VEED Fabric 1.0: the mouth pass after the picture exists (720p, 40cr)
VEED Fabric 1.0 wants a plate and a track. It is not how you invent the scene.
VEED Fabric 1.0 wants a plate and a track. It is not how you invent the scene.
VEED Fabric 1.0 is content type lipsync. Categories: audio-to-lipsync and image-to-lipsync. requires_image is true. Audio is true. Max output: 720p. Credits: 40. The catalog description also says "Professional fabric and texture video generation" — read that next to the categories, not instead of them. This row is a mouth pass. It needs a face (the plate) and a voice (the track). It will not design the set, the product, or the story.
Plate plus track
Image-to-lipsync is the still. Audio-to-lipsync is the voice. Together they are a talking head. Missing either one is the wrong label: a prompt-only "make a presenter" request belongs on a generate row, not here. requires_image: true is the gate on the spec. Bring the portrait. Bring the audio or the script you will turn into audio. Then run Fabric.
Forty credits is the catalog sticker. Qualities listed: 480p and 720p. Ceiling: 720p. Frame rates: 24fps, 30fps. Motion levels: Low, Medium, High. Styles listed: Cinematic, Documentary, Abstract. Those knobs dress a talking plate. They do not invent the plate.
The door is AI lipsync. The ranked lists are best lipsync model and best talking-head model. VEED's avatar and lipsync siblings sit on the VEED roster. Prompting: VEED Fabric 1.0 prompting. The edit task is lipsync a video.
720p lipsync, 40 credits
No duration list on the spec. Length follows the track. A 15-second VO is a 15-second mouth pass; a 3-second line is a 3-second mouth pass. That is why inventing the scene inside Fabric is so expensive: you will pay 40 credits (and whatever the per-second quote adds) to lip-sync a picture you do not like, to words you have not finished.
audio is true because the job is mouths matching sound. Write the line before you generate. A placeholder "hello welcome to my channel" is how you spend the 40 credits twice. The glossary term is lipsync.
720p is the ceiling. If the brief is a 4K keynote, this is the wrong row. If the brief is a 9:16 talking still at 720p, match the still's ratio to the platform and stay.
Mouth pass after the picture exists
Order:
- Lock the face still (a photoreal or illustration stills row — not Fabric).
- Lock the voice (TTS, clone, or an uploaded take).
- Run VEED Fabric 1.0 so the mouth follows the track.
- Caption, crop, publish.
Skipping 1 and 2 is "invent the scene." Fabric is step 3. Professional fabric/texture language in the description does not turn step 3 into a set dresser. If you needed a room, generate the room. If you needed a product turn, generate the turn. If you needed the person to say the line, this is the model.
Cheapest lipsync model is a different ranking. Fabric is 40 credits at the catalog field. Use it because you have a plate and a track, not because a roundup said VEED.
It is not how you invent the scene
A scene has blocking, lighting, a product, a gag. A mouth pass has phonemes. Keep them in different rows. When the picture exists and the track exists, spend the 40 credits once.
FAQ
Can VEED Fabric 1.0 generate a talking scene from a text prompt only?
Not as specced. requires_image is true. Categories are audio-to-lipsync and image-to-lipsync. Bring a plate and a track. Invent the scene on generate rows first.
What resolution do I get?
Max output 720p, qualities 480p and 720p. There is no 1080p or 4K on this slug.
Why 40 credits with no duration list?
40 is the catalog credits value. Length follows the audio. Check the in-app quote for the job. Do not convert it to USD. Do finish the script before you pay for the mouth pass.
Is this the same as VEED's avatar rows?
No. This slug is Fabric 1.0, lipsync, 720p, 40 credits, image required. Avatars are other slugs on the VEED roster. Pick the row that matches the job: a locked portrait plus a track, not a stock presenter you have not designed.