kling-kolors-image

What it does

Plans, generates, edits, references, and quality-controls still images using Kling AI IMAGE surfaces (3.0, Omni) or Open Kolors checkpoints via a Kling CLI/MCP workflow. It covers choosing between hosted surfaces and local Kolors, framing production decisions, and documenting requirements before a generation submission. It instructs how to frame deliverables, references, and invariants, and how to track rights, costs, and surface capabilities before producing a single image or a small batch.

How it works

• Resolve surface terminology between Kling image offerings (Kling IMAGE 3.0, Kling IMAGE 3.0 Omni, Open Kolors) and avoid assuming surface identity.
• Provide a production decision frame before generation, detailing deliverables, subject, environment, references, invariants, rights, and cost considerations.
• Choose a generation path based on the need: IMAGE 3.0 for single or edited images, IMAGE 3.0 Omni for series or 2K/4K outputs, local Kolors for on-premises data control, or legacy hosted models only when entitlement is verified.
• Use the official CLI/MCP workflow to run generation, with explicit steps to install the Kling CLI, login, and call help commands (kling who_am_i, kling text_to_image --help, kling image_to_image --help).
• Before paid submissions, collect model discovery, pricing estimates, and a generation plan hash; persist the plan and use task queries to reconcile credits.
• Follow example flows for a single-use CLI submission with explicit argv, including --model, --aspectRatio, --imgResolution, and --imageCount 1; provide one or more local --image inputs as needed.

When to use it

• Use when planning and executing still-image generation with Kling Kolors surfaces, ensuring you respect surface capabilities, licensing, and data locality.
• Use when you need to document requirements, costs, references, and invariants prior to production submission.

What it can touch

• Official Kling CLI and MCP tooling as the primary interaction surface (kling, who_am_i, text_to_image, image_to_image).
• Surface selection among IMAGE 3.0, IMAGE 3.0 Omni, Open Kolors, and legacy hosted models based on entitlement and live capability discovery.

Caveats

• Do not assume identity or parity between different Kling surfaces (e.g., 3.0 vs Omni vs Open Platform) without direct live verification.
• Ensure you use the correct CLI for the site (global vs China) and keep credentials in a secure store, not embedded in code.
• Price and token details are surface- and time-dependent; rely on live console/pricing surfaces rather than static assumptions.
• The plan requires explicit approval of a printed hash before execution and cannot be replayed automatically under the same attempt ID.