Flagship uncensored image-to-video with native audio, 1080p and per-second narrative control.
Spicy 1.0 Pro I2V (Wan 2.7 LoRA Spicy) turns a single image into a 2-15 second cinematic video with synchronized native audio. Script every second with the At 00:XX frame-by-frame structure, steer dialogue with emotion tags, and let the built-in prompt optimizer expand short briefs into production-grade scripts. Supports photoreal, anime and Asian-face LoRA styles.
Model Type:
No output yet. Run the model to see results.
Click 'Run Model' to generate output
Exact unit prices and billing rules for this model. All prices in USD.
| Resolution | No reference video |
|---|---|
| 720p | $0.2 |
| 1080p×2 | $0.4 |
720p · 8s output → 8 × $0.2/s
≈ $1.60
Real generations from this model — hover a clip to play, copy the prompt to remix it.
Sample clip
Endpoint
https://api.spicyrouter.ai/v1/videosAll requests require a Bearer API key.
Submit Task
curl -X POST \
"https://api.spicyrouter.ai/v1/videos" \
-H "Content-Type: application/json" \
-H "Authorization: Bearer YOUR_API_KEY" \
-d '{
"model": "wan2.7-i2v-spicy",
"audio_url": "https://example.com/your-asset",
"duration": 5,
"generate_audio": true,
"image_url": "https://example.com/your-asset",
"negative_prompt": "",
"prompt": "your prompt here",
"prompt_extend": true,
"resolution": "1080p"
}'Poll Status
# Replace TASK_ID with the data.task_id from the submit response
curl -X GET "https://api.spicyrouter.ai/v1/status/TASK_ID" \
-H "Authorization: Bearer YOUR_API_KEY"Request Parameters
| Field | Required | Type | Values | Description |
|---|---|---|---|---|
audio_url | No | string | Audio Url | |
duration | No | integer | 2 ~ 15 Default: | Duration |
generate_audio | No | boolean | Default: | Generate Audio |
image_url | Yes | string | Default: | Image Url |
negative_prompt | No | string | 1 ~ 2000 chars Default: | Negative Prompt |
prompt | Yes | string | 1 ~ 4000 chars Default: | Prompt |
prompt_extend | No | boolean | Default: | Prompt Extend |
resolution | No | enum | "720p", "1080p" Default: | Resolution |
Battle-tested prompts to get production-grade output in one shot.
At 00:00, the frame is a medium close-up from a slightly low angle, capturing a nude woman with long dark hair, soft makeup, a shy smile — sitting on the edge of a bed with white sheets; warm golden window light from the left; 4K, sharp focus, shallow depth of field, photorealistic. At 00:01, she looks up through her lashes, breath hitching. Camera remains completely static. No distortion. Face stable.
Why Pro
Pro I2V is the premium tier of the Spicy lineup — longer durations, native sound, and per-second narrative control. If Lite is your traffic engine, Pro is your monetization engine.
Dialogue, breath, ambience and BGM are generated in sync with the picture — emotion tags like [breathlessly] steer delivery and lip-sync.
The At 00:XX narrative structure lets you choreograph every second: micro-expressions, motion build-up, peak and release.
Built-in prompt optimization turns a short brief into a production-grade script automatically — v2.4 cut expansion latency in half.
Dedicated LoRA styles for anime aesthetics and Asian facial features — photoreal and stylized outputs from the same endpoint.
Use Cases
Cinema-grade 1080p clips with sound are monetizable content on their own — 15 seconds is enough to tell a complete micro-story.
Branching interactive stories where each user choice triggers the next clip — one API call per branch keeps the architecture trivially simple.
Ship one-click professional video generation to your creators — prompt expansion removes the learning curve, audio output removes the editing step.
1080p + native audio is a quality moat low-cost models can't cross — let creators sell premium clips while you take 15–30% commission.
Prompt Craft
Pose, nudity level and scene in your input image must match the prompt — the image defines the starting state. Subject fills 60–70% of frame, clean background, even lighting, ≥1024×1024.
The 00:00 frame must carry shot type, appearance, pose, anatomy detail, background and lighting. Later frames describe changes only — never repeat static information.
Lighting is the single biggest quality factor. Use the formula: source + direction + temperature + shadow — e.g. soft window side light, warm golden glow.
One camera move per shot, and static is safest. Orbit shots fail >80% of the time inside 5–15s clips. Default to Camera remains completely static.
5s = micro-expressions only; 8s adds hand movement; 10s+ allows full body motion. Over-ambitious choreography causes frame tearing.
Keep dialogue to 5–10 words per line with emotion prefixes. Describe physical sound cues (heels click on marble) and ambience explicitly.
Master Template
Fill the brackets, keep the structure. 50–200 words total is the sweet spot; the 5000-character hard limit truncates anything beyond it.
At 00:00, the frame is a [SHOT TYPE] from an [ANGLE], capturing a [nude/clothed] [ethnicity] woman with [hair], [makeup], [expression] — [pose and spatial relation to partner]; [anatomical details]; [partner: torso visible, face out of frame]; [background: objects, materials, colors]; [lighting: soft window side light, warm golden glow]; [4K, sharp focus, shallow depth of field, no censorship, photorealistic]. At 00:01, [subtle action + expression shift + breathing change]. At 00:02, [motion intensifies + posture change + sheen on skin]. At 00:03, [peak action + intense expression + muscle tension]. At 00:04, [continuation or transition + final state]. Camera remains completely static throughout, no zoom, pan, or tilt. No distortion. No jitter. Face stable, no deformation.
Master Template
Do
Don’t
FAQ
Diffusion models are inherently stochastic. Generate 2–3 versions and pick the best; more specific prompts and explicit quality constraints narrow the variance.
Add Fingers maintain correct anatomy. Hands stable, no extra digits., keep gestures simple, place hands naturally, and generate multiple candidates.
Motion coherence degrades and clips start looping. Split long narratives into two 5–8s generations and stitch them — never reuse the same descriptions in the second half.
Not currently — dedicated video upscaling is not offered on the platform right now; native output tops out at 1080p.
Upload one image. Get back motion, sound and heat. Production-ready in under two minutes.