ChatGPT IA Guide: Separate kurta Campaign Mood From Product Evidence

By TryBuy Editorial

Use ChatGPT to separate an expressive kurta campaign image from the sharp reference that supports product information. The output should assign each asset a role and audit the claims its caption can reasonably make. It should not treat motion blur as evidence of fabric behaviour or imply that a generated scene documents a real wearer’s performance.

Prepare two asset records

Identify the clear source view and the atmospheric candidate separately. Include the exact garment, authorised source, production type and intended placement for each. Add your own notes about what remains visible and what the blur obscures. Keep a planned motion treatment distinct from an observation of the resulting image.

Supply any product statements the article needs to support, using the actual record rather than the mood description. A claim about a neckline, border or sleeve needs appropriate evidence. Do not ask the model to infer material flow, breathability, comfort or durability from a dramatic visual effect. Those are not established by a source-inspired campaign concept.

Copy the role-and-claim prompt

Organise my sharp-reference and campaign-concept records into a two-asset role matrix. Return asset ID, production type, intended communication job, visible information, obscured information, supported caption and claim to avoid. Map each product statement to the appropriate source rather than the atmospheric image by default. Do not infer fabric performance, real movement or a completed event. Flag garment changes reported in my notes and distinguish them from intentional blur. Draft a clear handoff preserving both asset roles and unresolved questions.

Ask for unsupported claims to remain in a separate review list. A model should not solve a mismatch by weakening the caption so much that a visibly changed product still appears acceptable for its intended use.

Filled fictional example

Imagine the fictional campaign candidate preserves the visible collar but makes the lower border unreadable. The sharp source clearly shows the documented border. The matrix can assign atmosphere to the first asset and source-supported detail to the second. It should not claim that the blurred border is verified merely because the product is recognisable elsewhere.

If another candidate extends the border into a longer panel, keep that as a reported garment-change concern rather than a stronger expression of movement. The output should route it to source comparison or correction. The example is an editorial decision scenario, not an actual campaign test or proof that any image model preserves cloth accurately.

Use bounded revisions

• “Identify every caption phrase that turns a visual effect into a physical product claim.”
• “Keep the clear reference attached to the product statement even when the campaign image is the opener.”
• “Draft a correction request for the reported border change without removing the intended atmosphere from the separate concept brief.”

When a layout changes, update the asset-role record too. A clear image that has become a tiny thumbnail may no longer support the reader’s intended inspection. Keep the actual viewing role, not only the original production label, in the matrix.

Review both image and copy directly

Check the sharp source and campaign candidate yourself, then read the captions in the final layout. Confirm that any generated-concept label remains attached where needed and that a product card points to the actual garment. The text matrix cannot demonstrate fabric motion, verify a physical fit or inspect the final export automatically.

Use the original practical guide for the practical two-asset campaign plan and the men’s kurtas collection for current garment information. Save the role matrix, caption audit and selected version IDs together. The deliverable should preserve expressive storytelling while ensuring that product facts are supported by the appropriate clear reference, rather than by the persuasive mood of an image.

For general instructions and follow-up techniques, see official OpenIA prompting guidance.