nano banana prompts: model prompt

Define the model-specific prompt plan behind nano banana prompts

For Nano Banana Prompts, state the subject, environment, action objective, emotional beat, and end condition in plain production language. These invariants keep model-specific prompts variants comparable instead of producing unrelated ideas. Resolve the exact named model or version, then keep creative direction separate from claims about access, syntax, limits, price, audio, duration, quality, or availability. Use only dated first-party evidence for those changing facts. Write the decision owner and intended handoff beside the brief so a search phrase cannot be mistaken for a verified product capability or an instruction that has already been executed. For a controlled test, preserve the reference set, and ask the creative lead to record identity consent before the acceptance review.

Apply the model-specific prompts structure

Write a portable core prompt before adding any vendor or version vocabulary. Keep the subject, action, spatial relationship, shot order, continuity anchors, and end condition independent from a named model. Put model-specific controls in a separate dated note so an unsupported parameter cannot quietly become part of the creative brief. For “nano banana prompts”, keep a visible distinction between required content, optional treatment, exclusions, and facts that need evidence. That structure gives controlled variants a stable baseline and lets a reviewer identify which instruction caused a material change. For a controlled test, preserve the decision history, and ask the claims reviewer to record source fidelity before the workflow transfer.

• Portable shot contract At this stage, preserve the decision history, and ask the rights reviewer to record identity consent before the final sign-off.
• Dated model-control notes At this stage, preserve the handoff draft, and ask the model evaluator to record visible continuity before the final sign-off.
• Version and access verification At this stage, preserve the input snapshot, and ask the delivery owner to record revision intent before the final sign-off.

Prepare authorized inputs for general creative direction

This query calls for the intended medium, audience, purpose, authorized sources, subject, action, context, constraints, reviewer, and delivery condition. Turn broad or ambiguous words into observable choices. Define one outcome, order the important decisions, remove contradictions, and keep optional style language separate from required content. Record the source, permission, intended audience, protected details, and reviewer before testing language. The goal is a traceable creative proposal, not an assumption that a named platform, model, or workspace will perform the requested action. At this stage, preserve the review copy, and ask the production lead to record human approval before the scope confirmation.

Build controlled variants and observable acceptance checks

For this owner, the first review pass should make nano, banana observable rather than merely decorative. Change one meaningful dimension per version and label the hypothesis, invariant details, expected difference, and rejection condition. The brief is reviewable when its creative core remains portable and every model-dependent statement is either cited with a date or clearly labeled as an open test question. Keep failed variants in the decision record so the final wording is explainable rather than selected by impression alone. For the working record, preserve the delivery checklist, and ask the delivery owner to record revision intent before the evidence refresh.

Review authorization, safety, and changing product facts

Treat unclear product, entity, and platform references as questions. Do not invent capabilities, facts, identities, access terms, or production results to fill a gap in the query. The named third-party model or product requires dated first-party verification for identity, version, access, inputs, controls, limits, pricing, licensing, and output behavior; this page makes no comparison or SEELE availability claim. Treat every named model, competitor, access term, platform rule, identity use, commercial claim, and output specification as a dated evidence question. If authorization or first-party support is missing, keep the language editorial, stop the handoff, and record the unresolved gap instead of inventing a workaround. Before moving on, preserve the failure note, and ask the model evaluator to record evidence freshness before the production checkpoint.

Approve a reversible prompt handoff

Confirm the exact provider, version, access surface, accepted inputs, controls, and output behavior against current first-party material before a test. A model name in a query is an evaluation target, not proof that SEELE or any other product currently exposes it. Before approving “nano banana prompts”, check purpose, source rights, identity consent, factual support, continuity, disclosure, destination requirements, and the named review owner. Preserve the original brief and version notes. This handoff supplies authored guidance only and neither runs generation nor guarantees access, quality, speed, licensing, publishing, downloads, or outcomes. Before moving on, preserve the continuity note, and ask the source custodian to record source fidelity before the release review.