Design (image, video and music)

Design is the space in the top capsule for creating content with AI: you write what you want and the result appears in the thread, with its file saved on your computer. It is meant for anyone to use without knowing what a seed or a scheduler is: the technical settings are folded away and only show up if you ask for them.

The three tabs

Above the text box you choose what to create:

• Image: pick the provider and model, the size (aspect ratio) and how many images you want.
• Video: same as image, plus the clip length. Each model accepts only certain lengths; if you ask for one it doesn’t accept, the app tells you which ones work.
• Music: describe the style of the song and choose its length. Details below.

With the box empty, examples appear below it (“Try:”). Clicking one only fills in the text: nothing is generated until you send it, so you can tweak it first.

Creating music

1. Click the Music tab.
2. In the main box describe the style: genre, mood, instruments. For example, “upbeat pop with acoustic guitar and claps, summer”.
3. Choose the length: 30 seconds, 1, 2 or 3 minutes.
4. If you want vocals, click Lyrics and write what will be sung. You can mark the parts with [verse] and [chorus]. If you leave the lyrics empty, or click Instrumental, the song comes out without vocals.
5. Send. While it is being made you’ll see the “Creating song…” card; it usually takes a couple of minutes and you can cancel it.

The song appears in the thread with a player and buttons to copy the path, open the file or delete it. It is saved as audio (MP3 or WAV, depending on what the service returns) in the Design chat folder, marked in the file itself as AI-generated content.

Requirement

Music is created in the cloud with the ACE-Step model through fal.ai, which charges per use with your own key. Connect it first in Settings → Providers → fal.ai; if it isn’t connected, the app tells you when you send. Music generated on your own computer is not available yet.

Audio with ElevenLabs

In the Music tab, below the lyrics, there is the Generate with ElevenLabs panel. It needs your ElevenLabs key (Settings → Voice) and consumes credits from your account. Choose what to create:

• Sound effect: describe the sound (“old wooden door creaking open”), with automatic or fixed duration and a Loop option.
• Music: style and duration, with an Instrumental option.
• Voice design: describe the voice in at least 20 characters (“warm female voice, around thirty, slow pace”) and name it; it is saved to your account.
• Voice changer: pick an audio file and the target voice from your account.
• Isolate voice: pick an audio file and the noise and background music are removed.

The result goes into the Design chat thread like any other song, with a player and buttons to open or copy the file.

Advanced options

The Advanced options button unfolds what most people don’t need:

• Presets: save a named combination of settings and bring it back later.
• Seed: fix the starting point to repeat a similar result.
• Negative prompt: what you don’t want to appear (only on models that support it).
• Model that writes the prompt: which model “Enhance prompt” uses.

The app remembers whether you left the panel open or closed. Music has no advanced options.

Enhance prompt and tools

• Enhance prompt rewrites your description as a detailed prompt before generating. It works for Image and Video; for Music your text is sent as is.
• Tools groups operations on an image from the chat, such as upscaling it, turning it into a video or removing its background. Some run on your computer with local models from Context Forge and others depend on the provider you have configured.