Screen print: the shirt SKU cannot morph

Seedance I2V from the mock. The print is the SKU. Motion is allowed. A new illustration halfway through the clip is not. If the lockup swims, the ink colour drifts, or a third mascot appears at second twelve, you do not have a product video. You have a cousin of the shirt you cannot sell.

Seedance 2.5 Image to Video is the still-to-motion row: one approved still in, a continuous clip out, up to 720p, four to thirty seconds, native audio with the picture. Seedance I2V is still the still-to-motion job is the router. Text-to-video Elo moving next week does not change what a mock is for.

The mock is the product

A screen-print mock is a locked still: garment colour, placement, the actual art that will go through a screen. It is not a paragraph about "a vintage band tee on a model in natural light." Paragraphs invent tees. Image-to-video does not get to reopen the art direction. Frame one is the file you sent.

That is the same reason print-on-demand sellers test the design, not the inventory. Attention risk is whether the art stops a niche mid-scroll. Identity risk is whether the art in the video is still the art on the press. Those are different failures. A pretty eight seconds of the wrong eagle does not tell you if the real eagle converts.

Do not prompt "exact Pantone of our ink, distressed print, collar stays sharp" as a text-to-video job and hope the model honours a mock you never attached. Make the mock. Then animate it.

Motion around the ink, not through it

The prompt describes camera and physics only. Slow push. Fabric settle. A hangar turn. Steam, if the still already implies a room. Do not describe the illustration. The still already did. Re-describing the mascot is how the model treats the mascot as optional.

Keep the print large in frame. Fine linework is the first thing a generator loses — the same failure as a pack label on a product turn. If the art must read, it has to fill the frame at the start. A wide lifestyle wiggle that shrinks the chest print to a postage stamp is how you "keep identity" in theory and lose it in pixels.

Thirty seconds in one pass is why this family holds a SKU: stitching four short text-to-video clips around a tee is how the print drifts. One pass from a locked mock is how it does not. 720p is the ceiling on this endpoint. Feed and shop clips live there. Identity is still the I2V choice.

A last frame is optional. Add one only when the shot has a destination that must match a second approved still — back of shirt, a folded flat-lay you already shot. A last frame that is a different design is a morph of the SKU. Do not do that and call it the same listing.

First-last is not a free redesign

First-last frame as the product-turn model is for a bottle that must remain the bottle while it rotates. Two stills of the same print, front and three-quarter, is a turn. Two stills of two competing designs is a morph. Screen print cannot morph. The press will not.

If you need to A/B two art directions, generate two I2V clips from two mocks. Do not interpolate between them and hope the middle is a usable third SKU. The middle is where generators invent ink.

Lifestyle around the garment is a split file. I2V for the mock. A talking or scene model for the person, if you have a person. Comp them. Asking one interpolate to be the print and a model and a street is how the chest art becomes a suggestion.

Physical print still has to survive a screen. Gradients that look like one colour on a monitor separate into many passes. That is a merch-file problem, not a motion problem. Do not use video to hide a mock that will not separate. See merch artwork and substrate limits.

The listing test

Pause the clip on the print. Is it the mock you would send to the shop?

Yes → this job. Ship the motion.

No, the eagle grew a second wing → you prompted a scene, or you used text-to-video, or you first-last'd two different arts. Go back to the still.

The still is wrong → fix the still. I2V will not rescue melted type. Every second of video from a bad mock is waste.

Print-on-demand can validate a concept with mockups before a sample lands. It cannot validate a design that the video already redesigned. Hold the ink. Move the camera.

FAQ

Can I text-to-video "our shirt rotating on a hanger" and skip the mock?

You can. You will get a shirt. It will not be the shirt often enough to run as the listing. Seedance I2V exists because the mock is the spec.

Do I need a last frame for a simple hangar wiggle?

No. One hero mock plus a camera move is the usual job. A last frame is control when you must land on a second approved still of the same print.

Why not first-last between the design file and the worn sample?

Only if both plates are the same art, connectable in the duration, and you accept a warp if the crop is unrelated. A transparent PNG of the art to a lifestyle photo of a different placement is not a turn. It is a morph. Prefer I2V from the worn mock once you have a sample.

Is 720p enough for a product ad?

For feed, yes. For a print proof or a 2K plate, you are on a different generate — still start from the locked mock. Resolution is delivery. The SKU is the I2V choice.