Pack shots: lock the still, then decide I2V vs a hook pack
The body is one generate. Openings are a hook pack.
The body is one generate. Openings are a hook pack.
A product photographer already has the expensive part: a still someone would print on a box. The next credit is not "make more of the bottle." Animate that plate once for the demo that has to stay the SKU. Test the first three seconds as a separate object. Mixing those jobs is how a locked pack shot becomes five restaged cousins.
Lock the still before you spend on motion is the credit argument. Video meters are a terrible identity debugger. This page is the fork that sits on top of it: one body generate versus a pack of openings.
Sign the plate before you buy seconds
A still is not locked because the studio exported a JPEG. It is locked when label, colourway, angle, and crop would ship on a box. If the lockup is soft, every second of motion is a recall.
Image-to-video treats that plate as frame one. Composition stays. The model invents camera and physics forward from the picture you signed. That is the body: a slow orbit, a light rake, a three-quarter hold that still reads as the listing photo. Prompt only the move.
Text-to-video is how you start identity retries. "Our sage bottle on marble, label sharp" is a casting session. Photographers who already lit the bottle need the file they already retouched, moving. If the still would fail a listing audit, stop. Do not "see if motion hides it."
The body is one file, not a folder of I2V takes
Once the plate is signed, generate the body once. One continuous still-to-motion pass is the product demo: the SKU stays itself while the camera moves. Regenerating the same plate ten times hoping the first three seconds "feel more hooky" is ten bills for one idea.
That body does not have to be a film. It has to be the proof the opening will cut into — unbox, pour, a label-forward hold. If the ad is a Reel, generate vertical from a vertical still. Recropping a 16:9 pack after the fact is how the lockup you signed falls out of frame.
Five near-duplicate I2V takes are not openings. They all start on the same picture. Variance of camera speed is not variance of sentence.
Openings are a pack against that body
Create a branded hook pack is the named job for the first three seconds. Brief the brand in a sentence or two. The agent plans N distinct hooks so they do not repeat each other, charges once for the set, and submits every clip as a 1080p portrait job on Vidu Q3 with the line burned on as big TikTok-style text. That billing shape is why this is not five separate video generates.
The pack does not generate the body. It does not pick Seedance versus Veo for the bottle. Vidu Q3 is fixed so the variants are comparable. You pick a winner cold, then join it to the I2V keeper you already trust.
If you paste the five lines you already like, you have turned the pack into a renderer and thrown away planning. Brief the offer. Count three to read, five or more to test. Confirm the one charge. Animate the plate that ships. Test openings as a pack.
FAQ
Can the hook pack also be the product demo?
No. The pack is N portrait opens with burned-in lines. The demo is the SKU still in motion. Join them. Do not ask the pack to orbit a label, and do not ask I2V to invent five first sentences.
Should I I2V every still in the gallery?
No. Lock one hero plate. Animate that. The rest of the gallery is stills. Motion credits are for the shot that has to move, not for proving you have more JPEGs.
Does a five-hook pack cost five video jobs?
No. The pack is charged as a set, shown before you confirm. Width is a test-design choice, not a cart quantity.
What if the still is right and the body still drifts?
Then you have a motion problem, not a hook-pack problem. Shorten the shot, simplify the action, or change the I2V row. Do not "fix" a drifted label by generating ten new openings.