Image models

How Autotube uses Muse Image.

Meta's Muse Image is what a scene is generated with unless you pick something else: 2K frames, consistent characters, and ten reference images per scene.

3 credits per image

What it is good at

Why Muse Image is in the catalog.

  • The default scene model

    Every scene uses it unless you choose otherwise, so a run needs no model decision at all. Each step still shows its exact credit cost before it generates anything.

  • 2K frames

    It returns a 2K output class, which is what a 1080p scene frame wants — no upscaling between the model and the timeline.

  • Ten references a scene

    Up to ten reference images per scene, and cast and style references can go in the same request, so an ensemble shot and the channel's look hold together.

  • Every scene in one run

    It batches across a whole video, so a long shot list is generated in one run rather than a scene at a time.

Specs, from the app’s own catalog

Output size
2K images, 1080p scene frames
Consistent characters
Yes, from reference images
Reference images
Up to 10 per scene
Style reference
Yes, alongside character references
Match an input image
No, a preset ratio only
Prompt length
No fixed cap
Whole-video batches
Yes, every scene in one run

How this pipeline step works

What it costs

Priced in credits, before it runs.

3 credits per image

Muse Image sits in the value tier, which is what makes it a sensible default for a shot list that runs to a hundred scenes. The estimator prices the whole run before it generates anything, so the figure above is the one you can multiply by the scene count.

What a credit is, and what a video costs

Side by side

Muse Image against the other image models.

 Muse ImageNano Banana 2.1Grok ImagineGPT Image 2GPT Image 2.5 Flare
Credits3 credits per image11 credits per image6 credits per imagePriced from the request4 credits per image
Output size2K images, 1080p scene frames1K images, 720p scene frames2K images, 1080p scene frames1K images, 720p scene frames2K images, 1080p scene frames
Consistent charactersYes, from reference imagesYes, from reference imagesYes, from reference imagesYes, from reference imagesYes, from reference images
Reference imagesUp to 10 per sceneUp to 14 per sceneUp to 5 per sceneUp to 10 per sceneUp to 10 per scene
Style referenceYes, alongside character referencesYes, alongside character referencesYes, alongside character referencesYes, alongside character referencesYes, alongside character references
Match an input imageNo, a preset ratio onlyNo, a preset ratio onlyYes, instead of a preset ratioNo, a preset ratio onlyNo, a preset ratio only
Prompt lengthNo fixed capNo fixed capNo fixed capNo fixed capNo fixed cap
Whole-video batchesYes, every scene in one runYes, every scene in one runYes, every scene in one runNo, one image at a timeYes, every scene in one run
Limits

What Muse Image will not do.

  • It cannot match an input image's aspect ratio — it composes to a preset. Where a scene is built around a still you supplied, Grok Imagine keeps that still's shape.
  • It is a still-image model. For motion, the clip models animate a scene once the frame exists.
FAQ

Muse Image: common questions

Every model

The rest of the catalog.

All models, grouped by pipeline step

Try Muse Image on your next video.

Every plan reaches every model in the catalog, and the app shows the exact credit cost before it runs a thing.

See pricing