Visual Wardrobe and Outfit Concepts

Creator's demonstrated workflow

The supplied source demonstrates this sequence:

1. The creator gave 5.6 sol access to their camera roll.
2. The creator had 5.6 sol extract pictures of every piece of clothing they own from their photos.
3. The creator then told it to find new outfits for them.
4. The creator had the outfits rendered on me with GPT-image.
5. The creator describes seeing their entire wardrobe in a collection as “kinda cool.”

The source does not state how 5.6 sol identified clothing, how the collection was structured, how outfits were selected, or what controls were used for camera-roll access or GPT-image rendering.

XSkills operational adaptations

Use these adaptations to make the demonstrated sequence reviewable and executable. They are not claimed to be part of the creator’s workflow.

Inputs

Collect the following before beginning:

1. Authorized photo scope: selected uploads, a named album, or a user-exported folder containing the photos to inspect.
2. Wardrobe objective: confirmation that the user wants pictures of clothing they own extracted into a wardrobe collection and wants new outfits found from that collection.
3. Rendering request: whether the user wants selected outfits rendered on them with GPT-image.
4. Authorized reference image: only when rendering an outfit on the user requires a reference image and the available rendering setup requires one.
5. Optional constraints: occasion, desired number of outfits, garments to exclude, and preferred style direction.

Execution procedure

1. Confirm the scoped wardrobe task

Action - XSkills operational adaptation: Ask the user to define the photo set and confirm the intended result.

Example:

I will inspect only [the specified uploads/album/export] to extract pictures of clothing you own into a wardrobe collection, find new outfits from the identified clothing, and-if you request it-prepare selected outfits for rendering on you with GPT-image. Confirm this scope before I begin.

Check: The user identifies a specific photo set and confirms the wardrobe task.

Fallback: If camera-roll access is unavailable, use user-uploaded images or a user-exported folder instead.

Stop condition: Stop before inspecting photos if the image scope is not clearly authorized.

2. Extract clothing pictures into a collection

Action - XSkills operational adaptation: Inspect the authorized photos and create a provisional collection entry for each visible clothing item or clothing image.

Use conservative, visible-only descriptions. Distinguish among:

• a single garment image;
• multiple garments in one image;
• a worn outfit where components cannot be reliably separated;
• shoes, bags, or accessories, if the user wants them included;
• non-clothing, duplicate-looking, unclear, or unusable images.

Create entries using this reviewable schema:

Check: Each extracted entry links to an authorized source image and does not state unsupported details such as brand, fabric, size, or fit.

Output: A provisional wardrobe collection representing the clothing visible in the authorized photo set.

Fallback: When a garment cannot be separated from a worn outfit, retain one entry labeled full outfit / components uncertain rather than inventing separate pieces.

Stop condition: Stop extraction and request a clearer photo or user clarification when a required item cannot be identified sufficiently for the requested outfit task.

3. Review the wardrobe collection

Action - XSkills operational adaptation: Present ambiguous, duplicate-looking, multi-item, and low-confidence entries for user review before treating them as reliable wardrobe pieces.

Ask targeted questions, for example:

• Are these two similar trousers separate items?
• Should this full-outfit image be treated as one outfit or split into individual pieces?
• Should shoes and accessories be included when finding new outfits?

Check: Record user corrections and preserve uncertainty where no correction is provided.

Output: An approved working collection plus a clearly labeled pending or excluded queue.

Fallback: If the user does not review the collection, proceed only with high-confidence entries and label the resulting outfit suggestions as provisional.

4. Find new outfits

Action - XSkills operational adaptation: Generate new outfit combinations using the approved wardrobe collection. Reference the collection’s item IDs so the user can verify what each outfit contains.

For each proposed outfit, provide:

• a short look name;
• included item IDs and visible descriptions;
• the user-provided occasion or style objective, when supplied;
• a concise visual-coordination rationale based only on visible garments;
• any uncertain item or pairing requiring review.

Check: Each outfit uses collection entries rather than unverified garments. Do not state that the outfit will fit, flatter, or accurately reproduce real-world color.

Output: A set of new outfit suggestions traceable to the wardrobe collection.

Fallback: If the collection lacks sufficient confirmed pieces for a requested outfit, state the limitation and offer fewer combinations or combinations with clearly marked uncertainty.

Stop condition: Do not send a proposed outfit to rendering until the user selects or approves it for that purpose.

5. Render selected outfits on the user with GPT-image

The demonstrated rendering tool is GPT-image.

Action - XSkills operational adaptation: For each user-selected outfit, prepare a render brief that identifies the approved collection items and the authorized person to depict. Use GPT-image only when it is available in the current environment and the user has authorized the depiction.

Use a brief such as:

Render an outfit concept on the authorized user.
Outfit: [approved item IDs and visible descriptions].
Preserve only supported visible garment details, including confirmed colors or patterns.
Do not add unsupported logos, fabric composition, sizing, tailoring, or fit details.
Pose and setting: [user preference].

Check: Confirm all of the following before rendering:

• the user selected the outfit;
• the depicted person is the user, or the user has authority to use that person’s likeness;
• the selected outfit is traceable to approved collection entries;
• the render request is non-sensitive and appropriate.

Output: A GPT-image render, labeled as a visual outfit concept, when GPT-image is available and authorized.

Fallback: If GPT-image is unavailable or fails, return the selected item IDs and render brief for later use. Do not claim that a render was created.

Stop condition: Do not render a person without authorization to use their likeness. For sensitive depiction requests where age is unclear, stop and request a safe, non-sensitive alternative.

Delivery format

Action - XSkills operational adaptation: Return the following, as applicable:

1. The wardrobe collection, with approved, pending, corrected, and excluded entries clearly labeled.
2. New outfit suggestions referencing collection item IDs.
3. Selected GPT-image render briefs.
4. GPT-image outputs, if actually generated and authorized.
5. A correction path: add, correct, exclude, merge, recategorize, choose another outfit, or rerender from corrected entries.

XSkills safety adaptations

These safety measures are XSkills additions and are not established by the supplied source.

• Obtain explicit authorization for a defined photo set before inspecting a camera roll, photo library, album, folder, or export.
• Do not imply that camera-roll access exists unless it has actually been granted and is available in the current environment.
• Keep photo inspection limited to the wardrobe task. Do not inspect or disclose unrelated images or people.
• Treat wardrobe images and likeness references as sensitive personal data. Do not reuse them for unrelated tasks.
• Use a user-uploaded or user-exported photo set when direct access is unavailable; do not bypass device, account, or privacy controls.
• Make the wardrobe collection reviewable. Do not silently merge, delete, or treat uncertain items as confirmed.
• Describe only visible image details. Do not infer exact brands, authenticity, material, size, fit, body shape, ownership beyond the user’s stated objective, or precise real-world colors.
• Require authorization before rendering a person’s likeness with GPT-image. Label generated outputs as visual concepts rather than virtual try-ons or guarantees of garment fidelity.
• Do not generate sexualized or otherwise inappropriate depictions, particularly when age is unclear.
• Do not run scripts, alter photo collections, or create persistent artifacts without explicit user direction.
• If the requested access, likeness authorization, or photo scope is missing or ambiguous, stop and request clarification.

Source attribution

Source: source-1 - creator post describing giving 5.6 sol camera-roll access, extracting pictures of every owned clothing item, finding new outfits, rendering them on the creator with GPT-image, and viewing the wardrobe in a collection: https://x.com/cdngdev/status/2076812846793650485

Independently compiled by XSkills from the supplied source. Thijs did not author or review this skill. Attribution does not imply endorsement.