Wan 2.7 i2v Spicy: Uncensored 1080p Video From a Still Image

Key Takeawaysalibaba/wan-2.7-i2v-spicy is live on Siray. Input a still image; get back uncensored video at up to 1080p.This endpoint is i2v only — there is no t2v or ref2v variant in this family.prompt_expansion_enable defaults to true, meaning the model will rewrite your prompt before generation. Turn it off if you need exact prompt adherence.Billing rates have not been published; do not estimate from other endpoints.

Wan 2.7 i2v Spicy is available on Siray. The endpoint takes a still image as input and animates it into uncensored video at 720p or 1080p. Here is what the spec looks like and how to get the most out of it.

Endpoint Spec

Note on the i2v-only structure: This endpoint has no t2v or ref2v sibling. It is the only Wan 2.7 NSFW endpoint currently available. If your workflow needs text-to-video in the uncensored video pool, both Seedance 2.5 Spicy and Wan 3.0 Spicy cover that path with t2v, i2v, and ref2v variants.

The Parameter You Must Know: prompt_expansion_enable

This parameter defaults to true. When it is on, the model rewrites your prompt before generation — expanding, rephrasing, or adding detail the model judges as helpful. The output video reflects the rewritten prompt, not your original text exactly.

For most use cases this produces better results. For use cases where exact prompt adherence matters — a specific pose, a precise setting, a controlled scene element — set it to false:

{  
"model": "alibaba/wan-2.7-i2v-spicy",  
"image": "<base64 or URL>",  
"prompt": "...",  
"prompt_expansion_enable": false,  
"size": "1080p",  
"duration": 8
}

Using negative_prompt

The negative_prompt field lets you exclude elements from the output. Standard exclusion patterns apply — describe what you do not want to see in the same descriptive language you use for positive prompts. This is particularly useful for controlling background elements, unwanted objects, or motion artifacts.

Image Input

The source image drives the output significantly. A few things that consistently improve results:

• Clear subject framing. The model animates what is in the image; if the subject is partially cropped or obscured, the animation will inherit that ambiguity.
• Aspect ratio alignment. Feeding an image with a 16:9 composition into a 16:9 output size gives the model less to interpolate. Mismatched aspect ratios can cause cropping or stretching in the animated output.
• Minimal background complexity for character focus. If the goal is to animate a character, simpler backgrounds give the model fewer competing motion signals.

Where This Fits in the Uncensored Video Pool

The i2v-only structure is the key constraint. Use this endpoint when the workflow starts from an image — an AI-generated still, a reference photo, or a compositional sketch — and needs to be animated. For pure text-driven uncensored generation, Seedance 2.5 Spicy and Wan 3.0 Spicy both offer t2v endpoints. For a wider view of what is available, see the NSFW AI video API roundup.

Pricing

Billing rates for wan-2.7-i2v-spicy have not been published. Run a short test clip before scaling to calibrate your actual cost.

Create your free Siray account and start animating with Wan 2.7 i2v Spicy now.

Siray provides access to uncensored AI generation models strictly for legal adult creative use. All content must comply with applicable laws. Siray enforces a zero-tolerance policy on CSAM and any content involving minors. Uncensored access means no excessive content filtering beyond legal requirements — it does not mean illegal content is permitted.