Pixverse Transition vs Flux 3 first-last for a pack shot

Flux 3 first-last interpolates two stills you locked, 5–20s at 9 credits with native audio. Pixverse Transition is 5 / 8 / 10s at 1080p, 40 credits, audio unset — a clip-to-clip morph, not two catalog stills. A pack-shot A/B is first-last; a trailer smash is Transition.

The names both say "transition." The jobs do not. One row is image-to-video between two plates you already approved. The other is an edit-video morph between two clips you already cut. Routing a pack A/B through Pixverse is how a locked bottle becomes a morph. Routing a trailer smash through Flux 3 is how two finished shots get treated like hero stills they are not.

Two interpolation jobs, two catalogs

Flux 3 First Last Frame to Video is Black Forest Labs' interpolation row: a defined start frame and end frame, native audio, up to 1080p, durations from 5 through 20 seconds, listed at 9 credits. Category is image-to-video. The stills are the contract. The prompt describes the journey — how light, palette, and the pack travel from A to B — not a new scene.

Pixverse Transition is Pixverse's clip-to-clip morph: 5, 8, or 10 seconds, 1080p, 40 credits, audio unset. Category is edit-video. The catalog line is "create smooth transitions between video clips." That is a smash between two pieces of footage, not two product plates from the stills roster. Do not brief native audio on this row. The audio field is null.

The first-last-frame glossary is the Flux job: two stills, motion that has to land. Pixverse sits next to it in conversation and not in the catalog. Confirm the row before you attach files.

Pack-shot A/B is first-last

A pack-shot A/B is two locked stills of the same object in two states: closed carton and open carton, front label and back label, bottle upright and bottle poured. You already signed both plates. You want one clip that travels from one to the other without inventing a third bottle.

That is Flux 3. Attach the two stills. Write the in-between only: a slow rotate, a lid lift, a pour that lands on the open pack. Durations are every second from 5 to 20 — pick the length the move actually needs. Native audio is on, so write the mix (a quiet table, a cap click) or write silence. A silent brief still gets a soundtrack you did not choose.

The agent job for a first-and-last-frame transition is this path: two photos in, one interpolated clip out. It is priced per model, shown before you confirm. It is not a Pixverse smash. It is not a text-to-video pack that "kind of looks like" the stills.

If the two plates are not the same object in the same light, Flux 3 will still connect them. The connection will be a warp. A/B only works when A and B are plausibly the same shot at two times.

Trailer smash is Transition

A trailer smash is two clips you already have: the end of the chase, the start of the title card, a hard cut you want to feel like a morph. You are not introducing a pack. You are joining two pieces of motion with a graphic in-between.

That is Pixverse Transition. 5, 8, or 10 seconds — no 6, no 12, no 20. 1080p at the listed 40 credits. Audio unset: do not expect a designed whoosh from the row, and do not claim native dialogue. If the smash needs a hit, add it in the edit.

The inputs are the two clips (or the frames you pull from their join points), not two catalog stills of a pack. Re-describing the product in the prompt does not turn Transition into first-last. Style enums on the row (anime, 3d_animation, clay, comic, cyberpunk) are look, not easing. There is no match-cut toggle. Name the morph in prose.

Use this when the trailer already exists as shots and the join is the problem. Do not use it to invent the pack-shot from two moodboards.

Pick the row from the files you have

Ask what you are holding:

• Two approved stills of the same pack, two states, one continuous move: Flux 3, 5–20s, 9 credits, native audio. Run it from image-to-video or from the agent first-last job.
• Two finished clips and a join that should morph: Pixverse Transition, 5 / 8 / 10s, 1080p, 40 credits, audio unset.

If you have one still, you do not have either job. That is ordinary image-to-video. If you have a paragraph and no files, you do not have a pack-shot A/B. Generate or shoot the plates first.

Do not mix the durations. Flux will not sell 8 seconds as a special Pixverse length — 8 happens to exist on both menus, and that coincidence is how people pick the wrong row. Check the category: image-to-video versus edit-video. Check audio: true versus unset. Check the files: stills versus clips.

A pack you can ship is two plates plus Flux. A trailer you can punch is two clips plus Transition. Same word. Different generate.

FAQ

Does Pixverse Transition have native audio?

No. The catalog leaves audio unset. Do not brief a talking pack or a designed whoosh as if the row will author it. Add sound in the edit if the smash needs a hit.

Can Flux 3 first-last take two clips instead of two stills?

No. It interpolates a start frame and an end frame. Category is image-to-video. Two clips is the Pixverse job.

Why is Flux listed at 9 credits and Pixverse at 40?

Those are the published catalog numbers: Flux 3 first-last at 9 credits for the 5–20s interpolation with native audio; Pixverse Transition at 40 credits for 1080p on a 5 / 8 / 10s morph. They are not interchangeable rates for the same file.

Which one does the agent first-last job call?

The first-and-last-frame agent job wants two photos and an interpolation model. That is the Flux-shaped path, not a clip-to-clip smash. Attach stills. If you are holding clips, pick Pixverse Transition yourself.