GPT Image 1 OpenAI vs Wan 2.5 (Preview) Alibaba

GPT Image 1

• + Perfect adherence to the object arrangement
• + Clean and realistic glass refractions
• + High resolution with smooth textures

• − Lighting is a bit flat compared to the dramatic side light requested

Wan 2.5 (Preview)

• + Excellent handling of window light and shadows
• + Highly realistic textures on the book cover and wooden table
• + Dynamic atmosphere with visible dust motes

• − The sphere appears to be floating or poorly anchored on the glass base
• − Distracting artifacts/dots in the air that might be interpreted as noise

GPT Image 1

• + Excellent preservation of the man's facial features and distinctive hairstyle
• + Maintains the specific car model and color with high fidelity
• + Realistic motion blur and high cinematic visual quality

• − The man's clothing is changed from the plaid coat to a dark scarf and sweater

Wan 2.5 (Preview)

• + Successfully places the specific car into a California coastline environment
• + Good composition with palm trees and coastal scenery

• − Significant loss of detail and accuracy in the man's face
• − The car is positioned as a right-hand drive vehicle, making the man a passenger in the context of driving in California
• − The man's iconic hairstyle is less preserved compared to Model A

GPT Image 1

• + Excellent shallow depth of field and bokeh realism.
• + Superior skin texture and natural lighting on the subject.
• + Effective cinematic color grading that feels authentic to a 50mm street shot.

• − The 'imperfect framing' request led to a very tight crop that cuts off much of the bicycle.
• − The bicycle's mechanical structure is slightly abstract and nonsensical in the rear wheel area.

Wan 2.5 (Preview)

• + Better inclusion of the full bicycle and tools, showing more of the 'repairing' action.
• + Strong reflections on the wet pavement that enhance the rainy atmosphere.
• + Clearer street context and background depth.

• − The skin texture appears slightly smoothed and plastic compared to the other model.
• − The rain droplets are rendered as very distinct white lines, which looks a bit artificial.
• − Lacks the motion blur from cars requested in the prompt.

GPT Image 1

• + Exceptional engraved texture on the plate armor
• + Highly realistic facial skin texture and expression
• + Effective warm, moody lighting that integrates well with the character

• − The beads in the hair are very subtle and blend into the color palette
• − Lacks visible leather straps requested in the prompt

Wan 2.5 (Preview)

• + Excellent adherence to specific details like the beads in hair and leather straps
• + Clearer bokeh effect with distinct sparks and a visible torch source
• + Impressive detail on the frayed cloth and chainmail underlayers

• − The texture on the face looks slightly smoother/less aged than Model A
• − The engraving on the armor looks a bit more repetitive/pattern-like

GPT Image 1

• + Excellent typography rendering with readable fonts.
• + High-quality, appetizing, and realistic food photography.
• + Clean and professional layout that looks like a real-world design.

• − The grid is slightly cut off at the bottom of the frame.
• − The placeholder text 'Apperoiation descrigion' contains spelling errors.

Wan 2.5 (Preview)

• + Successfully includes all three requested sections (Appetizers, Pizza, Mains).
• + Effective use of grid layout and color-coded section dividers.
• + Captures the modern minimalist aesthetic well.

• − Typography is mostly unintelligible gibberish.
• − The food photos are repetitive and look slightly more artificial.
• − The text 'Mains' appears under a slice of pizza and several salads, which is illogical.

GPT Image 1

• + Excellent typography rendering with consistent fiery glow
• + Clean and photorealistic textures on the burger components
• + Perfect centered composition for a traditional advertisement

• − The price text is incorrect showing '.99' instead of '6.99'
• − The 'exploded' effect is a bit stiff and lacks dynamic energy

Wan 2.5 (Preview)

• + Highly dynamic 'exploded' composition with a true sense of motion
• + All text elements including the price are rendered correctly
• + Creative integration of melting cheese effects into the typography

• − The starburst element is slightly busier and less clean than the other text
• − Some minor blurring on the edges of the floating lettuce

GPT Image 1

• + Excellent chalk texture with grainy, realistic strokes.
• + Perfect adherence to the requested text and pricing.
• + Consistent letter size and spacing that feels authentic to a real menu board.

• − The handwriting leans more toward neat printing than 'elegant cursive' for the title.
• − Minimal environmental context, focusing almost entirely on the board surface.

Wan 2.5 (Preview)

• + Dynamic composition with a realistic cafe background and depth of field.
• + Strong 'handwritten' feel with varying strokes and chalk smudges on the board.
• + Applies a nice slant to the text as requested in the prompt.

• − Failed to include 'Herbs' in the second menu item title.
• − Repetitive text error on the final item, showing '$9' twice on separate lines.
• − The chalk texture looks a bit too smooth and digital compared to the grain in Model A.

GPT Image 1

• + Excellent cinematic lighting and texture on the horse's coat.
• + Strong anatomical rendering of the horse and suit textures.

• − Failed the core prompt instruction of placing the horse on top of the astronaut.
• − Visual composition is standard and lacks the requested surrealism.

Wan 2.5 (Preview)

• + High vibrant contrast and clear, detailed background elements.
• + Good sense of motion with the flowing mane and dust particles.

• − Completely ignored the counter-intuitive instruction for the horse to be on top.
• − Standard 'astronaut on horse' interpretation which is the opposite of the specific prompt.

GPT Image 1

• + Excellent preservation of the subject's face, hair, and unique skin features
• + Accurately recreates the layers, pattern, and style of the clothing from Image 2
• + Natural lighting and seamless integration with the existing background

• − The watch and hands are slightly blurred compared to the rest of the image

Wan 2.5 (Preview)

• + High quality clothing reconstruction including the watch and ring accessories

• − Completely failed to preserve the person from Image 1, replacing him with the man from Image 2
• − The edit ignores the explicit instruction to keep the base person's face and hair unchanged

GPT Image 1

• + Excellent photorealistic texture on the capybara's fur and the jacket.
• + Cinematic lighting that accurately reflects a nighttime taxi interior.
• + The passenger is correctly positioned in the back seat as requested.

• − The passenger is seated in what looks like the front passenger seat rather than the back seat.
• − The composition is a bit tight, losing some of the 'New York' environmental context.
• − The capybara's paws have a slightly muddy, less defined texture.

Wan 2.5 (Preview)

• + Includes a highly detailed New York backdrop with vibrant city lights and rain effects.
• + Sharp focus on the capybara and the steering wheel details.
• + The passenger is clearly visible and correctly following the prompt instructions.

• − The passenger appears to be floating or sitting in an anatomically impossible position relative to the seat back.
• − The capybara's hands/paws look slightly more like human fingers with long nails than natural paws.
• − The taxi sign on top is strangely placed relative to the interior perspective.

GPT Image 1

• + Excellent typography and perfect text rendering for all requested details
• + Cohesive vintage aesthetic with a dark, atmospheric color palette
• + Sophisticated border design that integrates well with the overall composition

• − The jack-o-lantern is a bit subdued compared to the large text
• − Less dynamic lighting than model_b

Wan 2.5 (Preview)

• + Vibrant, cinematic lighting with impressive fire effects inside the pumpkin
• + High level of detail in the twisted trees and thorny border elements
• + Strong 3D pop and depth in the central illustration

• − The text is somewhat generic and lacks the 'vintage gothic' elegance requested
• − A bit cluttered, with different artistic styles clashing between the border, background, and center
• − The scroll banner is small and lacks the same presence as model_a's

GPT Image 1

• + Excellent preservation of the original face and features
• + Matches the exact lighting of the original scene

• − The hair looks like a poorly blended toupee rather than a natural growth
• − The hairline and temples have visible masking artifacts and blurry edges

Wan 2.5 (Preview)

• + Natural-looking hair texture and professional blending with the forehead
• + Excellent integration of hair behind the ears and along the temples
• + Preserves the original facial identity perfectly

• − Slightly altered the top of the ear to accommodate the sideburns

GPT Image 1

• + Excellent adherence to the requested isometric 45-degree perspective.
• + Clean, legible text rendering with a centered flag icon.
• + Higher complexity of sushi types including nigiri and a maki roll.

• − The raised diorama base is a bit large for the plate size.

Wan 2.5 (Preview)

• + Beautiful soft lighting and subsurface scattering on the rice and fish.
• + High-quality text rendering and graphic layout.
• + Clean, professional aesthetic with refined materials.

• − Missed the 45-degree top-down isometric angle, opting for a lower perspective.
• − Failed to center the flag icon as requested in the prompt.
• − The 'diorama base' is less distinct, looking more like a simple tiered plate.

GPT Image 1

• + Captures the likeness and specific denim shirt from the source image accurately.
• + Excellent caricature style with traditional watercolor textures and expressive facial exaggeration.
• + Cleverly integrates all elements into a unified scene (dog on desk, hockey on screen, news desk).

• − The hockey stick on the desk appears slightly disconnected from the perspective of the rest of the image.

Wan 2.5 (Preview)

• + Clean vector-style graphic that is highly readable.
• + Incorporates multiple dogs and a clear hockey theme on the background monitor.
• + Preserves the denim jacket and black shirt combination from the original photo.

• − Facial likeness to the source person is weaker compared to the other model.
• − The composition feels a bit more like a sticker or logo than a cohesive caricature scene.

GPT Image 1

• + Excellent anatomical rendering with natural lighting and fur textures.
• + Realistic integration of animals into the lighting environment with believable god rays.
• + Dynamic composition that truly captures the tumbling and chasing action requested.

• − The rabbit is slightly smaller and more obscured compared to other subjects.

Wan 2.5 (Preview)

• + Vibrant color palette with clear dew sparkles and flower details.
• + All four animals are clearly visible and well-spaced in the foreground.

• − The fox kit has unnatural, glowing blue eyes that don't match the prompt's photorealistic requirement.
• − Floating water droplets and seeds feel like static overlays rather than part of a 3D scene.
• − The animal anatomy and fur textures look somewhat cartoonish compared to the competition.

GPT Image 1

• + Excellent hand-painted texture that mimics a watercolor or colored pencil style.
• + Perfect adherence to the 'soft pastel colors' and 'nostalgic mood' requested.
• + Successfully translates the facial expressions into a stylized Ghibli-esque simplicity.

• − The composition is slightly tighter/cropped compared to the original image.
• − The background architectural details are almost entirely lost to the painterly wash.

Wan 2.5 (Preview)

• + Very high fidelity to the original source image's composition and poses.
• + Captures the modern Ghibli digital animation aesthetic with clean linework.
• + Preserves background details and clothing patterns with high accuracy.

• − The lighting is a bit flat and lacks the 'dreamy' quality requested.
• − The 'hand-painted textures' prompt was largely ignored in favor of a clean digital look.

GPT Image 1

• + Excellent adherence to the 'windy' hair request with realistic strands and lift.
• + Added a high density of leaves to create a true sense of motion and energy.
• + Modified the dog's tail and fur to match the wind direction.

• − The leash has become slightly blurred and warped compared to the original.
• − Introduced some minor artifacts around the flower area in the background.

Wan 2.5 (Preview)

• + Successfully preserved the original quality and sharp details of the woman and dog.
• + Applied the wind effect to the hair while keeping the face perfectly intact.
• + Added leaves that contrast well with the background.

• − The added leaves look like flat, green shapes rather than organic, flying foliage.
• − The wind effect feels less 'dynamic' because the dog's fur is completely static.

GPT Image 1

• + Excellent typography with the correct accent on 'Caffè'
• + Precise rendering of 'Est. 1720' text within the banner
• + Clean, minimalist vector style that feels modern yet vintage

• − Failed to provide a 'light background' as requested, using black instead
• − Minimalist steam trail is a bit overly simplistic

Wan 2.5 (Preview)

• + Followed the 'light background' with 'subtle texture' constraint perfectly
• + Superior composition with decorative framing and multi-layered banners
• + Includes aesthetic highlights and shadows on the cloche dome

• − The accent on 'Caffè' is slightly misplaced or stylized as a dot
• − Banner perspective makes 'Florian' look a bit cramped

GPT Image 1

• + Excellent adherence to the requested flat-vector aesthetic with clean, chunky icons.
• + Legible text for most labels and accurate silhouette imagery.
• + Well-balanced composition that uses the full frame effectively.

• − Includes a misspelled word 'EARLLUNAR'.
• − The icons for Earth Orbit and Translunar are a bit abstract, with Translunar being just an arc.

Wan 2.5 (Preview)

• + Features a more complex diagrammatic flow that clearly maps out the trajectory.
• + Superior text rendering with zero spelling errors.
• + Greater detail in the icons, particularly for the Lunar Module and the Moon surface.

• − The 'Descent' and 'Landing' labels are not aligned with specific icons, creating a gap in the infographic flow.
• − The character portraits for the astronauts are slightly distorted and inconsistent in style.