Wan 2.5 (Preview) Alibaba vs Wan 2.6 Alibaba

Wan 2.5 (Preview)

• + Excellent photographic quality with realistic dust motes and soft bokeh.
• + High-quality texture on the red book and the wooden table.
• + Clean, minimalist composition that looks professional.

• − The blue sphere is completely opaque, lacking global illumination/refraction inside the glass cube.
• − Odd floating reflection/artifact directly under the book on the top glass pane.

Wan 2.6

• + Successfully captured the sphere as translucent glass rather than opaque matte.
• + Strong adherence to the 'window light from the left' instruction with visible window frame shadows.
• + The plant is clearly behind the cube and visible through the glass.

• − The glass cube has some structural inconsistencies in its edges.
• − The scale of the 'small' sphere is slightly large relative to the cube.

Wan 2.5 (Preview)

• + Successfully preserved the original car model and its orientation.
• + Accurately represents a generic California coastline background.

• − The man is poorly rendered and too small to detail.
• − The wheels suffer from severe motion blur artifacts that look unnatural.

Wan 2.6

• + The man is clearly visible and his likeness is well-preserved from the source.
• + High visual quality with a more dynamic and expansive landscape.

• − The car model was significantly altered and shortened from the original source.
• − The perspective shift resulted in losing the iconic front grille of the vehicle.

Wan 2.5 (Preview)

• + Excellent detail in the reflections on the wet pavement
• + Anatomically correct hands and realistic skin texture
• + Strong composition with a clear, sharp focus on the subject

• − Lacks visible motion blur from passing cars as requested
• − Rain effect is slightly less atmospheric compared to the competitor

Wan 2.6

• + Successfully incorporates motion blur on passing vehicles
• + Highly realistic depiction of water droplets on the man's jacket
• + Captures a gritty, cinematic street atmosphere with 'imperfect' framing

• − Minor anatomical inconsistency in the hands and how the tool is held
• − Face texture is a bit harsher and less natural than Image A

Wan 2.5 (Preview)

• + Excellent clarity on the engraved armor patterns.
• + Clean facial features with realistic textures.
• + Effective use of warm lighting on the shoulder plate.

• − The beads in the hair look more like modern piercings or studs.
• − Character looks a bit too young and clean for 'battle-worn'.

Wan 2.6

• + Superb adherence to 'battle-worn' with grit, sweat, and realistic dirt.
• + Excellent interpretation of hair beads as decorative jewelry.
• + Highly detailed texture on the leather straps and frayed cloth.

• − Slightly more cluttered composition compared to the cleaner focus of Model A.

Wan 2.5 (Preview)

• + Organized grid layout specifically following the three requested sections
• + Consistent lighting and plating style across all food photography
• + Cleaner font choice for a minimalist aesthetic

• − Nonsensical placeholder text even for large headers
• − Logical errors in categorization, such as placing pizza slices in the mains section

Wan 2.6

• + Includes realistic pricing and better text structure for a menu
• + Vibrant graphic design accents that match the prompt's request
• + High-quality, appetizing food photography with variety

• − Layout is cluttered with overlapping images and lacks clear whitespace
• − Misaligned headers where 'Appetizers' points to a pizza and 'Pizza' is listed twice in text lists

Wan 2.5 (Preview)

• + Excellent font rendering with a creative melting cheese and fire effect.
• + Clean, professional layout perfect for a digital advertisement.
• + Realistic burger textures and consistent light source from the embers.

• − The 'exploded' effect feels slightly static compared to the prompt's request for motion.

Wan 2.6

• + Dynamic 'exploded' composition with sauce dripping and ingredients angled for motion.
• + Intense fire and smoke effects that match the 'fiery' theme perfectly.
• + The text integration into the flames is very impactful.

• − The 'LIMITED TIME ONLY' text is partially obscured by floor flames, reducing legibility.
• − The burger toppings are slightly overlapping, making the individual components less distinct than Model A.

Wan 2.5 (Preview)

• + Excellent text legibility and accuracy
• + Captures the chalkboard aesthetic with realistic smudges

• − The text style looks slightly more like a digital marker brush than real chalk
• − Missing the 'Herbs' portion of the second menu item

Wan 2.6

• + Superb chalk texture with realistic dust and varying opacity
• + Follows the text prompt precisely including the final menu item
• + Excellent 'elegant cursive' interpretation for the title

• − Slightly lower contrast makes the bottom text a bit harder to read
• − The chalk dust at the bottom looks slightly clumped

Wan 2.5 (Preview)

• + High visual clarity and sharp textures on the horse's coat.
• + Cinematic lighting with a clear sense of depth and scale.
• + Dynamic composition with the horse appearing to gallop over the curvature of a planet.

• − Failed the primary negative constraint: the astronaut is riding the horse instead of the horse riding the astronaut.
• − Anatomical issues with the horse's legs, specifically the rear left leg extending awkwardly from the body.

Wan 2.6

• + Beautiful lighting effects and nebula colors in the background.
• + Detailed rendering of the space suit and horse tack.

• − Failed the primary negative constraint: the astronaut is riding the horse instead of the horse riding the astronaut.
• − The trailing reins appear to be floating or disconnected in an illogical way behind the astronaut.

Wan 2.5 (Preview)

• + Successfully replicates the outfit, jewelry, and sunglasses from Image 2.
• + Maintains high visual clarity and matches the lighting of the background well.

• − Failed the primary identity preservation constraint by replacing the face and hair of the person in Image 1 with the person from Image 2.

Wan 2.6

• + Correctly preserves the identity, unique skin patterns, hair, and sand on the face of the person in Image 1.
• + Accurately applies the coat, scarf, and sunglasses from Image 2 to the target subject.
• + Excellent source preservation of the background and wooden structure.

• − The scarf pattern is slightly modified (more red tones) compared to the original in Image 2.

Wan 2.5 (Preview)

• + Excellent photorealistic texture on the capybara's fur and the taxi dashboard
• + Dynamic composition with a more vibrant, neon-lit New York atmosphere
• + Strong focus on the interior details like the meter and steering wheel placement

• − The capybara's front paws look more like primate hands than rodent paws
• − The taxi roof light is positioned oddly as if visible through the windshield

Wan 2.6

• + Better anatomical representation of capybara paws on the steering wheel
• + Very convincing bored expression on the passenger that matches the prompt perfectly
• + Natural lighting and rain effects on the window glass

• − The composition is slightly less symmetrical with the passenger feeling very close to the driver
• − Slightly more blurring on the character details compared to Model A

Wan 2.5 (Preview)

• + Excellent text rendering with no spelling errors across all sections.
• + Clean, polished 3D-rendering style with high-quality cinematic lighting.
• + Well-defined border integration combining thorns, webs, and parchment.

• − The parchment texture is a bit too clean and light for a 'dark parchment' request.
• − Layout feels slightly crowded with the large curved title text.

Wan 2.6

• + Strong atmosphere with a dark, moody color palette and effective use of shadows.
• + Detailed twisted trees create a more immersive 'gothic' environment.
• + Good adherence to the 'dark parchment' texture requested in the prompt.

• − The small scroll banner has some minor graphical glitches/blending issues with the text background.
• − The text style for the event details is more generic compared to the title.

Wan 2.5 (Preview)

• + Excellent preservation of the original facial features and bone structure.
• + The hair texture matches the existing beard hair reasonably well.
• + Maintains the exact lighting and background from the source image.

• − The hairline on the forehead looks slightly artificial and stamped on.
• − The hair volume is a bit conservative compared to 'full and thick'.

Wan 2.6

• + Provides a very full, lush head of hair with great volume.
• + The hair flow and styling look very natural and well-integrated.
• + Successfully preserves facial identity and environmental context.

• − The hair slightly overlaps the glasses frame in a way that looks a bit less clean than Model A.
• − Slightly alters the shape of the upper head/forehead area to accommodate the volume.

Wan 2.5 (Preview)

• + Excellent typography with clean, bold execution and logical placement.
• + Vibrant rendering with high-quality PBR-style lighting and shadows.
• + Perfectly centered and clean composition.

• − The camera angle is lower than the requested 45-degree isometric view.
• − Text layout places the flag to the right rather than below 'SUSHI' as implied by the hierarchy.

Wan 2.6

• + Perfect adherence to the 45-degree isometric camera angle.
• + Superior miniature diorama feel with multiple sushi pieces and traditional wooden board.
• + Crisp text rendering and accurate flag icon.

• − The 'JAPAN' text is slightly off-center compared to the 'SUSHI' block.
• − The overall color palette is a bit more muted than Model A.

Wan 2.5 (Preview)

• + Features a classic big-head caricature style that leans more into the 'exaggerated' prompt.
• + Preserves the specific denim shirt from the source image better than the other model.
• + Incorporates a hockey broadcast on a screen to reinforce the professional setting.

• − The transition between the neck and the body looks somewhat disconnected.
• − The hockey stick is placed awkwardly to the side rather than being integrated into the character's actions.

Wan 2.6

• + Creates a more comprehensive and visually rich TV studio environment with professional lighting and equipment.
• + Stronger interaction with the props, showing the character holding the hockey stick and microphone simultaneously.
• + Variety in dog breeds adds more visual interest and character to the 'love for dogs' aspect.

• − The facial likeness is less distinct as a caricature of the specific source person compared to Model A.
• − Changes the character's clothing to a suit, losing the source image's denim shirt detail.

Wan 2.5 (Preview)

• + Excellent dynamic motion and energy across all four animals
• + Vibrant colors and clear focus on the foreground subjects
• + Accurately includes all four requested species with high-quality fur textures

• − The fox's eyes appear overly saturated and slightly unnatural in color
• − The water droplets/dew are disproportionately large

Wan 2.6

• + Beautifully soft lighting and atmospheric god rays that feel more integrated
• + More naturalistic fur rendering and eye colors for the animals
• + The 'tumbling' interaction requested in the prompt is better realized through closer proximity

• − The fox's front-right leg has an anatomical issue where it attaches to the body
• − The background kittens/fox feel a bit more cluttered in the composition

Wan 2.5 (Preview)

• + Excellent adherence to the Studio Ghibli cel-shaded animation style.
• + Preserves the composition and poses of the original meme perfectly.
• + The lighting and color palette feel nostalgic and warm.

• − The faces look more like modern generic anime than specific Ghibli characters.
• − The addition of floating leaves is a bit cliché.

Wan 2.6

• + Beautiful hand-painted watercolor texture that fits the prompt's request for textures.
• + Softer pastel color palette creates a more dreamy atmosphere.
• + Great preservation of the specific clothing details and facial expressions from the source.

• − The line art is slightly less defined, losing some of the structural clarity of the characters.

Wan 2.5 (Preview)

• + Successfully added wind-blown hair effect in a natural-looking direction.
• + Followed the instruction to add flying leaves throughout the scene.
• + Preserved the woman and dog's identity and posture perfectly.

• − The flying leaves are somewhat low-resolution and look like simple green shapes.
• − A small artifact appears above the dog's right ear where the tail of a leaf meets the fur.

Wan 2.6

• + Effectively rendered the hair blowing in the wind while maintaining facial clarity.
• + Added flying leaves with slightly better lighting and integration into the scene than Model A.
• + Excellent source preservation of the original environment and subjects.

• − The large leaf in the bottom left corner is slightly blurry compared to the rest of the image.
• − One small floating leaf near the woman's shoulder lacks a distinct stem, making it look a bit like a green blob.

Wan 2.5 (Preview)

• + Excellent typography including the requested accent mark.
• + Strong vector emblem aesthetic with a clear banner.
• + Balanced composition that feels like a complete brand identity.

• − The steam is inside the cloche which is physically unusual.
• − The background texture looks more like crumpled paper than a subtle logo texture.

Wan 2.6

• + More sophisticated and minimalist vector icon style.
• + Logical placement of steam rising from the cloche.
• + Beautifully rendered subtle 'distressed' texture on the background.

• − The Est. 1720 banner is much smaller and less prominent than requested.
• − Integration of the banner onto the side of the cloche is slightly unbalanced.

Wan 2.5 (Preview)

• + Excellent adherence to the infographic structure and steps.
• + Clean, professional flat-vector aesthetic with a solid NASA-inspired palette.
• + Accurate text rendering for labels and astronaut names.

• − The 'Descent' and 'Landing' text alignment is slightly floating and lacks connecting icons.
• − The Saturn V rocket illustration is a bit generic and resembles a modern shuttle mix.

Wan 2.6

• + Legible, clean typography for the title and astronaut names.
• + Correct use of the requested color palette.

• − Completely failed to generate the infographic steps and icons.
• − Image appears like a low-resolution printed towel or fabric texture instead of a crisp digital vector poster.
• − Lacks all visual elements requested except for the names.