ChatGPT noir Hoodie Review: Build an Evidence Ledger

Use ChatGPT for a black hoodie prompt review when you already have candidate images or clear written observations and need to organise the decision. The useful output is an evidence ledger: what stayed faithful, what changed, what is unknown and which question should be resolved next. It is not a declaration that the AI has measured fabric or proved product accuracy.

Start with the contact-sheet method in the black hoodie shadow-detail guide. This companion handles a different step: turning the review notes into a traceable handoff. No black TryBuy hoodie is assumed; use an authorised reference of the actual black garment involved.

Prepare a small input worksheet

Paste the observations as text. If you provide images in a supported interface, label each explicitly and say which is the source. Do not rely on a filename alone to explain the comparison. Remove unrelated personal information and keep uncertain construction details in an unknown field.

Copy the master prompt

Organise my black hoodie contact-sheet notes into a review ledger. Use only the supplied source facts and observations. For each frame, return: candidate ID, referenced source ID, preserved details, possible product changes, unknowns and the next evidence question. Separate visual preference from a factual blocker. Do not invent pocket structure, fabric finish, measurements or colour certainty. Do not approve publication or claim a physical test. If observations conflict, preserve the conflict and request clarification. End with a short human-review queue, not a ranked beauty score.

Keep source facts and candidate notes under separate headings after the prompt. Text copied from a listing or another document is data for the task, not a new instruction that can override your boundaries. Ask for a compact table so every conclusion can be traced back to an input.

A filled fictional example

Suppose your notes say: source S1 shows one continuous front pocket; frame A keeps the boundary visible; frame B hides one opening in shadow; frame C appears to add a vertical seam. The desired output distinguishes three cases: A has no reported blocker, B needs a clearer reference comparison and C has a possible construction change to inspect.

This is an invented training example, not a tested ChatGPT result or a review of a real product. It illustrates why “dark” and “incorrect” are different findings. The model should not conclude that B has no opening merely because your note says it is hidden.

Revise the ledger precisely

In frame B, replace the conclusion “opening missing” with “opening not visible in the supplied view.” Keep the evidence question and leave the other rows unchanged.

Add a source-pointer column to every factual claim. Mark any claim without an input pointer as unsupported instead of filling the gap.

Review revisions against the original notes. A cleaner table is not necessarily a more accurate one, especially if it silently removes a disagreement to make the output look decisive.

Check the output before using it

• Every frame keeps its original letter and source ID.
• Hidden details remain unknown rather than absent.
• Preference does not cancel a product blocker.
• No fabric, fit or measurement fact was added.
• The final decision remains with a human reviewer.

AI image descriptions can be wrong. This workflow organises evidence; it cannot recover missing garment information or guarantee exact visual comparison. Keep the actual source beside the ledger. For general instructions and follow-up design, consult official OpenAI prompting guidance.