Wan Preview 2.5 Text to Image: lock this still before you buy motion (4K, 5cr)

Wan Preview 2.5 Text to Image is the stills row. Animate after approval. Do not generate the pack inside the video model.

Wan Preview 2.5 Text to Image is Wan 2.5 in still-image mode — the video family's frame generator used standalone. It produces cinematic, video-native compositions that match Wan footage, for first-frame and thumbnail work. Category is text-to-image. Five credits. 4K. audio is null. No duration. requires_image is false.

Five credits for a frame the video row will recognise

This is not a generic poster model that happens to sit next to Wan. The catalog description is specific: video-native compositions that match Wan footage. If the next credit is Wan V2.6 Text to Video or another Wan motion row, this is the still you want to argue about — not a still from a different family that Wan will restage.

First-frame work is the point. Thumbnail work is the point. A pack of twelve product angles generated inside a 15-second video prompt is how you pay motion rates to discover the label is wrong. Make the frame here. Approve it. Then animate.

The text to image tool is the door. Wan's provider roster is the rest of the family. Best Wan model ranks the video siblings. This page is only the stills slug: wan-video-2-5-t2i, 4K, 5 credits.

Realistic or Artistic, then stop

Styles on the row are Realistic and Artistic. Two. Pick one. Do not also write a third school into the prompt and leave the field on default. Aspect is 1:1, 4:3, 3:4, 16:9, 9:16 — five shapes, no 21:9, no 8:1. If the cut is ultra-wide, this is the wrong stills row or you are cropping on purpose.

max_output_resolution is 4K. Generate the first frame at the resolution the motion pass will actually use. A 4K still you immediately squash into a 720p animate is wasted; a soft still you "fix in video" is how identity smears.

There is no audio because there is no time. Do not prompt foley. Do not prompt a voiceover. Those sentences belong on the video row, written in, because Wan's motion models can emit native audio. This row cannot.

Match Wan footage, then freeze it

The reason to be here instead of any other 4K stills model is continuity with Wan video. Same composition language, same cinematic framing, a thumbnail that looks like a paused clip from the family you will actually generate. If you are not going to Wan for motion, you do not need this sibling for its family resemblance — pick the stills row the job needs (type, mood, flash iteration) and stop cargo-culting the Wan name.

Lock the still before you spend on motion is the production rule. Register approved frames with reusable characters and products if they have to survive more than one shot. Then open image-to-video or Wan text-to-video. Do not generate the pack inside the video model.

Five credits. 4K. Text-to-image. Video-native. That is the whole machine.

FAQ

Is this a video model with the duration set to zero?

No. Content type is image. Catalog description: Wan 2.5 in still-image mode, used standalone. You get a still, up to 4K, for 5 credits. Motion is a different slug.

Why would I use this instead of any other 4K stills row?

Because the compositions are meant to match Wan footage — first frames and thumbnails that belong next to Wan clips. If the motion pass is not Wan, that resemblance is not a reason to be here.

Can I attach a reference photo?

requires_image is false and the only category is text-to-image. This slug is generate-from-text. An edit or image-to-image pass is a different row.

What do 5 credits buy?

One Wan Preview 2.5 Text to Image generate: a 4K-capable still, Realistic or Artistic, one of five aspect ratios. Not a clip, not audio, not a twelve-angle pack inside one call.