VEED Fabric on a 41.5-second voice note: $4.20 at 480p, $7.88 at 720p

A 41.5-second voice note turned into a talking still with Sume's VEED Fabric 1.0 is billed as 42 seconds. That is 42 x $0.10 = $4.20 at 480p and 42 x $0.1875 = $7.875 at 720p, which the reservation rounds up to $7.88.

The arithmetic

The catalog row for veed/fabric-1.0 prices ceil(audio seconds) after Sume's margin: $0.10 per second at 480p and $0.1875 per second at 720p, with a maximum of 300 seconds. The reservation is stored in USD micros and rounded up to cents for the balance check, so the 720p figure appears as 788 cents.

For comparison, the same note at 480p costs 42 x $0.10 = $4.20, which is $3.675 less than 720p. The gap grows with length: a 100-second note bills $10.00 at 480p and $18.75 at 720p, so the resolution choice is where long clips are won or lost.

Where the 10 MB limit bites

The same row says the audio must be Sume-hosted and under 10 MB. A 41.5-second note is far below that in compressed formats, but an uncompressed WAV can approach it. Check the file size as well as the length before you upload.

The body takes one visual source: an image_url of a still you accepted, or an avatar_handle if the user named that avatar, never both. Send the measured duration_seconds along with the audio_url, as the models overview describes.

• Audio: Sume-hosted, under 10 MB, up to 300 s.
• Visual: one of image_url or avatar_handle.
• Billing: ceil(duration_seconds) x resolution rate.

Draft at 480p, finish at 720p?

Running the same note at 480p first costs $4.20. Rerunning at 720p adds $7.875, for $12.075 in total. If you are confident in the still and the audio, go straight to 720p and save the $4.20. Draft at 480p when the still is new, because a bad pose is not fixed by a higher resolution.

For a note this long, consider whether one face for 42 seconds is what you want. A single still talks for the entire length; if you need cutaways, split the note and join the pieces in a timeline.

A practical rule for long notes is to render the face at the resolution you will actually show. A 480p talking head inside a corner overlay is indistinguishable at small size from 720p, and it costs 53 percent of the price ($0.10 / $0.1875). A full-screen vertical clip is the case for 720p.

If the voice note is much longer than the face needs to be on screen, consider cutting the note into two Fabric jobs and putting B-roll between them. Two 21-second halves at 480p cost 2 x 21 x $0.10 = $4.20, the same as one 42-second job, because Fabric is linear in audio seconds. The split costs nothing extra in Fabric charges; it only adds the timeline render for the join, at $0.10 per output minute, rounded up, so $0.10 for a clip under a minute.