GPT Image 1.5 OpenAI vs Wan 2.5 (Preview) Alibaba

GPT Image 1.5

• + Excellent adherence to lighting instructions with a clear source from the left.
• + The glass cube rendering is structurally sound with realistic reflections on the base.
• + The plant is very clearly visible through multiple facets of the glass.

• − The sphere is slightly large for the described 'small' blue sphere.
• − The book alignment looks a bit too perfectly centered, bordering on sterile.

Wan 2.5 (Preview)

• + Beautiful naturalistic lighting with visible dust motes and soft shadows.
• + The sphere has a very tactile, matte finish that contrasts well with the glass.
• + Includes the plant in a pot, adding more context to the scene.

• − The book is floating slightly above the glass cube rather than sitting on it.
• − The glass cube has distorted geometry near the top right corner where it meets the book.

GPT Image 1.5

• + Excellent preservation of the man's facial features and specific hairstyle
• + The man is clearly interacting with the vehicle
• + High quality rendering of the car's interior and materials

• − The car is positioned awkwardly facing the 'wrong' way compared to the road's curve
• − Changes the orientation of the steering wheel to the right-hand side

Wan 2.5 (Preview)

• + Perfectly preserves the exact look of the white car from the source image
• + Excellent sense of motion and dynamic composition
• + Accurately represents a California coastline road

• − The man is very small and difficult to see clearly
• − The man's unique hairstyle is not perfectly preserved at this distance

GPT Image 1.5

• + Excellent adherence to the 'imperfect framing' and 'candid' aspects of the prompt.
• + Highly realistic lighting and reflections on the wet pavement.
• + Naturalistic skin textures and convincing interaction with the bicycle tools.

• − The motion blur on the passing car is fairly subtle.

Wan 2.5 (Preview)

• + Strong bokeh and shallow depth of field effects.
• + Clear rain streaks visible in the background.
• + Correct inclusion of various bicycle parts on the ground.

• − The bike anatomy is slightly illogical with a floating kickstand and disconnected gears.
• − The lighting on the man's shirt feels a bit too clean and studio-lit for a 'candid' rain shot.
• − The composition is very centered, ignoring the 'imperfect framing' instruction.

GPT Image 1.5

• + Exquisite texture on the weathered leather and hammered metal
• + Highly realistic skin texture with believable dirt and scars
• + Subtle and professional color grading using warm torchlight

• − The braids are somewhat messy and loose compared to the prompt's bead-focused request

Wan 2.5 (Preview)

• + Very distinct and clear rendering of hair beads
• + Strong adherence to the 'ornate engraved' armor description
• + Good use of bokeh and shallow depth of field

• − Skin texture appears slightly smooth and plastic compared to Model A
• − Lighting on the face is a bit flat despite the presence of a bright torch

GPT Image 1.5

• + Excellent typography with perfect legibility and no spelling errors.
• + Realistic food photography that accurately represents the menu items.
• + Professional layout that mimics a real-world restaurant menu.

• − The color accents at the top are very subtle compared to the request for 'vibrant accents'.
• − The grid layout for photos is less symmetrical than Model B.

Wan 2.5 (Preview)

• + Creative use of a grid layout for food photography.
• + Strong adherence to the 'vibrant accents' and minimalist aesthetic.
• + Good spatial balance between images and text blocks.

• − Nonsensical text with numerous spelling errors and gibberish characters.
• − Visual glitches in some food images, making the dishes look artificial.
• − Poor font choice for small body text that is blurred and unreadable.

GPT Image 1.5

• + Excellent photorealistic texture on the meat patty and toasted bun
• + Dynamic, chaotic energy that perfectly matches the 'exploded' request
• + Strong integration of fiery glowing effects throughout the entire composition

• − The text 'LIMITED TIME ONLY' is slightly plain compared to the rest of the graphics
• − The starburst shape for the price is a bit busy with overlapping sparks

Wan 2.5 (Preview)

• + Very clean and readable typography with a creative 'dripping' fiery effect
• + Elegant, clear distribution of burger components that allows each ingredient to be seen
• + High-quality rendering of the price starburst as a distinct graphic element

• − The background is a bit too clean and lacks the intense 'fiery' atmosphere requested compared to the other model
• − The image feels a bit more like a composite graphic than a photorealistic physical explosion

GPT Image 1.5

• + Excellent chalk texture with realistic grit and varying opacity.
• + Highly accurate spelling and adherence to the prompt's specific date and items.
• + Convincing natural handwriting style that feels authentic to a café environment.

• − The composition is a tight crop, showing less of the 'cozy café' atmosphere.
• − The cursive for the title is somewhat simplified rather than 'elegant'.

Wan 2.5 (Preview)

• + Strong composition showing the chalkboard within a blurred, cozy café background.
• + Consistent, bold handwriting style with good contrast.
• + Effective use of layout with underlines for each menu item.

• − Failed to include 'Herbs' in the Grilled Octopus item.
• − The price for the cookies is repeated awkwardly on two lines.
• − Text appears significantly cleaner and less grainy than real chalk, bordering on a digital look.

GPT Image 1.5

• + High level of textural detail in the spacesuit and lunar ground.
• + Cinematic lighting with a gritty, realistic atmosphere.
• + Excellent composition featuring a variety of space elements like asteroids and a moon lander.

• − Failed to follow the specific spatial instruction (horse on top, not vice versa).
• − Anatomical issues with the horse's front legs and hoof structure.

Wan 2.5 (Preview)

• + Vibrant, surreal color palette with a beautiful nebula effect.
• + Clean rendering of the astronaut and horse with fewer anatomical glitches.
• + Dynamic sense of motion with the flowing mane and dust trails.

• − Failed to follow the specific spatial instruction (horse on top, not vice versa).
• − Less surface detail on the astronaut's gear compared to the competitor.

GPT Image 1.5

• + Successfully preserved the lower half of the original person's face and skin condition.
• + Maintained the background and the wooden structure accurately.
• + Correctly applied all components of the outfit including the scarf and watch.

• − Cropped the head off the person, failing to manage the full composition.
• − Could not verify if the eyes/hair were preserved due to the crop.

Wan 2.5 (Preview)

• + Applied the outfit, accessories, and jewelry with high fidelity to the source.
• + Great integration of lighting and shadows on the clothing layers.

• − Completely failed to preserve the person from Image 1, replacing him with the face from Image 2.
• − Ignored the explicit instruction to keep the person's exact face and hair unchanged.

GPT Image 1.5

• + Excellent photorealism with gritty, cinematic lighting
• + Fur texture and lighting on the capybara are highly detailed
• + Captures the bored expression of the passenger perfectly

• − The capybara's paws look slightly more like human hands/fingers in a capybara skin than natural paws

Wan 2.5 (Preview)

• + More colorful and vibrant background representing Manhattan night lights
• + Better rendering of the capybara's natural paws on the steering wheel
• + Good inclusion of the taxi roof sign and interior details

• − The capybara's face is a bit too smooth and symmetrical, losing some realism
• − The passenger's hands and phone interaction are slightly less natural than Model A

GPT Image 1.5

• + Excellent text integration and vintage typography that feels cohesive with the art style.
• + Superior adherence to the 'vintage dark parchment' aesthetic with a unified sepia-tone palette.
• + Atmospheric and spooky border that blends naturally into the composition.

• − The parchment texture is very heavy, slightly obscuring the clarity of the background graveyard.

Wan 2.5 (Preview)

• + High-quality rendering of the fire inside the jack-o-lantern.
• + Clean and legible event details at the bottom.
• + Good contrast between the blue night sky and the parchment border.

• − The text 'Halloween Party Invitation' has some slight layering artifacts on the 'H' and 'y'.
• − The border feels like a separate digital overlay rather than an integrated part of a 'vintage poster'.

GPT Image 1.5

• + Excellent texture and realistic hair density
• + Natural integration of the hairline with the existing forehead
• + Perfect preservation of the facial features, clothes, and lighting

• − The hair color is slightly cooler in tone than the beard

Wan 2.5 (Preview)

• + Successfully applied the requested edit
• + Preserved the composition and lighting well

• − The hairline transition on the forehead looks slightly artificial
• − The hair texture is a bit smooth and looks styled rather than natural hair growth
• − Small artifact in the hair strands above the forehead

GPT Image 1.5

• + Excellent adherence to the 'isometric miniature diorama' request with a detailed scene.
• + High-quality PBR material rendering on wood, ceramic, and food textures.
• + Perfect text layout including the flag icon centered as requested.

• − Slightly more complex than the 'minimal' request, though still very clean.
• − The background is not a perfectly solid blue, featuring a slight gradient.

Wan 2.5 (Preview)

• + Successfully captures the '3D cartoon' style with soft, rounded textures.
• + Very clean and minimal composition on a small raised base.
• + Accurate text rendering and placement.

• − Does not follow the 45-degree top-down isometric angle as well as Model A.
• − The background contains significant vignette/gradient rather than being solid.
• − Less detail in the sushi model compared to Model A.

GPT Image 1.5

• + Excellent caricature style that maintains a strong facial resemblance to the source image.
• + Rich, detailed background incorporating the news studio and hockey game seamlessly.
• + Creative integration of the prompt by putting one of the dogs in a hockey helmet.

• − Slightly crowded composition with many elements competing for attention.

Wan 2.5 (Preview)

• + Successfully incorporates all elements: news anchor, dogs, and hockey.
• + Preserves the subject's original denim shirt from the source image.
• + Clean, vector-like caricature style with high contrast.

• − The facial features lose the specific likeness of the subject in favor of a generic cartoon face.
• − The composition is a bit flat with a simple white background border.

GPT Image 1.5

• + Excellent fur texture details on all animals
• + Realistic interaction and tumbling poses
• + Natural integration of lighting and dew sparkles

• − The cat's anatomy is slightly messy with too many visible paw pads
• − God rays are a bit soft/blurry compared to the crispness of the animals

Wan 2.5 (Preview)

• + Dynamic sense of movement and 'chasing' as requested
• + Vibrant, distinct colors in the wildflower meadow
• + Consistent lighting across all subjects

• − The fox's eyes appear unnaturally blue and metallic
• − Dew drops are rendered as floating glass orbs instead of surface moisture
• − Fur texture looks somewhat flat or plastic compared to Model A

GPT Image 1.5

• + Excellent soft pastel color palette and warm, nostalgic mood.
• + Captures the 'painterly' texture and dreamy lighting requested.
• + Preserves the depth of field and composition of the source image well.

• − Faces are a bit generic and lose some of the specific character expressions of the source.
• − Overuse of lens flare/sparkle effects can be distracting.

Wan 2.5 (Preview)

• + Strong adherence to the Ghibli character design style with clean line art.
• + Perfectly captures the facial expressions of the original meme characters in an anime style.
• + High clarity and faithful preservation of the background elements and clothing patterns.

• − Lighting is a bit flat compared to the 'dreamy' request.
• − Color palette is slightly more desaturated/cool than 'warm pastel'.

GPT Image 1.5

• + Excellent adherence to the 'flying leaves' instruction with various sizes and depths
• + Modified the hair to show dramatic wind-blown movement
• + Maintained the overall composition and lighting of the source image

• − The leash handle has significant structural artifacts and looks broken
• − Some leaves look like they are floating static rather than moving due to lack of motion blur

Wan 2.5 (Preview)

• + Natural-looking hair movement that feels consistent with a breeze
• + Preserved the details of the woman's face and the dog very accurately
• + Clean preservation of the leash and park architecture

• − The flying leaves are sparse and look like flat lime-green stickers
• − Less 'energetic and lively' feel compared to the other model

GPT Image 1.5

• + Excellent typography with a hand-drawn feel
• + Strong vector-style shading and texture on the dome
• + Accurate rendering of the accent in 'Caffè'

• − Ignored the 'light background' instruction, delivering it on solid black
• − Text alignment on the bottom banner is slightly off-center

Wan 2.5 (Preview)

• + Perfectly followed the 'light background' and 'subtle texture' instructions
• + Clean, minimalist vector aesthetic that feels professional
• + Well-composed with a decorative border that adds to the vintage feel

• − The 'a' in 'Florian' is slightly malformed
• − The steam icon is a bit generic compared to the first image

GPT Image 1.5

• + Excellent text rendering for all panel titles.
• + Strict adherence to the requested grid-based infographic layout.
• + Consistent flat-vector style with a clean, cohesive NASA-inspired palette.

• − One minor spelling error in 'Tranquillity' (though both English spellings are technically valid).

Wan 2.5 (Preview)

• + Dynamic trajectory illustration linking the Earth and Moon.
• + Clean iconography for the astronaut profiles.
• + Follows the NASA-inspired color scheme well.

• − Confusing visual hierarchy where 'Descent' and 'Landing' text are floating without clear icons.
• − Inconsistent rocket designs; the ascent stage looks somewhat like a shuttle.
• − The trajectory lines are cluttered and cross over each other awkwardly.