magic-hour-product-visuals
Create or edit a Magic Hour product image with controlled composition, geometry, branding and copy space. Use for product shots, hero images, packaging and ecommerce visuals when Magic Hour is requested or is the project's chosen media provider.
- License 许可
- MIT
- 最近更新
- 2026-11-14
- Stars
- ★ 1.6k
- Forks
- 491
01 前言 Front Matter
- name
- magic-hour-product-visuals
- license
- MIT
- metadata
- { "author": "magichourhq", "version": "1.3.0" }
- description
- Create or edit a Magic Hour product image with controlled composition, geometry, branding and copy space. Use for product shots, hero images, packaging and ecommerce visuals when Magic Hour is requested or is the project's chosen media provider.
02 资源文件 4
- 参考文档agents/openai.yaml266 B
- 许可证LICENSE.txt1.1 KB
- 参考文档references/photoshoot.md3.7 KB
- 参考文档references/setup.md1.2 KB
03 SKILL.md 正文
Magic Hour product visuals
Deliver a product image that fits its intended placement and preserves the details that make the product recognizable. Treat a project ID as an intermediate result and inspect the finished image before calling it complete.
Generation consumes Magic Hour credits. Use a connected Magic Hour creation MCP at https://mcp.magichour.ai/ when available. Otherwise read references/setup.md. Discover the current schema before choosing a model, resolution, image count, or cost.
Infer the treatment from the customer's intended use; recommend one direction instead of asking them to choose technical parameters. Honor existing budget and delegated creative decisions, inspect intermediate results yourself and continue unless the user requested draft approval. Show the finished image first; retain technical records with the project. For catalog, lifestyle, detail, in-hand or coordinated sets, read reference photoshoot.
Define the deliverable
Extract or infer:
- where the image will appear, such as a product page, ad, marketplace listing, or website hero
- aspect ratio, resolution, crop tolerance, and required negative space
- product shape, material, finish, color, label, logo, and exact text
- camera angle, scale in frame, surface, background, lighting, and shadow
- details that must remain unchanged
- visible failures that make the image unusable
For a hero, specify which side needs empty space for copy. For an ecommerce listing, keep the full product inside the frame with a clean silhouette. When exact product identity, packaging, or branding matters, use an authorized reference and edit it rather than recreating it from text.
Choose generation or editing
- With no source image, call
generate_image. - With a usable source that needs a new setting, cleanup, relighting, repositioning, or other controlled change, call
edit_image. - AI Image Editor requires an input image. Never route a prompt-only creation request to it.
Use one image at a resolution that reveals the product's critical details and fits its placement. A low-resolution preview is useful for composition, not final label fidelity. Check cost against existing authorization and ask only when it does not cover the intended work. Prefer the live recommended model for a general request and choose a specialist only when the current schema describes the needed behavior.
Write a production brief
Write the prompt in this order:
- intended image type and aspect ratio
- one product with material, shape, color, and orientation
- exact composition and space allocation
- lighting, surface, background, shadow, and reflection
- required label, logo, or text in quotation marks
- elements that must remain fixed
- concrete rejection conditions
Use direct visual language. A useful pattern is:
Premium 16:9 product hero photograph. One [product] in the right third, fully visible, with clean negative space across the left 40 percent. [Camera and material]. [Lighting, surface, and background]. Preserve [shape, colors, logo, and exact label]. No extra objects, duplicate text, warped geometry, unintended marks, or watermark.
For editing, describe the change first and then state what remains unchanged. Compare against the original reference, not just the prompt: a prettier bottle with a different cap or viewpoint is still a failed preservation edit. Do not stack unrelated styles or ask one image to show several scenes. Keep promotional copy in a separate layout layer when a suitable tool is available; preserve text already printed on the product.
Inspect before delivery
Wait for the existing image project with wait_for_image_project. After it reports complete, use the exact returned download URL and inspect the actual output for:
- actual downloaded dimensions, aspect ratio, crop, product scale, and requested negative space; do not trust the requested ratio alone
- straight edges, plausible materials, and coherent reflections and shadows
- accurate product geometry, cap, handle, controls, packaging, and colors
- one correct instance of required text or logo
- no misspelling, duplicate label, invented brand-like mark, or distracting object; disclose any service watermark and do not promise an unwatermarked file without checking eligibility
- no anatomy or identity problem when a person is intentionally present
If the result fails, change the smallest useful variable. Use AI Image Editor for a localized correction when the rest of the image is strong; regenerate when composition or product identity is fundamentally wrong. Another generation spends credits, so stay inside the user's authorized budget.
When the product image will become video, accept the still first and use the image-to-video skill when installed. Otherwise discover create_video_from_image, pass the accepted image, prompt for motion, retain the ID, wait and inspect the complete clip. Do not require another skill installation to finish a supported request. Return the finished artifact, its saved location and any material limitation; retain project IDs for recovery.
04 同仓库更多 Skills
worktable-review-with-annotations
Create, inspect, reply to, update, or resolve situated Worktable annotations. Use when feedback belongs to a Doc section, block, text range, or HTML Doc, or when the user asks to address existing review feedback and coordinate a requested artifact edit.
worktable-find-and-synthesize-context
Find, assess, and synthesize context from Worktable Docs and Records. Use when the user asks what a workspace contains, wants workspace-backed research, needs current facts reconciled across artifacts, or requests a cited synthesis without changing workspace state.
worktable-create-or-update-docs
Create, edit, rename, or maintain narrative Worktable Docs. Use for plans, decisions, briefs, notes, research syntheses, and other durable prose when the user requests a new artifact or a safe update to an existing one.
worktable-create-or-manage-records
Organize repeated structured information as durable Worktable Records. Use for research datasets, inventories, directories, pipelines, trackers, and other collections whose items need identity, fields, filtering, links, validation, or individual updates.
worktable-create-or-manage-interactive-html
Create or improve durable visual and interactive artifacts in Worktable. Use for dashboards, trackers, calculators, mini apps, slide presentations, and other requests that benefit from a polished HTML Doc.
worktable-collaborate-threads
Read, answer, and hand off messages through durable Worktable threads with users or other authenticated participants. Use when the user asks to check addressed messages, reply to a participant, continue a thread, request another agent's help, wait briefly for a response, or watch for comments and messages sent to the agent.