Free AI Music Video Generator with Model Choice: Concept → Finished Music Video

Movie Maker at egaku-ai.com/movie-maker takes a single concept sentence ("An atmospheric music video: a lone figure walking through rain-soaked city streets at night") and produces a finished cut with scenes, video, and music. One concept in, one MP4 out.

The pipeline: concept → scene breakdown (LLM) → still generation per scene (Flux or SDXL) → i2v per still (Wan 2.6 or Seedance 2) → BGM generation (ACE-Step) → ffmpeg stitching. Every step is exposed if you want to control it. Every step also has a sensible default if you don't.

Two use cases keep pulling users into this flow:

• Indie musicians who can't afford a shoot. A rough concept turns into a mood video that they can post on YouTube or Instagram Reels while the real budget arrives.
• Content creators who need vertical loop content. The output is 9:16 by default. Great for TikTok, Reels, Shorts.

What Movie Maker does NOT do: it's not a lyrics-synchronized MV generator. The BGM is instrumental, driven by the concept description. If you need vocals synced to lip movement, look at Talking Avatar instead.

The music engine is ACE-Step, an open-source text-to-music model that runs on the same fal.ai infrastructure as the visual models. You supply a mood prompt ("cinematic epic, dark, orchestral") and a duration (up to 60 seconds), and it returns a 44.1 kHz WAV that gets muxed under the video.

Compared to Suno or Udio, ACE-Step is less lyric-focused and more mood-focused. That fits music video use — you want ambient tension, not verse-chorus-bridge.

ACE-Step is a paid feature (currently 15 credits per 60 seconds) but the visual side of Movie Maker works on the free tier. You can generate the video, download it, and add your own music separately.

Movie Maker exposes both text-to-video and image-to-video model choice. In our internal test set:

• Wan 2.6 i2v — best for realistic / cinematic music video. Identity preservation is strong across cuts.
• Seedance 2 Pro — best for high-motion, dance-adjacent visuals. 1080p output.
• PixVerse C1 — best for stylized / anime music video.
• LTX Video 2.3 — fastest, cheapest, but lower quality. Good for iteration.

Switching models between scenes is supported — if the intro should be cinematic and the drop should be high-motion, use Wan 2.6 on scene 1 and Seedance 2 on scene 2.

A three-scene vertical music video with BGM costs roughly:

• 3 × still generation (Flux Dev): 9 credits
• 3 × i2v (Wan 2.6, 5s each): 60 credits
• 1 × BGM (ACE-Step, 30s): 8 credits
• Total: ~77 credits ≈ ¥77

Compared to Suno + Runway ($30/month + $15/month subscription), one Movie Maker MV is roughly ¥100. If you make one a month EGAKU wins; if you make ten a day you probably want the subscription. Or you can generate the video on EGAKU and the music on Suno, mux them yourself.

egaku-ai.com/movie-maker — concept in, MP4 out. Five free generations anonymous, +50 credits on signup.