Stable Diffusion 3.5 Large Stability AI vs Wan 2.7 Alibaba

Stable Diffusion 3.5 Large

• + Excellent handling of glass transparency and realistic surface scratches
• + Correct lighting direction from the left with sharp, realistic shadows

• − Failed the spatial request by putting the red book inside the cube and the sphere on top of the book
• − The plant is not clearly visible through the glass as requested

Wan 2.7

• + Perfectly followed the complex spatial instructions
• + High-quality textures on the wooden table and the red book
• + Clear and accurate glass refraction and reflections

• − The glass cube has a strange vertical internal pane that wasn't requested

Stable Diffusion 3.5 Large

• + Excellent skin texture and realistic age-related details on the man's arms and face.
• + Beautiful lighting and reflections that convey a cinematic, rainy atmosphere.
• + The red bicycle color is vibrant and correctly executed.

• − Physical interaction between the hands and the bike is messy with some structural clipping.
• − The rain effect looks a bit like static or a digital overlay in some areas.

Wan 2.7

• + Very convincing street photography aesthetic with 'imperfect' framing as requested.
• + The environment feels authentic to a Japanese city street with believable background elements.
• + Good source preservation of the 50mm lens look with natural background figures.

• − The man's skin looks somewhat waxy and lacks the fine texture requested in the prompt.
• − The bicycles structure is distorted, particularly around the handlebars and front fork.

Stable Diffusion 3.5 Large

• + Excellent intricate engraving on the plate armor.
• + Sharp, lifelike eyes and facial expression.
• + Captures the sense of a large-scale battlefield in the background.

• − Missed the request for beads in the braided hair.
• − Armor appears almost too clean and reflective despite the battle-worn description.

Wan 2.7

• + Perfect adherence to the 'beads in hair' prompt detail.
• + Highly realistic skin textures, scars, and dirt application.
• + Strong implementation of warm torchlight, bokeh sparks, and textured leather straps.

• − The armor engraving is slightly less detailed compared to the other image.
• − The background wall is a bit generic.

Stable Diffusion 3.5 Large

• + Features a bold, experimental layout with high-quality food photography
• + Strong adherence to the 'grid' request for photos
• + Excellent rendering of the word 'Menu' in a bold sans-serif font

• − Most of the secondary text is illegible gibberish
• − The layout feels more like a poster than a functional restaurant menu

Wan 2.7

• + Highly realistic and functional menu layout with logical sections
• + Impressive text rendering for dishes, prices, and even social media handles
• + Cleverly includes professional elements like a QR code and restaurant info

• − The grid of photos is slightly smaller and less 'modern minimalist' than Image A
• − Adds environmental props (pen, rosemary) not explicitly requested in the design prompt

Stable Diffusion 3.5 Large

• + Excellent photorealistic texture on the meat and bun
• + Very coregent lighting with the fire background
• + Powerful, high-action atmospheric effect

• − Completely failed to include any of the requested text
• − The burger is stacked rather than being the requested 'exploded' view

Wan 2.7

• + Perfect adherence to all requested text elements with accurate spelling
• + Accurately represents the 'exploded' motion with suspended components
• + Creative fiery render on the typography

• − Image has a slightly more 'digital illustration' feel compared to the realism request
• − Some floating sesame seeds look a bit disconnected

Stable Diffusion 3.5 Large

• + Excellent contextual composition showing the menu within a realistic café environment.
• + Captures a very authentic chalk texture with dust and smudging on the board.

• − Significant spelling errors throughout the board (e.g., 'TODAAY', 'Ottpups', 'Chocoolalfe').
• − Incorrect date rendered as 2024 instead of the requested 2026.
• − Font style looks more digital/aligned than the requested organic handwriting style.

Wan 2.7

• + Perfect text accuracy and spelling for all requested menu items and numbers.
• + Successfully rendered the correct date (April 30, 2026) as specified in the prompt.
• + Strong adherence to the 'handwritten cursive' style with natural-looking ligatures.

• − The text has a slight 'glow' or drop-shadow effect that makes it look like a digital overlay rather than physical chalk.
• − The composition is a tight crop on the board, offering less environmental context than Model A.

Stable Diffusion 3.5 Large

• + Excellent cinematic atmosphere with realistic lighting and nebular dust.
• + High level of detail in the astronaut suit and horse's texture.
• + Strong composition that feels integrated into the space environment.

• − The astronaut's leg position on the saddle is slightly anatomicaly awkward.

Wan 2.7

• + Clear, sharp focus on all elements including the background planets.
• + Well-defined anatomy for both the horse and the astronaut's gear.
• + Vibrant colors and a clean, surreal aesthetic.

• − The horse appears to be floating staticly rather than moving dynamically.
• − Lighting on the subjects feels slightly detached from the background environment.

Stable Diffusion 3.5 Large

• + Excellent texture on the capybara's fur and clothing.
• + Vibrant color palette that captures a night aesthetic.
• + High resolution with sharp details on the interior hardware.

• − Completely failed to include the human businesswoman in the back seat.
• − The capybara's anatomy is slightly distorted with human-like legs/torso.
• − Only one paw is visible on the wheel despite the prompt's request for both.

Wan 2.7

• + Successfully included all elements of the prompt including the businesswoman and the capybara driver.
• + Perfectly captured the requested bored expression on the passenger's face.
• + Accurately depicted both paws on the steering wheel.

• − Lighting is a bit flat compared to Model A.
• − The textures are slightly less refined, especially the hair of the human passenger.
• − The capybara's face is a bit more static and less expressive.

Stable Diffusion 3.5 Large

• + Successfully captures a moody, high-contrast cinematic lighting
• + Includes the gothic title and scroll banner with mostly legible text
• + Excellent use of the 'twisted trees' and 'night sky' thematic elements

• − Missed all specific event details (Date, Time, Location) at the bottom
• − Text at the bottom of the scroll degrades into gibberish
• − The 'central' jack-o-lantern is offset to the side rather than the focal point

Wan 2.7

• + Flawless adherence to all text requirements, including specific event details
• + Superior composition with the jack-o-lantern appearing as the central focus
• + Intricate border featuring webs, thorns, and skulls that perfectly fits the gothic theme

• − Lighting is more illustrative than 'cinematic' or 'moody'
• − The color palette is slightly brighter than requested by 'dark parchment' and 'night sky'

Stable Diffusion 3.5 Large

• + High level of realistic detail in the sushi textures.
• + Coherent spatial arrangement of the objects.
• + Good lighting and shadows.

• − Failed to place text at top-center, opting for a physical card instead.
• − Composition is a bit cluttered with extra side dishes and garnishes.
• − Camera angle is too low for a requested 45 degree top-down view.

Wan 2.7

• + Perfect adherence to the text placement and layout instructions.
• + Clean, minimalist 3D cartoon aesthetic that matches 'miniature diorama' request.
• + Accurate 45 degree isometric perspective.

• − The shrimp sushi has some slight anatomical irregularities in the tail.
• − The sushi rolls are very simplistic in texture.

Stable Diffusion 3.5 Large

• + Excellent depiction of dynamic movement through blurred backgrounds and paws off the ground.
• + Strong lighting effects with prominent bokeh and soft, hazy god rays.
• + Captures a very joyful and whimsical expression on the animals' faces.

• − The kitten looks like a miniature fox or generic canine hybrid rather than a tabby kitten.
• − The environment is quite blurry, losing some of the 'lush wildflower' detail.
• − The animals overlap in a way that creates some anatomical confusion in the midground.

Wan 2.7

• + Perfect prompt adherence for all four distinct animals, including a clear tabby pattern on the kitten.
• + Excellent 'god rays' and dew sparkle effects that feel integrated into the physics of the scene.
• + Sharp detail across the entire frame, showing individual petals and grass blades.

• − The fox's face appears a bit stiff or grumpy compared to the 'joyful' prompt.
• − The butterfly sizes are slightly inconsistent relative to each other.

Stable Diffusion 3.5 Large

• + Strong minimalist vector aesthetic
• + Includes the cloche, steam, and banner as requested
• + Applies a nice subtle paper texture to the background

• − Spelling error in primary text: 'Cafféé' instead of 'Caffè'
• − The cloche illustration is a bit abstract and lacks clarity

Wan 2.7

• + Excellent typography with correct primary spelling 'Caffè'
• + Very clean vector execution with a classic circular emblem layout
• + Detailed and recognizable cloche dome illustration

• − Slight misspelling in the name as 'Florion'
• − The cloche is placed below the main text rather than having the text on a banner below it

Stable Diffusion 3.5 Large

• + Features complex, detailed planetary illustrations
• + Adheres well to the requested navy, white, and red color palette

• − Includes a Space Shuttle instead of the requested Saturn V rocket
• − Text is completely illegible and nonsensical
• − Fails to follow the sequential numbered steps requested in the prompt

Wan 2.7

• + Accurately depicts the six requested steps with relevant icons for each
• + Text and labels are highly legible and contextually accurate for the mission
• + Strictly follows the flat-vector design style and color palette

• − Minor spelling errors in text (e.g., 'Descript' instead of 'Descent' and 'Tranquiliry')
• − The background stars are somewhat cluttered compared to the clean vector aesthetic