Prompting Qwen Image 3 Spicy for Fantasy Character Portraits
alibaba/qwen-image-3-t2i-spicy generates NSFW fantasy character portraits from text at $0.035 per image. This article covers original, fictional character designs only — elves, demons, sorceresses, and similar archetypes — for legal adult (18+) content. Siray applies zero tolerance to any depiction of minors, including stylized or fictional framing, with no exceptions.
The endpoint has no negative_prompt field. That changes how a portrait prompt has to be built: instead of excluding unwanted terms, every detail the image needs has to be stated positively, element by element, in the prompt itself.
Why There's No Negative Prompt to Lean On
qwen-image-3-t2i-spicy ships without negative_prompt and without safety_tolerance. Both fields are absent from the request schema — not disabled, not hidden behind a flag, simply not part of this endpoint. The same is true across all six Qwen Image 3 endpoints, spicy and non-spicy alike.
The one assist available is prompt_expansion_enable, which defaults to true and lets the model fill in reasonable detail around a sparse prompt. It's not a substitute for structure — it smooths gaps in an already-organized prompt, it doesn't organize one for you. A vague single-sentence prompt with expansion turned on still produces an inconsistent character. A five-element prompt does not need expansion to hold together, but leaving it on rarely hurts.
Five Prompt Elements, One at a Time
Each element controls one part of the portrait. Write all five, in order, and the model has nothing left to guess at.
- Character archetype. Name the species or role before anything else — elf ranger, succubus, battle sorceress. This is the identity anchor everything else attaches to. Example: a horned demon sorceress.
- Pose and action. State a specific body position, not a mood word. "Sexy pose" gives the model nothing to render; "kneeling, one hand braced on a stone altar, head turned toward the viewer" gives it a pose. Example: kneeling on a cracked stone altar, torso turned toward camera, one hand resting on her own thigh.
- Wardrobe and explicitness. The core lever on this endpoint. Since there's no negative prompt to hold explicitness back, the wardrobe line has to state exactly the level wanted — from partial armor, to sheer fabric, to full nudity — rather than leaving it to inference. Example: cracked obsidian armor plates falling open at the chest, fully nude beneath.
- Lighting and mood. Rim light, candlelight, and moonlight each pull a portrait toward a different genre feel. Example: low red rim light from behind, dim candlelight across the front of the body.
- Art-style anchor. Naming a reference style keeps repeated generations consistent instead of drifting between painterly and photoreal. Example: dark fantasy oil-painting style, heavy shadow, painterly brushwork.
Combined, these five lines become one prompt: "A horned demon sorceress, kneeling on a cracked stone altar, torso turned toward camera, one hand resting on her own thigh, cracked obsidian armor plates falling open at the chest, fully nude beneath, low red rim light from behind, dim candlelight across the front of the body, dark fantasy oil-painting style, heavy shadow, painterly brushwork."
Batching Without Surprise Costs
n controls how many images come back from one request, from 1 to 6, default 1. Price scales linearly with n — it is not a flat per-request fee. A single image is $0.035; a batch of six is $0.21, six times the unit price, not a discounted bundle.
size and aspect_ratio are both required fields with no default value. size accepts only 1k or 2k — there's no 4k option on this endpoint. aspect_ratio accepts 15 values (1:1, 1:2, 2:1, 1:3, 3:1, 2:3, 3:2, 3:4, 4:3, 4:5, 5:4, 9:16, 16:9, 9:21, 21:9). Omitting either field causes the request to fail; there's no fallback default to catch a missing value.
Get Access
A single Siray API key covers qwen-image-3-t2i-spicy and every other model on the platform. Requests run through the async pattern: POST /v1/images/generations/async, then poll GET .../{task_id} until the result is ready.
Readers who need to feed a reference image into a portrait rather than generating one from text alone want the Qwen Image 3 edit endpoints instead — a separate endpoint family built for image-in, image-out editing.
FAQ
Q: Does Qwen Image 3 Spicy support negative prompts?
A: No. negative_prompt is absent from all six Qwen Image 3 endpoints, including qwen-image-3-t2i-spicy. Prompt structure — stating what should appear, element by element — carries the full weight instead.
Q: How much does a batch of images cost?
A: $0.035 per image, scaling linearly with n (1-6). A batch of six costs $0.21, not the single-image price.
Q: Are size and aspect_ratio optional?
A: No. Both are required with no default. size accepts only 1k or 2k; aspect_ratio accepts 15 fixed values. A request missing either field fails.
Summary
With no negative_prompt field, qwen-image-3-t2i-spicy puts the full weight of a consistent fantasy character portrait on the prompt itself. Five elements — archetype, pose, wardrobe/explicitness, lighting, and art style — stated in order, cover what a negative prompt would otherwise exclude by naming what should be there instead.
Create your free Siray account and start generating with Qwen Image 3 Spicy.
Siray enforces zero tolerance for content depicting minors in any form, including stylized or fictional depictions. All NSFW generation requires user attestation of adult status and consent, and is subject to ongoing moderation review.