FLUX releases Virtual Try-On with sub-4-second generations
Black Forest Labs released FLUX Virtual Try-On with demo and API access, claiming sub-4-second generations across thousands of products while preserving identity and garment details. The pitch targets catalog-scale apparel imagery, so teams should try it where logos, prints, and fit consistency usually break first.
TL;DR
- Black Forest Labs shipped FLUX Virtual Try-On with a public demo and BFL API access, according to bfl_ml's launch post.
- The core claim in the launch post is sub-4-second generation time across thousands of products, aimed squarely at catalog-scale apparel work.
- BFL's announcement says the model is tuned to preserve a person's identity while keeping garment logos, stitching, and prints intact.
- a follow-up demo clip shows the system handling multiple clothing pieces in a single try-on run.
You can jump straight into the demo, read BFL's engineering write-up, and watch the launch video focus on the hard part most try-on systems fumble, namely keeping faces stable while branded garment details survive the swap.
Demo and API access
Black Forest Labs launched FLUX Virtual Try-On as both a browser demo and an API product. The announcement links directly to the hosted experience, and says the model is already available through the BFL API.
That makes this a production tool story, not just a research teaser. The two day-one entry points are simple:
- Demo: public web interface for testing outputs
- BFL API: listed as available in the launch post
- Official blog post: BFL's write-up from its FDE team
Identity and garment details
The product pitch is unusually specific about what BFL optimized for. In the main announcement, the company says the system keeps the person's identity consistent while preserving logos, stitching, and prints on the garment.
Those are the exact details that usually break first in retail composites. BFL also frames the system around throughput, with the launch post claiming sub-4-second generations across thousands of products at low cost and low latency.
Multiple garments
A second BFL clip adds one more concrete capability: the model can try on multiple pieces at once. That demo post is brief, but it expands the launch from single-item swaps to more complete outfit changes.
For creative teams building apparel mockups, that is the most useful extra reveal in the thread because it points to styled looks, not just isolated tops or jackets.