Wan 2.6 Image to Video: the still is the contract (5s, 10s, 15s, 1080p, 15cr)
Wan 2.6 Image to Video is image-to-video. If the label is wrong, every second is waste.
Wan 2.6 Image to Video is image-to-video. If the label is wrong, every second is waste.
Wan 2.6 Image to Video is Alibaba WAN 2.6: a state-of-the-art image-to-video generation model with audio support, prompt expansion, and multiple resolution options. Category: image-to-video only. Requires image: yes. Audio: true. Durations: 5s, 10s, 15s. 1080p. 15 credits (the row is marked discounted on the snapshot). The image-to-video tool is the door. Text-to-video is a different door.
Upload the approved still or do not open the row
Image-to-video means the pixels of the hero are the brief. The model animates what you give it. A wrong SKU, a misspelled label, a face you have not cleared — those errors get 5, 10, or 15 seconds of life. That is the waste the claim names.
Do not "try it on a rough JPEG." Rough JPEGs become rough clips with audio. Approve the still on an image row first. Then this line.
Aspects: 16:9 and 9:16. That is the whole list. If you needed 1:1, you crop after or you pick another model. Qualities: 720p and 1080p. Ceiling: 1080p. Prompt expansion is in the description — describe the motion, not a new object. A new object belongs in a new still.
Audio is on. Silence is a sentence.
audio: true. You will get a soundtrack unless you steer it. Name room tone, the whoosh, the line, or "quiet, no music, no speech." A silent prompt on an audio-true row is how a 15-second product turn arrives with a choir.
Prompt expansion will also fill motion you left blank. Blank motion plus invented audio is two surprises. Write both.
5, 10, or 15 — not a continuous slider
Durations are three ticks. 5s for the orbit. 10s for the pour. 15s only if the action actually needs the hold. 15 credits is the catalog figure; longer is not "more value" if the still dies at second six.
This is Wan 2.6 image-to-video, not Wan 2.7 text-to-video. Different version, different input, different duration set (here there is no 2s, no 8s). The Wan roster is the map. Best image-to-video model is the ranked cut if this still-plus-audio-plus-1080p job is not actually the job.
If the label is wrong, fix the still. Every second you generate first is a second you will not ship.
FAQ
Can I run Wan 2.6 Image to Video from text only?
No. Requires image is true. Category is image-to-video. Without the plate, this row has nothing to honor.
What lengths can I generate?
5s, 10s, and 15s. 1080p. 15 credits on the snapshot. Audio support is on.
Will it add sound even if I do not ask?
Audio is true on the catalog row. Prompt the sound, including quiet. A mute assumption is how you get a bed you then have to replace.
Why not use Wan 2.7 for this?
Wan 2.7 text-to-video does not take the still as the contract. If the approved frame is the brief, stay on 2.6 image-to-video. If you have no frame, leave this page.