FASHN Try-On v1.5 Available on fal

Team fal

• 4 min read
FASHN Try-On v1.5 Available on fal

We're excited to announce that FASHN's Try-On v1.5 model is now available on the fal platform. To ensure a smooth transition, we plan to run both v1.5 and the current v1.0 side-by-side for a month before retiring v1.0, giving you time to adapt. This represents a significant upgrade to the existing FASHN Try-On v1.0 capabilities you're already using, offering substantial improvements in speed, quality, and ease of use. Get ready to elevate the virtual try-on experiences in your applications! 

What’s New in FASHN v1.5

FASHN v1.5 introduces a new state-of-the-art approach to virtual clothing fitting, designed to deliver more realistic results with less effort. A key innovation in v1.5 is its Segmentation-Free processing. Unlike traditional methods that require a separate step to digitally segment (cut out) the original clothing, v1.5's AI intelligently processes the model image and the new garment image in a single step. Key improvements include:

  • Better fit for long and loose clothing (e.g., tunics, layered tops)
  • Improved support for bulky outerwear (e.g., coats, jackets)
  • Easier handling of garments that cover the feet or have a wide bottom (e.g., mermaid dresses, long skirts)
  • More accurate results when hair or accessories partially cover clothing
  • Preserving head coverings and other cultural garments (e.g., hijabs) during swaps
  • Better support for layered outfits where only some pieces are being changed

Key Benefits for Your Applications

FASHN v1.5 brings several advantages directly impacting your development workflow and end-user experience:

Superior Fit Generation:

  • Complex Garments: Dramatically improved handling for baggy jeans, oversized t-shirts, puffer jackets, and voluminous dresses or gowns. The model accurately captures folds and volume.
  • Long & Flowing Items: Excels with tunics, layered tops, mermaid dresses, and long skirts, including those covering the feet.
  • Occlusions & Details: Better preserves fine details like hair strands over clothing and maintains the integrity of head coverings (e.g., hijabs). It also shows promise for partial outfit swaps (e.g., swapping only a jacket).

Enhanced Identity Preservation: Better accuracy in maintaining the model's body shape, skin texture, and unique characteristics like tattoos, leading to more relatable and realistic try-ons.

Simplified API Interaction (via the Model): The v1.5 model automatically handles many parameters previously requiring manual input (like Long Top, Adjust Hands, Cover Feet, background/clothing restoration). The AI is smarter at identifying photo type, category, and what needs removal/placement, potentially simplifying your API calls. Often, just the model image and garment image are sufficient.

Massive Performance Boost: Inference time is cut in half! This means faster responses for your users and potentially lower compute costs.

  • Quality Mode: ~12 seconds
  • Balanced Mode: ~6 seconds
  • Performance Mode: ~4 seconds

How to Access FASHN v1.5 on fal

  • Model Gallery on fal: Please refer to our platform's documentation for the specific model ID or tag required to call this v1.5 version.

Understanding the Trade-Offs

The Segmentation-Free approach excels at adding garments realistically. However, it has a limitation in completely removing the original garment in all cases, and remnants of the underlying clothing might sometimes be visible.

Developer Tip: FASHN recommends starting with model images where the original clothing is less voluminous than the target try-on garment for best results.

Perspective: While perfect garment removal is still being refined, v1.5’s strength lies in accurately fitting the new garment (the harder challenge). Minor artifact cleanup, if needed, is often simpler than correcting a poor fit.

Additional Flexibility: For use cases explicitly requiring removal of bulky garments, it is still possible to enjoy the benefits of v1.5. By passing segmentation_free: false in the API call, the try-on will revert to the previous segmentation-based mechanism while utilizing the stronger and faster v1.5 base model.

What's Next

We’re thrilled to offer this powerful upgrade on fal, enabling you to build even more compelling and realistic virtual try-on applications. In just a couple of weeks, FASHN will seamlessly deploy an output resolution upgrade to the same endpoint. This update will be fully backward-compatible, boosting resolution to the 1-megapixel range without any increase in runtime. Also, a gentle reminder: in about a month, the fashn/tryon endpoint will be retired. Since the new v1.5 model surpasses its predecessor in every aspect, there’s every reason to migrate now. Start experimenting with FASHN v1.5 on fal today!