仕様書

Wan 2.7 NSFW Image

High-fidelity NSFW image generation with multi-reference support.

Wan 2.7 is an image model, not a video model — it generates high-resolution NSFW stills. It introduces multi-reference guidance (up to 9 reference images), a "Thinking" mode for complex prompt adherence, and 2K/4K output (4K on Pro tier). Use it to produce reference images for I2V workflows or as a standalone image generator.

Technical Specifications

Output

Resolution	2K (standard) · 4K (Pro)4K requires Pro tier
Format	PNG / JPG
Batch	Multi-image pack generation

Generation modes

Text-to-image (T2I)	✓
Reference-guided generation	✓Up to 9 reference images
Text-to-video (T2V)	✗
Image-to-video (I2V)	✗

Quality traits

Prompt adherence	High"Thinking" mode for complex prompts
Identity consistency	HighMulti-reference anchor
Anatomy accuracy	Improved vs earlier models

Access

Free tier	✓2K output; 4K requires Pro
No download / install	✓

Strengths

Up to 9 reference images for precise identity anchoring — best in the Wan lineup for face/body consistency.
"Thinking" mode significantly improves adherence to complex or multi-subject prompts.
2K native output is sharp enough for most display and print uses.
Batch generation produces consistent sets in one run.

Limitations

Image only — does not generate video. Pair with Wan 2.6 I2V for motion.
4K output requires a Pro subscription.
Slower than basic T2I models due to Thinking mode overhead.

Best For

Generating reference images to feed into Wan 2.6 I2V.
High-quality standalone NSFW illustration or art.
Multi-character scenes requiring per-character identity control.

← 全モデルの仕様モデルを使う →