Model
FASHN.ai
Nano Banana Pro (Gemini 3 Pro)
Gemini 2.5 Flash (VTO / Mannequin)
IDM-VTON (Replicate)
Vertex AI — virtual-try-on-001 (Google)
All five — side by side
Scope
Top only
Bottom only
Full outfit / one-piece
Auto-detect (FASHN only)
Pose
Keep my pose
Adopt model's pose
Person Gender / Gemini Mode
Female — real person VTO (face + pose + background preserved)
Female — face-matched VTO (two-pass: mannequin + face transfer, 2× cost — fallback if single-pass is blocked)
Female — mannequin render (safe fallback, no person photo used)
Male — real VTO (person photo used directly)
Garment Style Profile
South Asian Luxury — kameez, lehenga, sherwani, heavy embroidery
South Asian Casual — shalwar kameez, kurta, lawn
Western Luxury — blazers, tailored suits, evening wear
Western Casual — t-shirts, jeans, streetwear
FASHN Quality Mode
Performance — ~5–7s · good for iterating
Balanced — ~8–10s · good quality
Quality — ~12–19s · best detail fidelity
Garment Photo Type
Auto-detect
Flat-lay (garment on surface)
On model (worn by a person)
Segmentation
Segmentation-free — best garment detail (hands fixed by post-process)
Human parsing ON — cleaner body boundary, slightly lower detail
Segmentation auto-switches when you change Scope.
Num Samples
1 sample (default)
2 samples — pick the best
3 samples
4 samples (max)
⚠ FASHN v1.6 removed pose-adoption as a parameter (deprecated at v1.5, now automatic). Selecting "Adopt model's pose" has no effect on FASHN output. Use Nano Banana Pro or IDM-VTON to test pose transfer.
Generate