Uncensored Z-Image to Wan 3.0 i2v: No 9:16 Aspect Ratio
Z-Image Turbo is the cheapest way to make a still on Siray, which makes it the obvious first stage of an uncensored image-to-video pipeline into wan-3.0-i2v-spicy. It also has exactly three output sizes and no aspect_ratio field, and that combination quietly rules out the one shape most short-form video needs.
Key Takeaways
- alibaba/z-image-turbo-t2i takes size as a three-value enum. The spec labels the field "Fixed size", and there is no aspect_ratio key in the request body at all.
- Those three values give you 4:3, 3:4, and 1:1 — nothing else. 16:9 and 9:16 are unreachable from a Z-Image still.
- alibaba/wan-3.0-i2v-spicy makes aspect_ratio required, choosing from adaptive, 16:9, 9:16, 1:1, 4:3, 3:4. Only three of the six line up with a Z-Image input.
- Z-Image's longest edge is 1024 px, so a 1080p Wan render starts from a source smaller than its own output. Specs verified 25 August 2026.
Three Sizes, Three Ratios
Z-Image size | Ratio | Matching aspect_ratio on either Wan i2v tier |
1024x768 | 4:3 | adaptive or 4:3 |
768x1024 | 3:4 | adaptive or 3:4 |
1024x1024 | 1:1 | adaptive or 1:1 |
That is the complete table. There is no fourth row and no parameter that adds one — the whole Z-Image request body is model, prompt, seed, and size. Its virtue is elsewhere: at $0.004 per image it is the cheapest still on the platform. See what Z-Image Turbo is built for.
Why adaptive Is Your Only Clean Choice
wan-3.0-i2v-spicy lists aspect_ratio among its required fields, alongside model, prompt, image, duration, and size. Its request body takes the same fields with the same accepted values as the standard tier — only the model string differs — so nothing here changes when you switch to Spicy: you still name a value on every call.
With a Z-Image input, adaptive is the choice that asks Wan for nothing it was not given. Naming 16:9 or 9:16 instead asks it to produce a shape your source does not have, and the spec does not state what happens then — crop, pad, and stretch are all plausible, none documented. Treat that as the genuine unknown it is: do not build a vertical pipeline on it without testing your own output.
Vertical Short-Form Is the Casualty
Run the numbers. 9:16 is 0.5625; the tallest thing Z-Image makes is 3:4, or 0.75. That is not a near miss you can crop away for free — reaching 9:16 from 768x1024 discards a quarter of the width, from a frame only 768 px wide.
The 1024 px ceiling compounds it. Wan i2v renders 1080p if you ask, but a Z-Image still has under 1080 lines either way, so the top size tier does more work than the source supports. At 480p and 720p the mismatch is academic; for 1080p vertical it is the wrong starting point twice over.
Three Ways Around It
- Resolve the ratio before the call. Pad or crop the still to your target shape yourself, then send adaptive. You control the framing instead of guessing at undocumented behaviour.
- Change the first stage. Qwen Image 3 exposes a 15-value aspect_ratio that includes 9:16 and 16:9, so a vertical still costs $0.035 instead of a workaround — that pipeline is written up here.
- Stay square on purpose. If the deliverable is a 1:1 loop or a 4:3 clip, none of this applies and Z-Image remains the cheapest route in. The same logic holds for the Z-Image to LTX pipeline.
FAQ
Can Z-Image generate 9:16 vertical images?
No. z-image-turbo-t2i accepts size as a three-value enum — 1024x768, 768x1024, 1024x1024 — which is 4:3, 3:4, and 1:1. There is no aspect_ratio field to override it.
Is aspect_ratio required on Wan 3.0 i2v-spicy?
Yes. wan-3.0-i2v-spicy requires model, prompt, image, duration, size, and aspect_ratio. The six accepted values are adaptive, 16:9, 9:16, 1:1, 4:3, and 3:4.
What resolution do Z-Image stills come out at?
1024 px on the longest edge, in all three sizes. That is below 1080p, so a 1080p Wan render starts from a smaller source than its output.
Build the Pipeline on Siray
Z-Image, Qwen Image 3, and all six Wan 3.0 endpoints run off one API key, and Wan is discounted this week. Create your free Siray account and check your target ratio before you pick the first stage.
Compliance: Siray serves legal adult creation only. The Spicy tier means fewer restrictions within the law, not none. Siray refuses all illegal content and operates a zero-tolerance policy on CSAM. All adult content must involve consenting adults aged 18 or over.