Wan 3.0 Spicy Aspect Ratios: Pick the Right Frame Fast

Key takeaway: alibaba/wan-3.0-{t2v,i2v,ref2v}-spicy ships six aspect_ratio values — adaptive, 16:9, 9:16, 1:1, 4:3, 3:4. Picking the wrong one is the fastest way to lose reach on the platform a clip is actually posted to, since the parameter sets the canvas but the prompt's composition language still has to match it.

Most creators leave aspect_ratio on adaptive and never touch it again. That works for a first pass, but every clip gets cropped, letterboxed, or algorithmically deprioritized the moment it lands somewhere with a fixed frame. Wan 3.0 Spicy makes the ratio an explicit, required parameter, so the framing decision gets made once, in the prompt, instead of fixed later in an editor.

Quick facts: Wan 3.0 Spicy

• Three live endpoints: wan-3.0-t2v-spicy (text-to-video), wan-3.0-i2v-spicy (image-to-video), wan-3.0-ref2v-spicy (reference-to-video)
• size: 480p / 720p / 1080p; duration: 2-30 whole seconds; audio_enable: optional boolean
• prompt_expansion_enable defaults to true — Wan rewrites the submitted prompt before generation unless set to false. This can quietly override careful framing language, so prompts built around a specific aspect_ratio should disable expansion or verify the expanded prompt first.
• No negative_prompt field on any of the three Spicy endpoints, unlike other Wan generations. Exclusions have to be written into the positive prompt itself ("clean single subject, no background clutter" rather than a negative).

The six aspect ratios, one at a time

9:16 for TikTok, Reels, and Shorts

9:16 is the default for TikTok, Reels, and Shorts. Push the subject toward center-frame with headroom at the top and footroom at the bottom, since vertical crops punish anything positioned for a horizontal frame.

Prompt: vertical portrait framing, subject centered, close-up to medium shot, shallow depth of field, direct eye contact with camera, soft studio lighting, aspect_ratio: 9:16

16:9 for widescreen players and YouTube-style embeds

16:9 fits widescreen players and YouTube-style embeds, giving room for environment and secondary action — useful when a clip sits on a landing page rather than a feed.

Prompt: wide establishing shot, two subjects with visible negative space on both sides, environmental detail in background, cinematic lighting, aspect_ratio: 16:9

1:1 for the Instagram grid and Threads

1:1 fits the Instagram grid and Threads. Square framing forces a tighter center of interest, so describe the subject as filling most of the frame rather than leaving open space at the edges.

Prompt: square framing, subject fills 80% of frame, tight medium shot, symmetrical composition, aspect_ratio: 1:1

4:3 and 3:4 for retro and archival framing

4:3 and 3:4 are the least-used tiers, fitting a specific niche: a boxier, "shot on an older camera" feel for retro content, or a taller-than-square crop for platforms that reject full 9:16. Lean into the retro cue directly — film grain, boxier framing — rather than treating these as fallback ratios.

Prompt: 4:3 boxy framing, retro film grain, warm vintage color grade, medium shot, aspect_ratio: 4:3

adaptive for i2v and ref2v jobs with a strong source image

adaptive fits i2v and ref2v jobs where the source image's own proportions should drive the output rather than forcing a mismatch crop. For t2v jobs with a specific platform already in mind, an explicit ratio is more reliable.

Full example: TikTok-ready clip

Prompt: vertical portrait framing, subject centered with headroom, medium close-up, direct camera engagement, soft rim lighting, subtle motion, size: 1080p, duration: 8, aspect_ratio: 9:16, prompt_expansion_enable: false

This combines the ratio, the composition language that matches it, an explicit duration, and a disabled prompt expansion so the framing instructions aren't silently rewritten.

Developer notes

aspect_ratio is a required string field, not a prompt keyword — pass it alongside prompt, size, and duration. ref2v-spicy bills differently: billable duration is input video duration (capped at 5s) plus output duration, so a longer reference clip doesn't scale cost like a longer output does.

On pricing: Siray's console currently shows a "Wan 3.0 Launch Month" promotion on the Spicy tier. The badge says "up to 33% off," but the actual per-second billing on the model-detail page is more precise: 480p $0.045/s, 720p $0.09/s, 1080p $0.18/s — a uniform ~31% cut across all three tiers, available through Siray with one API key across all three Spicy endpoints.

This piece covers aspect_ratio specifically. For Launch Month pricing mechanics across the full Wan 3.0 lineup, see Wan 3.0 Launch Week: Up to 25% Off Video Rates. For a migration decision between Spicy and the previous Wan release, see Wan 3.0 Spicy vs Wan 2.7 i2v Spicy. For a cross-model cost comparison on micro-drama workloads, see Micro-Drama Video: Wan 3.0 vs Seedance.

Summary

The aspect_ratio field is the cheapest lever for matching output to a platform — free to set correctly, costly to get wrong when the algorithm is already deprioritizing mismatched framing. Pick the ratio for where the clip is going, match the composition language to it, and disable prompt expansion when the framing has to hold exactly as written.

Create your free Siray account and start generating with Wan 3.0 Spicy across all three endpoints, one API key, no per-model integration work.

CSAM zero tolerance. Siray refuses all illegal content without exception. "Uncensored" describes the absence of over-censorship for legal adult (18+) content — it is not permission for anything unlawful. Every generation must depict consenting adults only.