Attachments aren’t just context — each upload type activates a specific behavior in the pipeline.
Images
A person’s photo becomes an identity reference: the person appears in the video with their real face, consistent across every scene where they’re cast.
A product photo becomes the canonical product reference. The real item — shape, colors, label text — is preserved in every scene that features it. Wavemaker even transcribes the brand text on the product so wordmarks render letter-perfect.
A style image (a mood frame, a film still, a texture) steers the look. Wavemaker classifies uploaded images as subject vs. style automatically and uses each accordingly.
Practical notes:
- Up to 10 MB per image; JPEG/PNG/WebP/HEIC supported (HEIC is converted automatically).
- Clean, well-lit, single-subject photos anchor identity best. For products, a straight-on shot without background clutter works better than a lifestyle collage.
Audio
A music track becomes the video’s score. Wavemaker listens to the song — detecting real BPM, song structure (verse/chorus), energy, and lyrics — then cuts scenes on the beat, maps movements to sections, and syncs imagery to lyric moments. If you give no duration, the song’s length drives the video, and the track plays once with a clean fade instead of looping.
- Up to 50 MB; MP3/WAV/AAC/OGG/FLAC/M4A supported.
Video
A reference video (an ad you like) can be analyzed as a creative reference: cut rhythm, structure, shot grammar, and style inform your video’s plan. You can paste a URL instead of uploading — YouTube, TikTok, Instagram, X, Vimeo, and direct file links are supported.
Important legal guarantee: reference analysis is structural only. Wavemaker never copies the reference’s footage, people, wording, or logos into your video — it learns pacing and style, not pixels.
Your own footage can be used directly in the composition, cleaned up (denoise/stabilize), extended, or edited.
- Up to 100 MB per file; MP4/MOV/WebM supported.
Documents
PDF, DOCX, XLSX, and CSV files are parsed and read. Paste a product spec sheet, a report, or a press release and the video is grounded in its actual content — great for “turn this report into a 60-second explainer.”
- Up to 50 MB per document.
How to reference uploads in your prompt
You usually don’t need to — attached assets are automatically summarized for the planner and preferred over generated stand-ins. But being explicit removes all doubt:
“Use the attached photo as the founder — she narrates on camera in scenes 1 and 5.” “The attached song is the soundtrack; cut the whole video to it.” “The attached CSV is our product list — feature the top 3 rows.”
Limits at a glance
| Type | Max size | Count |
|---|---|---|
| Images | 10 MB each | up to 5 per message |
| Video | 100 MB each | up to 5 per message |
| Audio | 50 MB each | up to 5 per message |
| Documents | 50 MB each | up to 5 per message |