TEXT AND IMAGE INPUTS
Wan 2.7 Spicy: Wan 2.7 image to video and text inputs
Wan 2.7 Spicy is not a separate official edition established by this page. The local Wan Spicy catalog checked on 10 November 2026 labels both task entries Wan 2.7. Choose image-to-video for an existing composition or text-to-video for a written scene, then use the controls belonging to that task.
Maintained by Wan Spicy · Last reviewed:
What Wan 2.7 Spicy means in the model selector
The reviewed Studio exposes entries named Wan 2.7. A Spicy search qualifier does not, by itself, identify an official edition or a universal content policy.
When comparing a service, read its full model name and task. In Wan Spicy, choose the input mode first and then inspect the Wan 2.7 entry exposed for that mode.
Choose the input mode before writing the prompt. Text-to-video must establish the subject and setting from words, while image-to-video begins with an existing composition and should concentrate on movement and continuity. Mixing those two jobs into one evaluation makes it difficult to explain why a result changed.
Label each saved attempt with its input mode so a later comparison does not silently combine two different generation tasks.
The Wan Spicy models separates the catalog label from legacy search references.
Choose a first frame or a written scene
Use an image when appearance and framing have already been chosen. Describe the movement around what is visible instead of asking the model to redesign the source.
Use text when you want to establish the subject, environment and shot together. Write the scene’s initial state before describing its movement. The two modes should not be assumed to share every setting.
For image input, note what must stay recognizable and what is allowed to move. A concise instruction can identify the primary body action, the camera behavior and a stable wardrobe or background cue. Avoid adding several new objects when the purpose of the attempt is to animate the supplied frame.
Return to Wan Spicy for the platform overview, then choose the input mode inside Studio.
When choosing a first frame, use the NSFW image to video source checklist before writing the motion.
Record the input mode before comparing attempts
Save whether the attempt used a written scene or a source image alongside the exact Wan 2.7 selector label. An image supplies composition that a text request must describe, so changing the mode changes the comparison.
Keep the prompt, selected controls and acceptance condition in the same record. For a fair task comparison, note any option available in only one mode instead of treating the requests as identical.
For text input, establish the adult subject, scene, framing and action in that order. Then add one pacing or camera cue. Keep the first attempt simple enough that you can tell whether a later improvement came from the wording, a setting change or a different model.
Use the Wan video prompts for scene and motion wording; this guide focuses on the Wan 2.7 task selection.
Check audio and framing in the selected mode
The local catalog checked on 10 November 2026 offers 5, 10 or 15 seconds at 720p or 1080p for both Wan 2.7 tasks. Both expose prompt enhancement, defaulting off; the image task additionally exposes single- or multi-shot selection, defaulting to single. Some selections require subscriber access.
Neither reviewed Wan 2.7 task exposes a separate audio toggle or aspect-ratio selector. Their absence is not proof of silent output or guaranteed framing. The image entry accepts one source image up to 20 MiB. These are local configuration facts, not proof of active Production availability or output quality.
Keep an experiment log with the task type, model label, duration, resolution, prompt and decision. Compare complete clips at the same review points rather than selecting a single attractive frame. Wan Spicy does not publish a controlled quality winner for every prompt, so your recorded criteria should explain the choice.
For a source-image trial, inspect identity, pose, limbs, clothing and a distinctive background cue at the opening, middle and end. For a text-only trial, also check whether the requested subject and scene were established. Those are different evaluation questions even when the model label is the same.
Do not interpret a locally visible selector as proof that a real provider request was completed. Production evidence needs an actual task record and, for a quality claim, the source, settings, attempts and selection method. Until then, the page documents the reviewed interface and a repeatable way to test it.
When the result will be compared with another input mode, use equivalent scene requirements rather than identical wording that favors one task. Note any unavailable controls and avoid assigning a winner when the source, duration or output size could not be matched.
Publish the limits with any example: a single selected clip does not show the rejected attempts, the range of possible outputs or how another source would behave. A transparent note about selection and missing data makes the example useful without overstating certainty.
Compare Wan 3.0 Spicy settings when the decision concerns the separately named image-to-video entry, not an unrecorded quality ranking.
QUESTIONS, ANSWERED
Frequently asked questions
What is Wan 2.7 Spicy called in the model selector?
The page covers the search phrase, while the reviewed local task entries are named Wan 2.7. Use the selector’s exact label when recording a test.
Should image and text modes expose identical settings?
Not necessarily. The input mode changes the request, and controls must follow the selected task entry.
Has Wan Spicy benchmarked Wan 2.7 quality?
No controlled real-provider benchmark is published here. Local simulated generation verifies interface behavior, not model quality.
Which Wan 2.7 duration and resolution options were checked?
The local catalog checked on 10 November 2026 lists 5, 10 and 15 seconds with 720p or 1080p for both text and image tasks. Some selections require subscriber access. Inspect the current task before submitting; these facts do not prove Production activation.
Does Wan 2.7 have an audio switch or aspect-ratio selector here?
Neither reviewed Wan 2.7 task exposes those separate controls. Do not copy them from a Wan 3.0 entry or an external provider. The absence of an audio switch does not establish whether an output will contain sound.
How do Wan 2.7 text and image controls differ?
Both reviewed tasks have prompt enhancement. The image task additionally accepts one source image up to 20 MiB and offers single- or multi-shot selection. Its source image supplies composition that the text request must describe.