Wan 2.7 Image Prompt Guide for Consistent Image Sets

Wan 2.7 Image prompts for a series must define both continuity and each frame's job. The Wan 2.7 image and video overview covers the generation routes; this guide focuses on writing coherent still-image sets.

Separate what stays fixed from what may change, then describe the sequence before decorative details. Sequential generation can help images relate, but it does not guarantee identical faces, products, typography, or geometry. This guide reflects the workflow available in August 2026.

TL;DR

• Write one reusable Set Bible for identity, product truth, style, palette, lighting, and camera grammar.
• Assign every uploaded reference one role and state which reference wins when two sources conflict.
• Build a Frame Map that gives each requested image a distinct composition, action, or narrative beat.
• Limit each frame to one or two deliberate changes through a Change Budget; repeat the locked details instead of relying on "same as before."
• Generate a small set first, review drift at full size, and revise only the weakest layer of the brief.

A practical prompt order is Deliverable, Set Bible, Reference Map, Locked Elements, Frame Map, Change Budget, Exclusions, Delivery. This order turns "make these consistent" into visible decisions that can be checked after generation.

1. Why Consistent Image Sets Drift

Most series prompts define a theme but not a system. "Create six cinematic images of a ceramic artist" leaves the face, clothing, studio, clay, lens feel, and time of day open to change. The word consistent does not rank those properties. Name the invariants and repeat them where the frame instructions begin.

Drift also appears when one frame changes pose, location, wardrobe, camera, lighting, and props at once. Too little visual evidence connects it to the previous image. Change one or two axes per frame while protecting the set's recognizable anchors.

References add ambiguity when their roles overlap. A portrait may have the right face but the wrong jacket; a palette board may contain unwanted people or furniture; a composition reference may conflict with the product angle. Input capacity is not role clarity, so state what to borrow, what to ignore, and which source has priority.

Alibaba describes Wan2.7-Image as a unified generation and editing model with personalization and multi-image creation in its official announcement. That is context, not a promise of production-ready consistency. Use an explicit brief and review every frame.

2. The Set Bible and Frame Map Framework

A Set Bible lists the reusable properties that make the series recognizable: identity, object geometry, wardrobe, environment, medium, palette, light, lens family, and finish. Each item should be visible enough to compare across frames.

A Frame Map defines each image's shot size, composition, action, orientation, negative space, and story beat. A Change Budget states which variables may change and which remain locked. Together, the three layers create variety without losing continuity.

Prompt layer	Question it answers	Example
Deliverable	What is the set for?	Four 3:4 editorial campaign images
Set Bible	What gives every frame the same visual identity?	Same person, green jacket, cool daylight, 50mm editorial photography
Reference Map	What does each uploaded image control?	Image 1: identity; Image 2: jacket; Image 3: palette only
Locked Elements	What must not drift?	Face, hair, jacket cut, camera finish, skin tone
Frame Map	What is distinct in each image?	Establishing, working detail, portrait, closing still life
Change Budget	How much may vary per frame?	Change pose and crop; keep wardrobe, location, and light fixed
Exclusions	What recurring failures should be blocked?	No duplicate subject, new accessories, or location changes
Delivery	What output shape is required?	Related set, 3:4, 2K, requested count four

The prompt need not be long; a simple set may use one sentence per layer. Separation makes a failed layer easier to identify and revise.

2.1 Write the Set Bible in Visible Terms

Replace "keep the vibe cohesive" with visible criteria. For portraits, specify identity, hair, wardrobe, palette, light direction, background, focal-length feel, and finish. For products, prioritize silhouette, closure, label position, material, reflections, and brand colors. Verify exact small text or add it during layout.

Rank competing anchors: "Priority 1: identity and freckles. Priority 2: indigo jacket and silver pin. Priority 3: cool window light and restrained grain." This tells the model and reviewer what cannot be traded for composition.

2.2 Give Every Reference One Job

Write the Reference Map before the scene: "Image 1 defines identity and hair only. Image 2 defines the jacket only. Image 3 defines the palette only; ignore its people and objects. Image 4 defines framing and negative space only." Add a priority rule when references disagree.

Use only images you own or have permission to edit, and use consenting adults for identity-led sets. Do not present generated portraits as documentary evidence or use them to impersonate another person. Reference control can improve creative continuity, but it does not change the responsibility attached to the source material.

2.3 Control Variation With a Change Budget

The Change Budget limits new information. A campaign might change pose and crop; a storyboard, action and camera distance; a product set, object rotation and one prop. Keep identity or geometry, location, palette, and light fixed.

Do not rely on "same character as Frame 1." Repeat the key anchors in the Set Bible and after the Frame Map so continuity is explicit.

3. Reusable Wan 2.7 Image Prompt Templates

The templates below are designed for sequential image generation. Replace bracketed fields, remove irrelevant lines, and request only the number of frames the project can clearly distinguish. A four-image set with four strong jobs is a better first test than twelve frames filled with near-duplicates.

3.1 Character or Editorial Series

Create a related set of [number] [ratio] editorial images for [campaign or story]. Set Bible: the same [fictional or authorized subject], with [identity anchors], wearing [wardrobe anchors], in [one environment]. Keep [palette], [light direction], [lens feel], [texture], and [finish] consistent across the set. Reference Map: Image 1 defines identity and hair only; Image 2 defines wardrobe only; Image 3 defines palette and light only. Frame Map: Frame 1: [establishing composition]. Frame 2: [action or detail]. Frame 3: [portrait or alternate crop]. Frame 4: [closing beat with negative space]. Change Budget: change pose and shot size only; keep identity, wardrobe, location, time of day, and color treatment fixed. No face drift, new accessories, duplicate subject, wardrobe changes, or unrelated background objects.

This structure makes the subject and visual treatment persistent while the Frame Map supplies progression. If a side profile is essential, include a reference that actually shows that angle. Asking one frontal portrait to support several extreme viewpoints increases the burden on both generation and review.

3.2 Product Campaign Set

Create a related set of [number] commercial images featuring the referenced [fictional product]. Set Bible: preserve the exact product silhouette, proportions, closure, material, packaging colors, label position, and logo orientation. Use [surface], [background], [palette], [light], and [camera finish] across every image. Reference Map: Image 1 controls product geometry and packaging; Image 2 controls material and reflections; Image 3 controls composition only. Frame Map: Frame 1: clean hero view. Frame 2: close material detail. Frame 3: product in use with one supporting prop. Frame 4: wide closing layout with copy space. Change Budget: change camera distance and one prop; keep product truth, surface, light direction, and color grade fixed. No redesigned packaging, extra products, new text, warped label, changed cap, or floating objects.

Product consistency should be judged at full size, not from a contact-sheet thumbnail. Check the silhouette, closure, label boundaries, highlights, and repeated decorative marks frame by frame. A coherent palette cannot compensate for a bottle or device that has changed shape.

3.3 Environment or Storyboard Set

Create a related [number]-image visual sequence set in the same [location] during [time and weather]. Set Bible: preserve the same architecture, floor plan, key props, horizon, light direction, palette, medium, and camera grammar. Frame Map: Frame 1 establishes [location and subject]. Frame 2 shows [one action]. Frame 3 reveals [new information] from [shot size]. Frame 4 ends on [specific composition]. Change Budget: change subject position and camera distance; do not change the location, weather, time of day, materials, or color treatment. Keep spatial relationships believable. No room redesign, extra doors, duplicate props, sudden season change, or inconsistent light direction.

Environment prompts benefit from spatial anchors such as "the tall window remains on the left wall" or "the kiln stays behind the worktable." These statements are more useful than "same room." When continuity matters across several angles, include references that reveal the layout instead of asking one view to define unseen geometry.

4. Copy-Ready Prompts for Consistent Image Sets

These prompts are production starting points, not benchmark claims. Keep the first returned set, record the model and settings, and review every frame against the stated anchors before deciding what to revise.

4.1 Four-Frame Ceramic Artist Story

Create a related set of four 3:4 editorial photographs about the same fictional ceramic artist in one daylight studio. Set Bible: a woman in her early thirties with an oval face, warm brown skin, short natural curls, and a small crescent-shaped birthmark below her left eye; indigo chore jacket over a cream shirt; pale stoneware clay; white plaster walls; one tall north-facing window on the left; soft cool daylight; muted indigo, clay beige, and eucalyptus green palette; natural 50mm editorial photography; fine film grain and restrained contrast. Frame Map: 1) medium-wide establishing portrait beside the worktable, looking toward camera; 2) close three-quarter view of both hands shaping one bowl on the wheel; 3) waist-up side portrait inspecting the same unfired bowl near the left window; 4) quiet still life of the finished bowl on the worktable with the artist softly visible behind it and clear space in the upper right. Change Budget: change action and crop only. Keep the same person, hair, birthmark, jacket, studio layout, window position, bowl material, light direction, palette, and finish. No extra people, wardrobe changes, jewelry, duplicate bowls in Frames 1-3, warm sunset light, or redesigned room.

Expected direction: four related editorial beats with a stable person and room, moving from introduction to process, inspection, and a usable closing image.

Illustrative AI-generated example. Review notes: This is the strongest of the three sets because the artist's face, short curls, indigo jacket, clay material, cool daylight, and restrained editorial finish remain recognizable from frame to frame. The sequence also moves through distinct shot roles instead of repeating one composition. Small details still vary, so it should not be read as proof of exact identity or object preservation. Treat it as a useful first pass that would still need full-size inspection before publication.

4.2 Square Skincare Launch Set

Create a related set of four 1:1 commercial photographs for a fictional mineral skincare launch. Set Bible: preserve the referenced low cylindrical frosted-glass jar, brushed silver lid, centered vertical coral label block, and small embossed wave mark; pale gray limestone surface; soft off-white background; cool daylight from upper left; crisp controlled reflections; restrained coral, silver, mineral gray, and sea green palette; premium studio photography without readable packaging copy. Frame Map: 1) centered front three-quarter hero view; 2) macro crop of the frosted glass, lid edge, and cream texture; 3) jar opened beside one smooth green stone, lid placed directly behind it; 4) wide top-down layout with the closed jar in the lower left and clean negative space in the upper right. Change Budget: change camera distance, lid state, and one prop only. Keep jar geometry, label block, wave mark, surface, background, light direction, palette, and finish fixed. No extra jars, new text, flowers, splashes, floating objects, changed lid color, or altered proportions.

Expected direction: a launch-ready square set with enough compositional variety for carousel use while the fictional package remains recognizable.

Illustrative AI-generated example. Review notes: The frosted jar, coral label block, mineral palette, limestone surface, and controlled studio lighting stay coherent enough to read as one launch set. The composition does not follow every requested beat, however: Frame 2 shows cream texture but omits the requested lid edge, while Frame 4 is a centered front-facing product view rather than a top-down layout with the jar in the lower left. This is useful prompt drift, not a perfect result. In a controlled follow-up, restate only those two frame requirements while keeping the successful product and lighting anchors unchanged.

4.3 Four-Beat Observatory Storyboard

Create a related set of four 16:9 cinematic storyboard frames in the same mountaintop observatory before dawn. Set Bible: one fictional young astronomer with a blunt black bob, round amber glasses, rust-red field jacket, and charcoal trousers; one brass telescope on a dark timber platform; circular concrete observatory with an open roof slit; deep blue pre-dawn sky; one warm practical lamp on the right; realistic cinematic still photography; 35mm lens family; cool blue shadows, restrained amber highlights, and subtle grain. Frame Map: 1) wide establishing frame with the astronomer entering from the left and telescope centered; 2) medium side view as she adjusts the telescope, roof slit visible above; 3) close portrait as reflected starlight catches her glasses, telescope edge in foreground; 4) wide rear view of her facing the brightening horizon, telescope on her right and empty sky reserved above. Change Budget: change action and camera distance only. Keep character identity, glasses, jacket, telescope geometry, observatory layout, lamp position, pre-dawn continuity, palette, and film finish fixed. No extra people, duplicate telescope, daylight jump, new furniture, changed jacket, or different architecture.

Expected direction: a readable four-beat visual narrative that progresses without changing the character, location, or photographic language.

Illustrative AI-generated example. Review notes: The astronomer's identity, bob haircut, amber glasses, rust jacket, telescope material, lantern position, and pre-dawn palette remain recognizable across the set. However, the images do not preserve the requested circular observatory or open roof slit, and the closing frame places the telescope on the opposite side from the Frame Map. That makes this a useful drift example rather than a perfect match. Consistent color and character are not enough when location geometry and prop placement are also locked elements, so audit those anchors separately.

5. Troubleshooting When the Set Looks Inconsistent

When identity drifts, shorten the Frame Map, strengthen the first two identity anchors, and state which reference has priority. Keep early poses within the reference pack's covered angles. If one frame is wrong, revise its instruction in a controlled follow-up request without rewriting the Set Bible, which could shift the whole series.

When style drifts, replace broad adjectives with one medium, palette, light direction, contrast level, lens family, and finish. Repeat them after the Frame Map. "Cinematic" is ambiguous; "realistic 35mm still photography, cool blue shadows, one warm practical light, restrained contrast, subtle grain" is reviewable.

When frames are too similar, give each image a job instead of asking for "four variations." Change shot size, action, or negative-space placement while protecting the anchors. Spend the Change Budget on the difference that advances the story or supports another placement.

When product geometry or room layout shifts, reduce viewpoint ambition. A front view cannot define a package's side, and one room image cannot establish every wall. Add an authorized reference for the missing angle and stay close to known geometry.

Finally, do not treat the requested image count as a guaranteed delivery count. In Wan 2.7 Image's sequential mode on Vofy, you can request up to 12 related images, but the model determines the actual number returned. Build the brief so a smaller set still has useful frames, then fill missing beats in a controlled follow-up rather than changing the whole visual system.

Illustrative diagnostic diagram: the locked anchors remain fixed across the sequence, while framing and action use the limited Change Budget. Read it as a review checklist for separating intentional variation from continuity drift, not as a model output benchmark.

6. Using the Framework in Vofy

Open the Wan 2.7 Image model page to review its current workflow, then launch Wan 2.7 Image in Vofy Studio. As of August 10, 2026, the reference-guided workflow accepts up to nine input images and supports 1K or 2K output. Standard mode creates one to four independent images, while sequential mode can request up to 12 related images; the model determines how many images are actually returned. Assign each reference a clear role, choose a supported ratio, and begin with a small set before scaling the brief.

Save the Set Bible outside the prompt field before revising. After generation, compare the outputs in this order: highest-priority identity or product truth, environment geometry, palette and light, frame-specific composition, then small artifacts. Change one layer at a time and keep the strongest wording fixed. The independent multi-reference image generation test provides a stricter role-separation method when several sources must survive in one image.

Vofy uses Credits for generation, and rates vary by model, resolution, quantity, and selected settings. Check the estimate shown in the current interface before expanding a test into a larger set. For upstream model information beyond the Vofy integration, use the official Wan website as the current vendor reference.

Disclosure: Vofy is an all-in-one AI creative studio and provides the practical workflow used in this section. The prompt framework and examples are editorial methods, not guarantees of identical output.

7. Conclusion

Consistent image sets are built from controlled repetition. Write one Set Bible, assign every reference a role, map each frame, and allow only a small number of deliberate changes at a time. Sequential generation can help the images relate, but the prompt still has to define what relationship means for the project.

The warning is straightforward: do not confuse a matching color grade with real continuity. Review faces, products, props, architecture, and light direction before approving the set. A series is only as consistent as the detail that matters most to its intended use.

FAQ

How do I prompt Wan 2.7 Image for a consistent character?

Use one clear identity reference, list visible identity anchors, and state that identity has the highest priority. Keep early changes limited to pose, action, or crop, and include references for important side or three-quarter angles. Repeat the identity anchors after the Frame Map instead of relying only on "same character."

What is the difference between standard output and sequential image generation?

Standard mode creates independent results, with one to four outputs documented in Vofy's Wan 2.7 Image workflow. Sequential image generation requests a related set for a shared story or campaign direction and can request up to 12 images. The actual returned count and fine-detail consistency can vary.

How many reference images can I use with Wan 2.7 Image?

Vofy's reference-guided Wan 2.7 Image workflow accepts up to nine input images as of August 10, 2026. Use only the references that have a defined job. More inputs can create more conflicts when identity, palette, composition, or style roles overlap.

Can Wan 2.7 Image keep product packaging exactly the same?

It can use product references and preservation instructions, but exact geometry, label details, and small text still require inspection. Define the package as the highest-priority reference, limit camera changes, and verify each output at full size. Apply regulated or exact campaign copy during final layout when necessary.

Should every frame use the same prompt?

Every frame should share the same Set Bible and locked elements, but each needs a distinct Frame Map instruction. Repeating one prompt without frame roles often produces near-duplicates. Rewriting everything for each frame often causes drift. Keep the anchor stable and vary only the planned composition or action.

Does a consistent prompt guarantee identical results?

No. Generative output varies with model updates, references, settings, and sampling. The framework reduces ambiguity and makes drift easier to diagnose; it does not turn image generation into deterministic compositing.

Related Reading

• Wan 2.7 Image and Video Models on Vofy
• Multi-Reference Image Generation Test
• How to Write Better GPT Image 2 Prompts