works for everything else: "product silhouette stays intact", "cup shape remains stable", "sharp focus throughout". The help center lists which models do have a negative prompt, and PixVerse is not among them.

What do the parameters do?

Five, and two of them are unusual for this catalog.

• sound is on by default. It decides whether audio is generated at all; what that audio contains still has to be written into the prose, as content ("audio includes soft room tone and faint spoon clink"), never as a promise.
• seed exists here, which is rarer than you would expect—most video families in the catalog have none. Reuse one to keep a shot reproducible while you change a single clause.
• quality steps through 360p, 540p, 720p and 1080p, and it moves the price: the same five seconds went 350, then 600 at the default, then 1,150 at the top.
• duration runs from 1 to 15 seconds, 5 by default.
• aspect_ratio offers eight values, including 9:16 for vertical and 21:9 for a wide crop.

None of them belong in the prompt text. Writing "16:9" or "10 seconds" into the sentence spends words on something a control already carries.

What changes for image-to-video, transitions and extends?

The image does the describing, so the prompt stops describing. With a starting frame, composition, wardrobe, subject and light are already locked: write the motion, the camera move and one stability rule, and never re-describe what is visible. With a start and an end frame, prompt the transformation between them rather than either endpoint. With an extend, write only what happens next—recapping the source clip gives the model nothing it does not already have.

How we checked

• Models: PixVerse v6 Text-to-Video and c1 Text-to-Video, with the other modes read from the catalog.
• Settings: five-second clips at 360p, 720p and 1080p, 16:9.
• Date and what was read: September 22, 2026—the models' parameter contracts and live credit quotes in LUVI, LUVI's own prompting guide for the PixVerse family, and PixVerse's platform documentation for text-to-video.
• Results: 350, 600 and 1,150 credits for five seconds at 360p, 720p and 1080p on v6; 650 at 720p on c1. Neither model exposed a negative prompt field. No outputs were generated for this post, so it makes no claim about output quality.
• Drawbacks: the four words named above come from LUVI's prompting guide rather than a published ablation, and the 50-to-80-word target is a working range, not a measured optimum.

In LUVI

Open v6 or c1 in the Workspace, write the three sentences in order, and let the feather button check them against LUVI's prompting guide for the model before you generate—it catches the broad words and the negations. Then set quality and duration deliberately, because those two decide what the clip costs; the help center explains how resolution and duration change the price.

New accounts start with 200 credits, and verifying a phone and a card adds up to 1,500 more. You can create a LUVI account and test the three-sentence shape on your own subject.

Sources

• Text-to-video guide — PixVerse Platform Docs, official documentation. Accessed September 22, 2026.