Makeartly
0
Upgrade

Wan 2.7 AI Video: Official Capabilities and Makeartly Availability

MODELWan 2.7

Reviewed 2026-08-11

Open the creator

Alibaba Cloud documents Wan 2.7 image-to-video, reference-to-video, and video-editing workflows. This article separates those official API capabilities from Makeartly’s current catalog and does not treat an undocumented Wan 3.0 as a released model.

Studio status: Under evaluation

The live catalogue did not confirm a model-specific route. Studio will open without preselecting this model.

Compare available model guides
Indie Film Car MorningPhone Glow BedroomCowgirl CosplayRemove Clothes HDBar Night FlashBathroom Panty Tease

Makeartly template inspiration — not Wan 2.7 output.

Release brief · verified against Alibaba Cloud

Wan 2.7 is a workflow release, not merely a prettier text-to-video checkpoint. Alibaba Cloud documents keyframe control, reference-to-video and video continuation alongside generation. This evidence board separates those official capabilities from what can actually be selected inside Makeartly today.

Editorial review:

Official Wan 2.7 workflow evidence

The following media is hosted by Alibaba Cloud documentation. Input clips and documented result clips are labelled separately; none is presented as an Makeartly generation.

01
Official continuation input: the source clip whose final motion and framing establish the next segment. Alibaba Cloud Model Studio
02
Official continuation result: inspect the join, direction of travel and whether scene geometry remains stable. Alibaba Cloud Model Studio
Wan 2.7 image-to-video input example 03
Reference-frame example from the official guide, used to anchor subject, palette and composition. Alibaba Cloud Model Studio
Wan 2.7 workflow interface example 04
Documented workflow frame showing how a controlled video task is assembled before generation. Alibaba Cloud Model Studio
Wan 2.7 start-frame example 05
Start-frame evidence: composition is supplied explicitly rather than left for a text prompt to invent. Alibaba Cloud Model Studio
Wan 2.7 end-frame example 06
End-frame evidence: a destination frame constrains where motion and composition need to arrive. Alibaba Cloud Model Studio

Three ways to brief a Wan workflow

These are reproducible planning briefs based on the documented modes. They are not claimed benchmark results; use them to decide which input control your shot actually needs.

01

Continue an existing movement

Preserve direction and rhythm beyond the source clip.

Prompt

Continue the same forward camera movement through the greenhouse aisle; the subject keeps walking at the same pace, leaves move gently, morning light remains consistent, no cut or change of lens.

Settings
  • Video continuation
  • Source clip
  • Match camera speed
  • No scene change
Finding

Continuation succeeds when the source ends on readable motion. A chaotic final frame gives the model no stable direction to carry forward.

02

Guide the destination with keyframes

Use explicit start and end compositions for a visual transition.

Prompt

Move naturally from the seated start pose to the standing end pose, preserve the fictional adult character and wardrobe, keep the camera locked, smooth weight transfer, no new objects.

Settings
  • Start frame
  • End frame
  • Locked camera
  • Short duration
Finding

Keyframes constrain composition but do not specify the path between them. Describe the physical transition and keep camera movement out of the prompt unless it appears in both frames.

03

Reference a repeatable character

Build a new shot while maintaining visual elements from references.

Prompt

The referenced fictional adult character crosses a quiet hotel lobby at dusk, same face, hairstyle and charcoal coat, low tracking shot, polished stone reflections, one continuous take.

Settings
  • Reference-to-video
  • Character reference
  • 16:9
  • Tracking shot
Finding

Reference media should agree with itself. Conflicting hairstyles, lenses or lighting across references force the model to average details rather than preserve a clean identity.

Wan workflow modes at a glance

Pick the smallest amount of control that makes the shot possible. More references are useful only when each one answers a different question.

DecisionText / image to videoKeyframesContinuation
ControlsIdea and opening frameStart and destinationWhat happens next
Best forNew short shotsPlanned transitionsLonger sequences
Main riskComposition driftUnnatural in-between motionVisible join or direction change
Inspect firstSubject and framingWeight and trajectoryBoundary continuity

How to evaluate Wan 2.7 without confusing availability

  1. Confirm the selectable version

    Open Makeartly Studio and check the model picker. Editorial coverage and a live selectable integration are separate states.

  2. Choose one control mode

    Start with text/image-to-video, keyframes or continuation according to the shot. Do not add references simply because the workflow accepts them.

  3. Prepare clean boundary media

    For keyframes and continuation, match aspect ratio, subject scale and lighting before generation. Input disagreement becomes motion drift.

  4. Draft the shortest useful segment

    Test the transition and camera direction at short duration before extending the sequence or raising quality.

  5. Inspect the join

    Watch the boundary repeatedly. Look for a speed jump, changed focal length, shifted horizon or subject-scale change.

Where the documented controls matter

Sequence extension

Continue a proven clip while carrying its final camera vector, lighting and subject movement into the next segment.

Designed transitions

Use start and end frames when a shot must arrive at a specific product position, pose or composition.

Repeatable campaign characters

Reference-to-video is the relevant mode when the same fictional character or product must appear across separate scenes.

Core capability

Official capabilities documented by Alibaba Cloud

Alibaba Cloud documents Wan 2.7 image-to-video with text, image, audio, and video inputs. Its guide covers first-frame generation, first-and-last-frame generation, and video continuation; it also describes automatic or custom audio, multi-shot prompting, and 2–15 second output at 720p or 1080p. A separate official API reference describes instruction-based video editing and style transfer.

Makeartly availability

Wan 2.7 is available in Makeartly’s custom generator. The production allowlist includes Wan 2.7 text-to-video, image-to-video, reference-to-video, and video-editing IDs. No curated exact-version template is currently attached to this article; generator availability and curated-template availability are different states.

Makeartly test and alternative evidence

No curated Wan 2.7 article test gallery is attached or verified. That absence does not change the custom-generator availability above, and Alibaba Cloud examples are not Makeartly runs. If you generate with Wan 2.7, evaluate the result against its own prompt, input, and selected mode.

Evidence

Prompt guidance and use cases

For a Wan 2.7 custom-generator test, begin with one subject, one setting, one action, and a single camera direction. Add a clear first frame when composition matters; use a final frame only when the ending matters enough to constrain it. This is editorial prompt guidance based on the documented workflow, not a prediction of any particular result.

Compare related video models

Related pages include the Wan AI Video Generator guide, the Seedance 2.5 release explainer, and the Kling 3.0 release explainer. Each page has a separate availability statement.

FAQ

Does Makeartly run Wan 2.7?

Yes. Wan 2.7 is available in Makeartly’s custom generator. Its production allowlist includes text-to-video, image-to-video, reference-to-video, and video-editing IDs; this article does not have a curated Wan 2.7 template gallery.

Is Wan 3.0 released?

No official Alibaba source for a Wan 3.0 release was verified for this article’s review date. This page therefore does not present Wan 3.0 as released.

What does Alibaba Cloud document for Wan 2.7?

Its official documentation covers multimodal image-to-video inputs, first-and-last-frame generation, video continuation, reference-to-video, and instruction-based editing. Availability varies by region in Alibaba Cloud.

Are there Makeartly tests for Wan 2.7?

No curated Wan 2.7 article test gallery is attached or verified. That is not evidence that the custom-generator option is unavailable.