Fertilizer: bag still, no fake lawn science presenter

SKU + captions. The product is the bag: analysis, net weight, the colourway a retailer already knows. A generated agronomist in a polo explaining "lawn science" is a health-adjacent expert in a costume. YouTube's third inauthentic-content bucket names AI personas that present as human experts giving advice on health. Wellness remedies are in the Help example on purpose. A synthetic lawn doctor is that shape wearing turf.

Synthetic presenters on health and finance are a YPP kill is the policy. This post is the bag. Photograph the SKU. Caption the grade. Do not mint a scientist.

The bag is the contract

Front of pack, three-quarter, any panel that carries the analysis. Even light, plain background. Those stills are what must not change. A generate that restyles the N-P-K block is a mislabelled product, not a style pass.

Keep the bag large in frame. Fine print is the first thing a model loses. Zoom to 100% and read the analysis before you keep a take. If the still would not ship on a pallet, do not buy motion.

Text-to-video of "lush suburban lawn, golden hour, bag of premium fertilizer" will invent a bag. Retailers do not buy invented bags. Identity is the photo. Motion is optional.

Motion from the still, not from a lecture

Once the pack is locked, image-to-video is the next credit: the upload is frame one, the prompt is camera and atmosphere only — a slow push, granules catching light, nothing else in the paragraph restating the brand. Seedance 2.5 Image to Video is one catalog row built for that job: a still in, a native clip out, durations from 4 seconds through a 30-second single pass, up to 720p, with an optional last frame if you own the closing composition. It requires the image. It does not require a presenter.

Do not use a 30-second hold as a place to hide a talking head. A long I2V pass on a bag is still a bag. Stretching it so a generated expert can "explain the soil food web" is how you spend a long-form row on a YPP problem.

The live door for "animate this still" is AI image-to-video. Describe only what moves. Leave the analysis out of the prompt so the model is not invited to redraw the numbers.

No lawn-science presenter

An AI "doctor" on wellness remedies is the Help example. A lab-coat generate prescribing a spring protocol is the same grammar: synthetic persona, health-adjacent advice, expert framing. "Here's what you should put down this weekend" from a constructed face is the bucket. A bag with the grade burned on is not.

You do not have to abandon lawn content. You have to stop using a fake expert as the source. Third-person on-pack facts — analysis, coverage statement you already print, a real seasonal calendar as text — belong on the plate. Second-person treatment advice belongs to a named human you actually employ, or it does not ship.

Captions: if a real employee spoke, transcribe that tape. If the plate is silent, burn the SKU line you wrote. Auto-captions will not typeset an N-P-K block from a mute bag loop.

The test

Mute it. Can you read the SKU? Is anyone playing scientist?

If the bag is readable and no persona is prescribing, you have the ad. If a generated face is giving a protocol, you have a presenter problem the bag did not need. Ship the SKU.

FAQ

Can we use a mascot instead of a scientist?

A mascot that is obviously a character, not a human expert, is not the Help example. The moment it puts on a lab coat and tells the viewer what to apply, you are back in the bucket. Keep it as a stamp, not as an agronomist.

Should the analysis live in the motion prompt?

No. The still already has it. Repeating numbers in text is how they get redrawn. Caption or overlay the line you approved after picture lock.

Is a 30-second bag loop the point?

Only if the still holds for that long. Seedance 2.5 Image to Video can run a native pass up to 30 seconds from one frame. If the label crawls at second twelve, shorten. A clean 6-second pack beat beats a long take of soup.

Does labelling the presenter as AI fix it?

No. Disclosure does not move a video out of the third bucket. Change the format: bag, captions, no synthetic clinician of turf.