How to write first-frame notes that keep an AI storyboard grounded

Observe the frame before describing the advertisement

Look at the selected image without considering the campaign idea. Count the products, name their visible state, locate the support surface and identify any person or contact. A concise observation might say: one closed blue travel cup on a pale desk, its front label facing the camera, window light from the left, no hands. Each statement can be checked by another reviewer.

Avoid adding a desirable interpretation to that record. A cup beside a laptop establishes a desk setting; it does not prove that the cup is designed for long working days or keeps a drink hot. Put verified specifications in the product-facts field and put the intended feeling in the creative idea. This division makes later disagreements easier to diagnose.

Use three different containers for information

The same distinction applies to lighting. You can observe a bright patch on the left of a bottle. You cannot necessarily identify the exact lamp that caused it. Write only the level of detail the image supports; a guessed softbox position can introduce an unnecessary contradiction.

Describe uncertainty without turning it into a feature

If the source is too small to read the model name, record that limitation and inspect a better copy. Do not ask the text planner to resolve the letters from context. If a reflection hides the edge of a silver lid, describe the edge as unclear rather than specifying a shape you cannot verify.

For an illustrative fabric pouch, the visible front may show a zipper pull but not how the compartment opens. The useful note is that the pouch is closed and only the front is supplied. An opening demonstration remains unsupported even if the product description confirms that the pouch has two compartments. A fact about construction is not a visual record of the missing movement.

Enter the notes without duplicating the entire brief

In Product videos, use the optional first-frame field for a short inventory and any important spatial relationship. The current field accepts up to 500 characters. Spend those characters on details that constrain the shot: an already visible person, a closed lid, a single object or a label near an edge. Repeating the whole product description reduces the space for useful observations.

Runway distinguishes the visual information supplied by an image from the motion described in text. Itemlume adds a customer-written observation step because its storyboard planner does not inspect the uploaded image. These are different stages with different inputs. Read Runway’s image-to-video guidance.

Replace the image and review the notes together

Changing from a product-only photograph to a model scene changes the evidence even if the SKU stays the same. Recheck the notes, the people setting and every proposed action before approving video generation. The image check can report dimensions or a framing mismatch; it does not certify that your description matches the pixels.

A final read-through should answer two questions: could someone verify every observation from this image, and does the storyboard ask for anything the image lacks? Correct a mistaken observation in the notes. Prepare a different source for a genuinely missing element. Keep the resulting plan specific enough to review without presenting the planner as an eyewitness.