Untitled Workflow
Asynchronous AI processing pipeline designed for high-performance automation.
This workflow has no configurable inputs. Check the visual builder to add task inputs.
Overview
About this workflow
This workflow is an asynchronous AI processing pipeline. Submit your inputs and receive results when the pipeline completes.
Quick Integration
Use via REST API
Submit inputs to POST /workflow/virtual-try-on-seedream-flux-kling-o1/api-execute and poll GET /workflow/run/{run_id}/status until the run completes. Switch to the for full code examples.
About Virtual Try On with Seedream v1.5 + Flux 2 Dev + Flux 2 Flex + Flux 2 Pro + Qwen Image + Kling o1
This is the most comprehensive virtual try-on workflow on Muapi, using six models in sequence to achieve accurate garment placement on a model figure. Seedream v4.5 generates the base garment image, multiple Flux variants (Dev, Flex, Pro) progressively refine the fit and fabric detail, Qwen Image handles fine-grained face and body adjustments, and Kling o1 delivers the final edited composite.
The multi-model approach produces noticeably better garment edge preservation, fabric drape, and body proportion accuracy than single-model try-on methods. It's designed for fashion brands that need consistent, high-quality output across an entire product catalogue.
Models Used
How It Works
- 1
Garment image generation
Seedream v4.5 generates or edits the base garment image with accurate colour and texture.
- 2
Progressive fit refinement
Flux 2 Dev and Flux 2 Flex iteratively refine the garment shape, drape, and fit to match a realistic body outline.
- 3
High-fidelity detail pass
Flux 2 Pro and Qwen Image Plus sharpen fabric details, adjust skin tone accuracy, and clean up edge artefacts.
- 4
Final composite
Kling o1 produces the final edited composite with precise garment-to-body alignment.
Use Cases
- Display clothing on model figures without photoshoots
- Generate size-variant try-on images for product pages
- Produce lookbook content for seasonal collections
- Show garments across diverse model types
- Reduce returns by giving shoppers accurate fit previews
Who Is This For
Frequently Asked Questions
- Does this work for all garment types?
- Yes — tops, bottoms, dresses, outerwear, and accessories all work. Structured garments (blazers, trousers) produce the sharpest results.
- Can I use my own model image?
- Yes. Provide a reference model image as input, and the workflow will place the garment on that figure.
- How accurate is the garment fit?
- The multi-model pipeline significantly reduces common artefacts like garment distortion and unnatural draping compared to single-model approaches.
- How long does the full pipeline take?
- Approximately 60–120 seconds end-to-end, depending on image resolution and pipeline stage load.