WAN 2.2 Spicy Image-to-Video
Animate a photo with WAN 2.2 Spicy: expressive motion, 5–8s clips at 480p/720p, 20–40 credits/s. See how Turbo Spicy and “unlimited” offers compare.
WAN 2.2 Spicy is an image-to-video model built on Alibaba's WAN 2.2 mixture-of-experts architecture and tuned for more expressive motion than the standard WAN 2.2 image-to-video checkpoint. Upload one image, describe the movement, and the model returns a 5- or 8-second MP4 at 480p or 720p. The Spicy tuning pushes motion amplitude, lighting contrast and color grading, so shots feel more dramatic and less restrained. It suits cinematic reveals, stylized characters, fantasy scenes and social clips. On Conut.ai it runs in the same workspace as the rest of the WAN family, at 20 credits per second for 480p and 40 credits per second for 720p.
- Resolution: 480p / 720p
- Duration: 5s / 8s
- Price: 20–40 credits/s
- Audio: None
How to use WAN 2.2 Spicy Image-to-Video
Spicy is the expressive WAN 2.2 variant. Pick a strong subject, describe one clear motion, and keep the clip short.
- Pick an image with one obvious subject Portraits, products and single characters give the model a clear subject to move. Busy scenes turn into motion soup once the Spicy tuning pushes amplitude.
- Write a motion-first prompt Lead with what moves and how: hair lifting, camera pushing in, fabric rippling. Mood words help, but concrete motion is what the checkpoint responds to.
- Choose 5 or 8 seconds Five seconds is the default and the safer length for exaggerated motion. Go to 8 only when the action needs a beat longer than a simple gesture.
- Pick the resolution by budget A 5-second 480p clip costs 100 credits and the same clip at 720p costs 200. Draft at 480p when you are still testing the prompt.
- Lock a seed once the motion works When a run lands, copy its seed and rerun with small prompt changes. That keeps the motion path while you tune the look.
Settings reference
WAN 2.2 Spicy keeps the control set small and puts the tuning into the checkpoint: one image, one motion prompt, two lengths and two resolutions.
| Setting | Value | Notes |
|---|---|---|
| Input image | 1 image, required | A clear, well-lit subject works best; high-contrast or stylized photos amplify the effect. |
| Prompt | Required | Describe motion, mood and camera movement. The model leans expressive, so keep the action legible. |
| Duration | 5s / 8s | Default is 5 seconds. Short clips keep exaggerated motion coherent. |
| Resolution | 480p / 720p | 20 credits per second at 480p, 40 at 720p. 720p is the default. |
| Seed | Optional | Reuse a seed to reproduce the motion path of a run you liked. |
| Negative prompt | Not available | This checkpoint exposes no negative prompt; state exclusions in the positive prompt instead. |
| Last frame | Not available | Unlike standard WAN 2.2 Image-to-Video, there is no last-frame control. |
What “Spicy” changes compared with standard WAN 2.2
Standard WAN 2.2 image-to-video is tuned for stability: the subject stays on model, the camera move is gentle, and the result looks like a controlled cinematic shot. WAN 2.2 Spicy keeps the same architecture but shifts the output toward larger movement and stronger tonal contrast. In practice, the same 5-second clip usually contains more subject motion, deeper shadows, more saturated color and a more willing response to dramatic camera directions.
The trade-off is that extra motion also gives the model more chances to warp. Faces, hands, text and fine repeating patterns are the first places where Spicy can slip. If you need a clean product turntable or a subtle expression change, standard WAN 2.2 is the safer choice. If the standard model looks flat and you want the shot to feel alive, try Spicy first.
- Larger subject movement in the same clip length
- Stronger light-and-shadow contrast and richer color
- More willing to follow dolly, orbit, tilt and tracking directions
- Better fit for fantasy, anime, dance and stylized fashion shots
- Less conservative on motion, which also means less margin for error
WAN 2.2 Spicy vs WAN 2.2, WAN 2.5 and WAN 2.6
The WAN family overlaps, so pick the model by what the shot needs rather than by version number. The table uses the settings exposed on Conut.ai.
- Choose Spicy when the shot needs energy and the source image is clean.
- Choose WAN 2.2 I2V when you need a negative prompt or a defined last frame.
- Choose WAN 2.5 when the subject speaks or needs synchronized audio.
- Choose WAN 2.6 when the clip must run longer than 8 seconds or reach 1080p.
| Model | Resolution | Duration | Credits | Audio | Best for |
|---|---|---|---|---|---|
| WAN 2.2 Spicy | 480p / 720p | 5s, 8s | 20–40 /s | No | Expressive motion, stylized shots |
| WAN 2.2 Image-to-Video | 480p / 720p | 5s, 8s | 30–60 /s | No | Stable cinematic motion, negative prompt, last-frame control |
| WAN 2.5 Image-to-Video | 480p / 720p / 1080p | 5s, 10s | 20–40 /s | Yes, one-pass A/V sync | Talking shots, lip-sync, multilingual dialogue |
| WAN 2.6 Image-to-Video | 720p / 1080p | 5s, 10s, 15s | 30–50 /s | Yes | Longer cinematic clips at higher resolution |
Turbo Spicy, Unlimited, and WAN 3.0 Spicy: what the names mean
“WAN 2.2 Turbo Spicy” usually refers to a turbo-distilled build hosted by a different provider. Turbo variants cut the number of inference steps to run faster and cheaper, but they can give up fine detail and some of the motion nuance of the full checkpoint. The endpoint on this page is wavespeed-ai/wan-2.2-spicy/image-to-video, the standard Spicy model, not a turbo-distilled build. When you compare outputs across sites, compare the model ID rather than the marketing label.
“WAN 2.2 Spicy unlimited” is a hosting phrase, not a separate model file. It normally means an uncapped generation quota or a relaxed content policy on a particular site. Conut.ai does not advertise an unlimited plan: generation is metered in credits, with 100 credits for a 5-second 480p clip and 320 credits for an 8-second 720p clip. Every job on this page uses the same per-second pricing.
There is no WAN 3.0 Spicy model listed on Conut.ai. If you saw the term in a search result, the newest WAN image-to-video model here is WAN 2.6, which supports 1080p and clips up to 15 seconds.
How to prompt WAN 2.2 Spicy for cleaner motion
Spicy responds best to a prompt that separates what moves from how the camera moves. A practical formula is: subject + action + camera + light and atmosphere + style. Keep one dominant action per 5-second clip. A turn, a step, a wingbeat or a head lift is enough; stacking three or four actions usually makes the model rush the last one.
- Put the camera move in its own clause: slow dolly-in, 30-degree orbit, low tracking shot, tilt up.
- Name the light: backlit, rim light, volumetric fog, sunset side light, neon reflections.
- Keep the source image simple: one subject, clear silhouette, no text and no busy collage.
- Test motion and seed at 480p, then re-run the winning prompt at 720p.
- Reuse the seed when you want to compare prompt changes without changing the starting noise.
- Slow down anything that must stay readable: hands, faces, logos and small props.
Limitations and when to use a different model
WAN 2.2 Spicy is deliberately a short-clip model. It tops out at 720p, produces 5 or 8 seconds per job, and does not generate audio. This endpoint also does not expose a negative prompt or last-frame control. Those are not bugs; they are the boundaries that keep the model focused on expressive motion.
- Need 1080p? Use WAN 2.5 or WAN 2.6 image-to-video.
- Need a clip longer than 8 seconds? Use WAN 2.6 (up to 15s) or chain clips in an editor.
- Need dialogue or synchronized sound? Use WAN 2.5 or WAN 2.6.
- Need a negative prompt or a defined end frame? Use standard WAN 2.2 image-to-video.
- Fast lateral motion, hands, faces and text are the usual failure points. Keep them small in frame or slow them down.
- The source image sets the quality ceiling. A soft or cluttered image stays soft no matter how detailed the prompt is.
- Acceptable-use rules still apply. “Spicy” describes motion and style tuning; it is not a promise that every kind of content is allowed.
Prompt examples for WAN 2.2 Spicy Image-to-Video
- Ice dragon glide (the example on this page): Animate the ice dragon gliding forward through the storm, snowflakes streaming past its wings, clouds swirling around the mountain peaks, camera slowly orbiting the dragon's flight path, cold blue rim light, cinematic fantasy realism. The example output above uses this prompt. One subject, one action, one camera move.
- Neon portrait turn: A woman in a red coat turns slowly toward the camera, wind lifting her hair, neon signs sliding across her face, slow dolly-in, teal and amber cinematic grade, shallow depth of field. A small action with a clear camera move is the safest way to keep a face stable.
- Product orbit: A glass perfume bottle on wet black stone, camera orbits 30 degrees around the bottle, water ripples spread from the base, soft studio highlights, luxury commercial lighting, crisp reflections. Rigid objects and controlled lighting hide the model's weakest areas.
- Storm landscape: Storm clouds rolling over a snow-covered mountain ridge, sunlight breaking through in shifting beams, slow aerial push-in, dramatic contrast, timelapse feel, cold desaturated palette. Landscapes tolerate large motion, so Spicy's extra energy has room to work.
- Anime rooftop: Anime heroine standing on a rooftop, coat flapping in the wind, petals swirling past, camera tilts up as the sun sets behind her, hand-painted background, warm rim light, film still. Stylized art hides small warping that would be obvious in live-action footage.
- Rain dancer: A dancer spins in a rain-soaked street, water droplets caught mid-air, slow motion, backlit by a single streetlamp, reflections stretching across the asphalt, high contrast, music-video mood. Backlighting and rain add motion cues without asking the model to animate many objects.
Pro tips for WAN 2.2 Spicy Image-to-Video
Spicy trades control for energy. Lean into that instead of fighting it.
- Expect more motion than you asked for The tuning amplifies movement, so prompts written for standard WAN 2.2 tend to come back busier. Reduce the number of actions rather than adding qualifiers.
- 720p is the practical default 480p is fine for prompt tests, but the expressive motion shows compression artefacts first. Deliver at 720p unless the clip is a background element.
- Keep social clips at 5 seconds A tight 5-second loop reads better than an 8-second clip with a weak ending, and it halves the cost of iterating.
- Switch to standard WAN 2.2 for control If you need a negative prompt or a specific last frame, use WAN 2.2 Image-to-Video. Spicy cannot do either.
- Do not fix artefacts with a longer prompt Extra clauses usually add new motion rather than remove the old. Rerun with a different seed or a cleaner source image instead.
Frequently asked questions
What is WAN 2.2 Spicy?
WAN 2.2 Spicy is a variant of Alibaba's WAN 2.2 image-to-video model tuned for larger motion, stronger contrast and more dramatic color than the standard checkpoint. It turns one image into a 5- or 8-second clip at 480p or 720p.
Is WAN 2.2 Spicy free or unlimited on Conut.ai?
New accounts get free credits to try it. After that, generation is metered: 20 credits per second at 480p and 40 credits per second at 720p, so a 5-second 720p clip costs 200 credits. Conut.ai does not offer an unlimited plan.
What is the difference between WAN 2.2 Spicy and WAN 2.2 Turbo Spicy?
Turbo Spicy is usually a distilled build from another provider that runs in fewer steps for speed. This page runs the standard wavespeed-ai/wan-2.2-spicy/image-to-video endpoint, not a turbo-distilled checkpoint, so compare model IDs before comparing outputs.
Does WAN 2.2 Spicy support 1080p or audio?
No. This endpoint supports 480p and 720p and produces silent video. For 1080p, dialogue or sound, use WAN 2.5 or WAN 2.6 image-to-video.
How long can a WAN 2.2 Spicy clip be?
Each job produces 5 or 8 seconds. For longer shots, generate multiple clips with the same seed and source image, then join them in an editor, or switch to WAN 2.6, which supports up to 15 seconds.
Can I load a custom LoRA with this model?
Not on this endpoint. Conut.ai exposes the standard Spicy image-to-video model without custom LoRA loading. The LoRA variant is a separate model file hosted elsewhere.
Is there a WAN 3.0 Spicy model?
No WAN 3.0 Spicy model is listed on Conut.ai. The newest WAN image-to-video model here is WAN 2.6, which adds 1080p, longer durations and sound.
What images work best with WAN 2.2 Spicy?
Use a sharp image with one clear subject and a simple background. Avoid text, logos, busy collages and heavy motion blur in the source. A clean, well-lit image with a clear silhouette gives the model the most to work with.
Compare with other image-to-video models
- WAN 2.2 Image-to-Video
- WAN 2.5 Image-to-Video
- WAN 2.6 Image-to-Video
- Kling Image-to-Video
- Sora 2 Image-to-Video
- Browse all AI video models
Specifications
- Provider: Wavespeed
- Model type: Image to video
- Credit pricing: 480p: 20 credits · 720p: 40 credits
- Capabilities: popular, cinematic, realistic, multi-shot
About Conut.ai
Conut.ai is an AI image, video and audio generation platform with 50+ creative tools. It supports text-to-image, image-to-video, face swap, motion control, relighting, inpainting and video effects through models including Nano Banana Pro, Seedream, GPT Image, Sora 2, Kling, WAN and Seedance. Start free — no credit card required.