Fabric Collage Outfit Transitions · Open Skill 공개 스킬
Fabric Collage Outfit Transitions
Give each outfit its own fabric-inspired backdrop. This workflow creates four full-body fashion collages, with three fabric swatches behind the model and matching samples scattered at their feet. You can make the images on their own or use four finished collages for an outfit-transition video. **What to upload** - A clear photo of an adult for the model’s face and identity. - One horizontal outfit board with four looks, arranged left to right. - A fabric-collage layout sketch to guide the composition. Already have four finished collages? Upload them in A–B–C–D order and use the video workflow. **How it works** 1. Confirm the person, outfit order, and layout reference. 2. Create one collage for each outfit. Use the outfit board for clothing, the identity photo for the face, and the sketch for composition. 3. Review all four images before moving on to video. 4. For video, request one fast clockwise camera orbit with three outfit changes: A → B → C → D. The model stays in place while the camera moves, and the fabric backdrop changes with each look. **Make it your own** Try a seasonal wardrobe, a coordinated color palette, or four different takes on your personal style. Provide your own identity reference for each new person; the template does not include a fixed face or the creator’s private photos. **Before you run it** Model availability, native resolution, and supported video length must be checked before generation. For new images, the template checks for image2 at native 4K. If that setup is unavailable, use finished collages or choose another supported setup before generating. Video motion and identity consistency depend on the selected model. The updated example images were generated in LibTV using a fictional adult model and show the intended style. This template has not yet been tested end to end on Digen, including a different-person run. **Example requests** - Create four fabric-collage fashion photos using my selfie, outfit board, and layout sketch. Keep the person’s identity consistent. Images only for now. - Use these four approved collages, in order, for a fast clockwise camera-orbit outfit transition. Keep the model in place and switch the fabric backdrop with each outfit.
TryOn
3 1 1
https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/4274163/2026-10-13/56n445k4.jpg
--- name: fabric-collage-fashion-orbit display_name: Fabric Collage Outfit Transitions description: Turn an identity photo, a four-look outfit board, and a layout sketch into full-body fabric collages, then create a fast camera-orbit outfit transition. allowed-tools: - generate_image - generate_video - recognize_image - list_models - create_document - update_document - get_document --- # Fabric Collage Outfit Transitions ## When to use this workflow Use this workflow to: - Create four separate 9:16 fashion photos from an adult identity reference, a four-look outfit board, and a fabric-collage layout sketch. Each image shows one full-body model centered in front of three large fabric swatches, with matching samples scattered at their feet. - Turn four approved collages into a short video with one fast clockwise camera orbit and three outfit changes: A -> B -> C -> D. - Use a different person or wardrobe for a new run. Never hard-code the creator's face, private asset IDs, or previous user inputs. ## Inputs Build an asset table with each input's role, user-provided filename, actual asset ID or URI, purpose, order, and version. Upload order is not outfit order. If an input is missing, identify the missing role. Never invent URLs or put a user's local file paths in the public template. - `identity_reference`: A clear photo of an adult or a multi-angle face reference. Use it for the person's face, visible hairstyle, and identity only. Do not copy a studio background, split panels, or divider lines into the final image. - `outfit_board`: A horizontal board with four outfits, ordered A/B/C/D from left to right. Follow the board for clothing, shoes, and accessories. When changing the person, use the board for the wardrobe, not the previous model's face. - `composition_reference`: The original black-and-white layout sketch. Pass this actual image into all four collage generations. Use only the placement, foreground/background layering, and relative size of the person and fabric swatches. Ignore the sketch's person, clothing, text, arrows, and drawing style. Never omit this reference and claim it was used. - `attached_canvas_screenshot`: An optional overview of the workflow, colors, and layout. It is not a per-image generation reference. Do not copy browser controls, connections, titles, or a multi-image canvas layout into the output. - If the user supplies four finished collages, check each image's identity and A/B/C/D order, then use the video workflow without regenerating the images. Optional inputs include season, clothing style, color palette, pose, images-only or video output, and target video length. By default, retain the supplied four outfits and target a 7-second video, subject to the live channel's supported parameters. ## Workflow ### 0. Verify available capabilities Before generating media, call `list_models` for images and video to verify the available channels and their limits. Do not assume a model name or output specification. Record the verified resolution, aspect ratio, and duration limits in the run log. ### 1. Create four 9:16 collages - Reuse an approved outfit board without regenerating it or changing its faces. - If there is no outfit board, first establish the four outfits and identity reference. Identify any missing model capability or required user choice. - For a newly created internal clothing or body reference, use a neutral gray studio background, a full-body view, and normal proportions. Cover the face with an opaque white circle and supply a separate unmasked face reference. If making a three-view sheet, include FRONT, SIDE, and BACK. Do not alter an already approved board. Final collage photos and videos must show the person's face, with no reference mask or three-view layout. - The default for new images is image2 at native 4K. Verify that the platform provides both before generating. If either capability is unverified, reuse approved collages or wait for the user to explicitly select a sup