Ace-Step 1.5 - Cover Art and Model Switch

This workflow will primarily use Ace-Step 1.5 Models to generate Music and an Cover Art image using Flux2.Klein 9B or Krea2 in one run.

This workflow tries to be a comfortable one click generation WF with organized and well named output.

Meta Lyrics Version:

This one is a comprehensive all in one workflow which has the following features:

• Switch between ACE-Step models (Checkpoint / diffusion / unet)

• Switch between t2i models (Flux.2 klein 9B / Krea2)

• Save mp3 including metadata (cover-art inside the mp3 output)

• Save mp4 Lyrics video (warning: large file size and 2 minutes generation time)

warning - dependencies required and a particular custom node mandatory for this one:

GitHub - mattjohnpowell/comfyui-audio-expo · GitHub

FLUX.2 9B and Krea2 Version:

As a result the following will be saved by default:

Root Output folder → Folder named after your Music Title →

Filename will be: Artist - Title - Seed - Output type

Example:

Adaptalabor - Homing Flames - 619329538607327 - Audio_00001.mp3

Adaptalabor - Homing Flames - 619329538607327 - Cover_Art_0001.webp

Adaptalabor - Homing Flames - 619329538607327 - Static_Video_00001_.mp4

The Title will be shown in the Cover and influences the cover art design.

You can add an additional prompt to enhance the cover-art outcome.

Required Reccources:

https://civitai.red/models/2425543/ace-step-15-music-generation-for-low-vram?modelVersionId=2727094

Any CustomNodes that are used in this workflow.

Just make sure you have the Custom Nodes manager installed and

use the missing nodes download function.

Have Fun and have a nice day.

The Process:

It might be helpful that I posted a table (created with Copilot) that shows typical key-range/bpm/ and scales for each genre.

Furthermore I use a local llm on a seperate text-generation client to create the lyrics. My favorite Model right now is: MS3.2-The-Omega-Directive-24B-Unslop-v2.0 (16GB VRAM-Card)

A promt can looks like this:

"create long lyrics for a lo-fi style music song from a female's perspective. make it rhyme. make it erotic and poetic. Here are some outlines:

Verse1

[Description of the topic / theme / story]

Chorus:

[Description of the topic / theme / story]

Verse2

[Description of the topic / theme / story]

Chorus:

[Description of the topic / theme / story]

Verse3

[Description of the topic / theme / story]

Chorus:

[Description of the topic / theme / story]"

This usually works well."

I could not keep the whole thing to myself and hope it is as helpful to you as it was for me.

Enjoy!