Single Scene AI Clip

Plan One Shot, Or Animate One Image Directly

Build a single scene end-to-end with reference images, or skip straight to direct image animation. Saved images stay in your library so you can reuse them without uploading again.

Workflow review

Reference uploads and generated outputs are now designed around stable `/uploads/...` media paths so the app can serve them locally or from S3-backed object storage without changing the UI.

Generation Mode

Choose whether we should create a fresh frame first or animate an existing image directly.

1. Reference Stack

Upload product, person, and support references. Place the strongest anchor first.

Upload your first anchor image

Start with the product or face that must stay the most consistent. Add more references for environment, styling, and extra details afterward.

Saved Library

2. Direct the Frame

Generate a visible OpenAI frame preview, regenerate until it is right, then continue to animation.

No references selected. We will generate the frame from prompt only.

Frame prompt

Tip: you can refer to Step 1 reference titles directly (example: "Use Ref 1 - Inverter hero as the main product").

Generate a frame preview first to review composition before animation.

3. Direct the Motion

This becomes the animation brief for xAI. Keep it focused on one continuous shot with synchronized speech and music.

Motion prompt

Optional spoken line

16/10 words

Leave this blank if the clip should be silent or if no spoken line is needed.

Speech language

Background music direction

Audio mix

xAI video model

Use this to test alternatives quickly. Keep `grok-imagine-video` if you are unsure.

Aspect ratio

Frame quality

Video duration

Video resolution

Prompt Plan Preview

This is the exact plan that will be sent when you click generate.

Credit required: 15 per one-shot generation

No frame preview generated yet. Generate Clip will ask OpenAI to create a fresh frame first.

Reference summary

No references yet.

OpenAI frame prompt

Create one polished keyframe for a single-scene AI clip.

Preserve the recognisable identity, product geometry, surface details, logos, and visual cues from the uploaded references.

Reference priority: treat reference 1 as the strongest anchor, then blend the remaining references only where relevant.

If the scene direction mentions a reference title, match it to the exact title in the reference brief.

Scene direction: Create a cute 3D cartoon character based on this product, smiling with simple arms and legs, standing in a sunny rice field near a strong PVC water flow, with solar panels in the background and the Eman controller mounted under the panel. Vertical 9:16 composition for TikTok.

Vertical 9:16 composition sized for TikTok, Reels, and Shorts.

Image quality target: high.

Keep the composition coherent as one real shot, not a collage.

Do not add extra featured products, duplicate people, swap logos, or change facial identity unless the prompt explicitly asks for it.

No subtitles, speech bubbles, UI chrome, watermarks, or text overlays unless explicitly requested.

xAI motion prompt

Animate the finished keyframe as one continuous shot.

Preserve the same person identity, product design, styling, props, and background layout from the keyframe throughout the clip.

Motion direction: A Thai farmer stands in a bright rice field, turns toward camera, points to the solar panel and PVC pump setup, then speaks confidently to camera in one continuous shot.

Audio direction: generate synchronized spoken dialogue and background music in the same clip.
Dialogue language: English. Pronunciation should sound native and natural.
Required spoken line (keep wording exact, no paraphrase): "Hot days should power your farm. Use solar DC pumps with Eman and enjoy warranty support.".
Background music style: Warm uplifting commercial underscore with modern acoustic instruments, no lyrics.
Avoid distorted voices, gibberish phonemes, clipped syllables, and abrupt audio pops.
Audio mix: keep speech clearly in front and duck background music under dialogue.

Clip target: 5 seconds, 9:16 vertical, 720p.

Avoid jump cuts, morphing, duplicate limbs, face drift, product deformation, floating objects, and unnecessary camera shakes.

Latest Output

Track the current run with a visible timeline, then inspect the frame and final clip.

Start a generation to see the live timeline, frame preview, and final output here.

Generation History

Review previous runs, reopen outputs, and inspect failures for debugging.

0 local

Loading generation history…