Fawna Muse

Muse is an image-to-video specialist. It requires an input image, supports first and last frame, runs up to 15 seconds, and has the most permissive content policy of any Fawna tier.

Tier label

Muse

Price

From 22 credits per second

Aspect ratios

16:9, 9:16, 1:1

Resolutions

720p, 1080p

Durations

2, 4, 6, 8, 10, 15 seconds

Audio

Toggle

Quality tiers

Fast, Standard

Character refs

9-grid composite (one image, up to 9 panels)

Style ref

Not supported

Keyframes

First frame required, last frame optional

Negative prompt

Supported

Magic Prompt

Supported

When to pick Muse

• You already have an image and want to animate it.
• Your content needs permissive handling: Muse has the most relaxed content policy in the lineup.
• You want the longest possible clip (15s).
• You have a character sheet you can compose into a 9-grid for best-in-class consistency.

The input image is required

Muse is I2V only. There is no text-to-video mode. If you select Muse and send a prompt without an image, the composer shows a warning and the generation will not start. Always drop a first-frame image first.

For best results, the image should be clean, well-composed, and framed in the aspect ratio you want. The model preserves composition tightly.

The 9-grid reference

Muse's character consistency is driven by a single composite reference image with up to 9 panels arranged in a 3x3 grid. Each panel shows the same subject from a different angle or pose. This gives the model dense information about who the character is, without needing 9 separate ref uploads.

You can build a 9-grid manually in Compose, or use the Studio Library's "Build 9-grid" action on a character you have already generated. Attach the grid to the first-frame slot and Muse treats it as identity anchor plus scene input.

Strengths

• Longest clips in the lineup (15s).
• Most permissive content handling.
• Dense 9-grid character lock is excellent for consistency-critical shots.
• Smooth frame interpolation when both first and last frames are set.

Where it struggles

• No text-to-video mode. You must supply an image.
• No style ref slot. Style has to come from the first-frame image itself.
• Audio is synthesized, not natively synced like Cinema or the Film family.

Prompt recipe

[What happens in the clip: subject action, change, and motion].
[Single camera move]. [Any new lighting or mood cues].

Muse is already looking at the first frame, so you do not need to redescribe the scene in detail. Focus on what changes during the clip.

Example

Maya stirs her tea slowly, then lifts the cup to her lips and
smiles at something off-camera. Soft handheld drift forward.
Natural golden-hour warmth deepens as a cloud passes.

Tips

• If your clip feels still, bump the duration. Muse uses the extra frames well and avoids the "loop and stop" artifact that other models can produce at long durations.
• Describe change, not the scene. The scene is in the image.
• Use Magic Prompt for quick briefs. It will expand a short action into a full prompt.
• Build your 9-grid once. Reuse it in every character scene. Consistency compounds.

Where to go next

• Using references for details on the 9-grid input.
• Fawna Compose to build the first frame and character grid.