SAM 3 segment before background remove

SAM 3 Video Segment is 1 credit to cut a matte. Bria video background removal is 1 credit. VEED video background removal is 3. Segment first when you need the key. Remove when you need the back gone. Do not pay 3 credits to discover you needed a mask.

Sam 3 Video Segment is content type video, category edit-video, 1 credit, audio null, no duration chips, no resolution listed, image not required. Description: intelligent video segmentation for object detection and precise video editing. The output is a cut — an object, a person, a region — you can use. It is not a generate. It is not, by itself, a green screen.

Bria Video Background Removal (display name Bria; catalog slug brio-video-background-removal) is edit-video, image required, 1 credit, audio null. Description: AI-powered video background removal and replacement.

VEED Video Background Removal is edit-video, image required, 3 credits, audio null. Description: isolates the subject from moving footage frame by frame, no green screen required; clean cutout ready for compositing.

Three edit rows. One family of “do something to picture we already have.” Three jobs people mash into one button.

Segment versus key

A segment is a matte. SAM 3 is the cheap matte. You need to know which object, over time, so an editor can grade it, hide it, track it, or composite it. The background may stay. The point is the selection.

A key is the back gone (or replaced). Bria and VEED are keys. The subject is the keep. The rest is empty, replaced, or chroma, depending on the card.

Run segment first when:

• You will composite later and you want control of the edge in an editor, not a baked cutout.
• You need one object out of a busy frame (the product, not the person; the person, not the dog).
• You are not sure the remover will pick the right subject. A 1-credit matte is the cheap way to see.
• The next step is a grade, a blur, a highlight — not a new cyclorama.

Run a remover when:

• The job is “subject on a new back” and you do not need an intermediate matte.
• You will composite on a plate you already have and a baked cutout is enough.
• You needed chroma specifically — that is VEED’s green-screen sibling, not SAM 3.

The remove video background task is the key door. AI video editor is the suite. Background removal is the term. Segmentation is the step people skip because the remover is the word they know.

Skipping the matte looks like this: VEED at 3 credits, the model keys the wrong person in a two-shot, you run it again. SAM 3 at 1 credit would have shown you the object. Then you key.

One credit, one credit, three credits

SAM 3 and Bria are both 1 credit. They are not the same 1 credit.

SAM 3: no image required on the published card, because the input is video to segment. Empty duration list — you are not picking 5s of generate. Empty resolution — output follows the plate. Audio null — this will not write a bed. Features live in the description: object detection, precise editing. That is a matte sentence.

Bria: requires an image (the plate). Edit-video. 1 credit. Removal and replacement in the description. If the label is wrong — you typed a scene instead of attaching the file — every second is waste. The existing Bria trap is treating it as I2V generate. It is not. It cuts or replaces what is behind the plate.

VEED remove: requires an image, 3 credits, baked cutout, no green screen required on this slug. Three credits is the VEED key, not a SAM discount. Pay it when you want VEED’s cutout and you do not need a matte first.

VEED’s other card, VEED Video Background Removal Green Screen, is also 3 credits, also edit-video, also requires a plate. It replaces the background with solid chroma green so you can key downstream. That is still a key, not a segment. SAM 3 does not emit a guaranteed green field just because you asked in the prompt.

Do not “upgrade in the prompt” from SAM to VEED. Change the row. A 1-credit matte will not become a 3-credit VEED cutout because you wrote “green screen” in the box.

Audio is null on all three. Keep the source mix in the editor. Native audio is a generate feature elsewhere.

The plate is the contract on the removers

Bria and both VEED removers require a plate. SAM 3 does not list that gate the same way. Practical result: you still need footage. Nobody segments a paragraph.

Bring a clip whose subject is actually separable. Motion blur, hair against a matching wall, a translucent product — those fail every key. A 1-credit SAM pass is the cheap diagnostic: if the matte falls apart, a 3-credit VEED pass will not save it. Relight or reshoot or generate a cleaner plate, then key.

Replacement in Bria’s description is not “put them in Tokyo” from a sentence. It is “put this file on a new backplate.” Bring both sides of the composite if the job is a swap. Generate rows invent Tokyo. Edit rows cut.

After a key or a matte, you still ship. Captions on a cutout need a back, or they float on checkerboard. Add captions to video. Safe area still applies when you composite onto 9:16. Captions for silent autoplay still apply when the source had speech; the remover did not invent a VO.

Photoreal cutouts of real people in ads can still be synthetic performances when the back is fake and the audience would take the scene as a real location. EU Article 50 from 2 August 2026. Article 50 disclosure. A 1-credit matte does not exempt the composite.

The wider green-screen versus model split is background removal vs green screen. This page is narrower: segment first, then pick the 1-credit or 3-credit key.

Confirm the job on the chip

Open the catalog. If the description says segmentation and object detection, you are on SAM 3 at 1 credit. If it says background removal and replacement, you are on Bria at 1. If it says cutout ready for compositing, you are on VEED at 3. If it says chroma green, you are on the VEED green-screen sibling at 3.

If the confirm screen is a text-to-video generate, you drifted. Back out. These four credits of edit (1+1+3 as a mistaken stack) are still cheaper than a 49-credit silent generate of a cyclorama you will then try to key.

Order of operations for a composite you actually control:

1. Lock picture.
2. SAM 3 at 1 credit if you need to see the object.
3. Bria at 1 or VEED at 3 if you need the back gone.
4. Composite.
5. Captions.

Skip 2 when the subject is obvious and a baked cutout is enough. Do not skip 2 when the frame has two people and you only wanted one.

FAQ

Is SAM 3 a background remover?

No. It is a 1-credit edit-video segmenter: object detection and a matte for precise editing. The back may stay. Bria (1 credit) and VEED (3 credits) are the keys.

Why is Bria’s slug brio-video-background-removal?

That is the published catalog slug. The display name is Bria. Link the slug. Do not hunt a second URL.

Should I always run SAM 3 before VEED?

No. Run it when you need the matte or you are unsure which subject a remover will pick. If the job is a clean single subject on a new back and a baked cutout is enough, Bria at 1 or VEED at 3 is the row.