Cold-brew concentrate: first-last bottle to diluted glass

Own both stills. Ratio copy is overlay, not VO.

Cold-brew concentrate is a two-state product. Frame A is the bottle you actually fill. Frame B is the glass at the dilution you print on the label. The ad is the travel between those two photographs. A model that invents a pour from a paragraph will invent a second bottle, a second ratio, and a voice you did not approve. Photograph both ends. Interpolate. Burn the ratio as type.

Own the bottle and the glass

Shoot the closed bottle where it lives in the listing: label large, cap readable, no lifestyle kitchen the art director did not light. That is frame A. Mix the dilution you sell — the printed ratio, the same glassware every time — and shoot it in the same lighting family, same camera height, same back edge. That is frame B.

The two frames have to be connectable. Same counter, same lens family, same colour of bottle glass. A hero pack shot morphing into a cafe window is a warp, not a dilute. Keep the set.

First-last frame turns an open pour prompt into a bounded landing. Ordinary image-to-video knows where it starts and guesses the rest. First-last has to finish on the glass you attached. If you never attach the glass, there is no ratio. There is only a bottle wiggling.

Do not generate the "after." A model-made tumbler is a different drink. The label already committed to a ratio. The photograph of that ratio is the proof.

Interpolate; do not prompt a pour

Create a first-and-last-frame transition video is the named agent job: two photos in, one clip that opens on the first and lands on the second. The agent passes both stills and picks a first-last-capable model. Cost is priced per model, shown before you confirm. This is not "turn one pack shot into a video."

Flux 3 First Last Frame to Video is the catalog row for that interpolation: start frame and end frame, native audio, up to 1080p, durations from 5 seconds through 20. The listed rate is 9 credits per second. Audio is on. Write the mix — ice, a quiet counter — or write silence. A silent brief still gets a soundtrack you did not choose. Do not put the ratio in that mix.

Prompt the travel, not the objects. The stills already contain the bottle and the glass. Name the mechanism: hold, then the pour-state change, camera locked, no new kitchen. If the endpoints share geometry, the morph reads as dilution. If they do not, you get a mushy crossfade you could have made in an editor without a video bill.

Duration should match the morph, not a Reels quota. Five honest seconds between two labeled stills beats fifteen seconds of invented condensation. Inspect the midpoint. If the middle grows a second bottle, reroll. Do not regenerate both stills "to match the motion." The stills are the contract.

Ratio copy is overlay, not voice

"1 to 4," "add water," the yield line — those are claims on a concentrate label. They do not belong in a generated mouth, and they do not belong as Flux dialogue. Native audio on this row is a bed, not a nutrition panel.

Add a text overlay burns one authored line for the whole clip: top, center, or bottom. Type the ratio exactly as the label prints it. Place it off the bottle face. This tool does not transcribe. Nothing in the interpolation is speaking the number.

Do not:

• Prompt "a warm narrator explains the dilution."
• Ask Flux to letter the ratio onto the glass. It will misspell it for five to twenty seconds.
• Film a talking host holding a cousin bottle.

Muted autoplay still needs the number. Overlay is how the number survives mute. A voiceover of the same line is a second job on a real person after picture lock. Most concentrate ads never need that mouth. Two lines — ratio, then a CTA — is a timed overlay on another pass, not a talking-head row.

This is not cafe /for or roaster /for

Cafes sell a morning ritual in a room: pour-over, window seat, a barista calling a regular. Four posts a week, 9:16, Reels and TikTok. That plate is the shop. A concentrate SKU that I2Vs a generated marble counter and captions it as your bottle has used the wrong still.

Coffee roasters are origin, roast drum, cupping notes. That is process on green and brown, not a diluted glass of this concentrate. Do not steal the origin-story playbook and paste a first-last of two random tumblers into it.

This page is only the two-state pack: closed concentrate, then the glass at the printed ratio. Your bottle is not their espresso machine. Their corner table is not your dilution.

Do you have both stills, and is the point to travel between them without inventing a kitchen? Yes → first-last, then overlay. Only the bottle → you do not have the product yet. A cinematic "cold brew lifestyle" paragraph → you are in the wrong family.

FAQ

Can I animate one bottle photo and hope it becomes a glass?

You can. The ending is then the model's guess. Concentrate is the diluted glass at the ratio you print. Pin that glass as the last frame or you are not selling dilution.

Why not put the ratio in the Flux 3 soundtrack?

Flux 3 First Last Frame to Video has native audio. That is a bed you either write or accidentally receive. Ratio copy is a label claim. Burn it with add a text overlay. Do not hope a mix letters a number.

Do the two photos have to be shot on the same day?

They have to share geometry. Same bottle glass, same counter family, same lens height. Noon bottle into a night cafe window is a time-lapse you should not sell as a ratio. A different tumbler is a morph.

Is this the same as uploading one image to image-to-video?

No. One image is open-ended motion. Two images is a bounded landing. The agent job notes that this capability needs first frame and last frame. Missing the glass is a different job.