GPT Image 2 OpenAI vs LongCat-Image Meituan

GPT Image 2

• + Excellent adherence to the 'plant behind the cube visible through the glass' instruction.
• + High detail on the texture of the red book's cover.
• + Realistic soft lighting and shadows that anchor the objects to the table.

• − The glass cube appears more like a frame or open box due to the lack of strong reflections on the front faces.

LongCat-Image

• + Beautiful glass material with realistic thickness and edge refraction.
• + The blue sphere is rendered with semi-translucency, adding to the visual appeal.
• + Correct positioning of all requested elements.

• − The plant is barely visible through the cube itself compared to the background.
• − The lighting on the cube edges is slightly inconsistent with the soft window light.

GPT Image 2

• + Excellent photographic realism with natural skin textures and plausible clothing.
• + Strong adherence to the 'imperfect framing' prompt with foreground objects obstructing the view.
• + Highly accurate anatomical details in the hands and facial features.

• − The 'red' color of the bicycle is slightly muted.
• − Motion blur on the background car is subtle compared to Model B.

LongCat-Image

• + Atmospheric lighting and clear reflections on the wet pavement.
• + Good sense of motion blur on the passing vehicle.
• + Strong color contrast with the vibrant red bicycle.

• − The bicycle geometry is nonsensical with multiple wheels and frames merging.
• − The rain is rendered as overly uniform digital streaks.
• − Features a more 'AI-stylized' appearance rather than a realistic 50mm photo.

GPT Image 2

• + Exceptional photographic realism in skin texture and eyes
• + Masterful lighting and subtle torchlight reflections on weather-beaten armor
• + Excellent use of depth of field and soft bokeh

• − The beads in the hair are very subtle and blend in with the texture
• − Composition is very close-cropped, hiding some of the armor details

LongCat-Image

• + Strong adherence to the 'beads' and 'leather straps' requirement
• + Vivid depiction of bokeh sparks and a clear torch source
• + Shows more of the ornate armor design

• − Skin and hair textures look significantly more synthetic and plastic-like
• − The scar on the face appears like a dark hole/artifact rather than lifelike skin damage
• − Overall lighting is harsh and lacks the tonal depth of the competitor

GPT Image 2

• + Exceptional text rendering with perfect spelling and legibility.
• + Clean, professional layout that strictly adheres to the requested grid and sections.
• + High-quality, appetizing food photography that looks realistic and consistent.

• − None notable; very strong adherence to the prompt.

LongCat-Image

• + Bold, vibrant use of color and large blocks of graphic design.
• + Interesting use of negative space in the layout.

• − Incoherent and gibberish text throughout the design.
• − Food photos are somewhat distorted and lack professional clarity.
• − Poor organizational flow compared to a functional menu.

GPT Image 2

• + Excellent adherence to the 'exploded' instruction with all components clearly separated and suspended.
• + Superior integration of text using the requested fiery, glowing effect throughout all copy.
• + Highly detailed food textures, especially the charred patty, sesame bun, and droplets of sauce.

• − The composition is very crowded, with the text overlapping some of the fiery background elements.

LongCat-Image

• + Clean and readable starburst graphic for the price point.
• + Good use of glowing coals at the bottom to create a sense of heat.

• − Failed the 'exploded' prompt as the burger remains mostly assembled.
• − The text effects are inconsistent, with the price starburst lacking the 'fiery, glowing' render requested.
• − Visual quality of the food looks more like plastic or CGI than the photorealistic textures in the competitor image.

GPT Image 2

• + Excellent text rendering with perfect spelling for all requested menu items.
• + Detailed chalk texture with realistic variation in stroke opacity and grain.
• + Consistent, elegant cursive handwriting that matches the 'cozy café' aesthetic.

• − The frame is slightly cropped at the bottom.
• − The background lighting is a bit dark, focusing primarily on the board.

LongCat-Image

• + Bright, well-composed café background with good depth of field.
• + Realistic chalk dust buildup at the bottom of the chalkboard frame.

• − Severe spelling errors and garbled text across every line.
• − The lettering looks like a digital brush stroke rather than natural chalk on a board.
• − Failed to follow the specific order and phrasing of the menu items provided in the prompt.

GPT Image 2

• + Excellent adherence to the specific 'horse on top' request
• + High visual fidelity and cinematic lighting on the spacesuit texture
• + Clever use of reins and saddle on the astronaut to enhance the theme

• − The horse's front legs look slightly unnatural where they meet the saddle

LongCat-Image

• + Dynamic composition and sense of movement
• + Correctly depicted a horse and astronaut in a space setting

• − Failed the negative constraint by putting the astronaut on top
• − Obvious AI artifacts like the strange planes in the sky and extra horse legs

GPT Image 2

• + Excellent cinematic lighting and photorealistic fur texture on the capybara.
• + Highly accurate adherence to the 'bored expression' of the businesswoman sitting in the back.
• + Perfect composition that emphasizes the 'inside the taxi' feeling requested by the prompt.

• − The capybara's paws are rendered as slightly amorphous shapes rather than defined claws/paws.

LongCat-Image

• + Vibrant colors and clear details on the taxi exterior and top light.
• + Good rendering of the capybara's claws on the steering wheel.
• + Successfully depicts the business attire and phone usage.

• − Included two passengers instead of the requested single businesswoman.
• − The capybara's head is disproportionately large compared to its body and the driver's seat.
• − The lighting feels a bit more like a composite than a single photorealistic scene.

GPT Image 2

• + Perfect text rendering for all lines including the address and date.
• + Superior vintage goth atmosphere with a cohesive parchment texture and intricate border.
• + Excellent composition that blends the background elements like the bridge and castle into the design.

• − The jack-o-lantern is slightly off-center, though it still feels balanced by the lanterns.

LongCat-Image

• + Good adherence to the requested elements like thorns and twisted trees.
• + Clear and legible title text and scroll banner.

• − Failed to render the address correctly, writing 'The Armiees' instead of 'The Arches'.
• − The composition feels disjointed with a cutout window effect rather than a cohesive poster.
• − Visual style is a bit more like a modern clip-art collage than a 'vintage gothic' invitation.

GPT Image 2

• + Excellent text rendering with clean 3D effects
• + Highly detailed and realistic textures on the sushi fish
• + Well-executed isometric diorama base with 3D depth

• − The scene is a bit cluttered, partially ignoring the 'minimal garnish' request

LongCat-Image

• + Perfect 3D cartoon style with soft, clay-like textures
• + Follows the 'minimal' requirement very well
• + Clean flag icon and bold text rendering

• − The text is 2D and lacks the 3D 'pop' seen in the other model
• − The sushi rice looks somewhat like plastic beads rather than food

GPT Image 2

• + Excellent adherence to the prompt, successfully including all four specific animals.
• + Superior fur textures and realistic interaction between the subjects and the environment.
• + Effective use of 'god rays' and lighting to create a cohesive, masterpiece-quality scene.

• − The fox kit has one paw that looks slightly distorted.
• − The butterfly in the center is positioned somewhat awkwardly near the dog's head.

LongCat-Image

• + Bright, cheerful colors with very large, expressive eyes that fit the 'cute' theme.
• + Clean rendering of the golden retriever puppy's face.

• − Failed to include a separate baby bunny, instead merging ears onto the kitten to create a hybrid creature.
• − The 'dew sparkles' look like floating digital blobs rather than actual morning dew.
• − The fox's tail is positioned oddly, appearing to sprout from its upper back/neck area.

GPT Image 2

• + Excellent typography with perfect spelling and accent marks.
• + Sophisticated woodcut-style engraving details on the cloche and banner.
• + Balanced professional composition within a classic frame.

• − The 'FLORIAN' text is slightly larger than a standard minimalist logo might dictate.

LongCat-Image

• + Strong 'vintage' paper texture in the background.
• + Effective use of the cloche as a central framing element for the text.

• − Poor typography with repetitive words and messy letter rendering.
• − Low visual quality with blurred lines and messy steam trails.
• − Lacks the minimalist elegance requested, appearing more like a cluttered sketch.

GPT Image 2

• + Excellent typography with perfectly rendered, legible text for all mission stages and crew names.
• + Strict adherence to the 6-step chronological sequence requested in the prompt.
• + Professional graphic design layout with clean lines, consistent iconography, and a clear NASA-inspired color palette.

• − Includes a few photographic textures (like the moon surface) which deviates slightly from a purely 'flat-vector' style.
• − The landing module icon appears slightly different in style compared to the sleek Saturn V rocket.

LongCat-Image

• + Captures a simplified flat-vector aesthetic with clean shapes.
• + Uses the requested color palette of navy, white, and muted red effectively.

• − Failed to include the specific 6-step sequence, resulting in nonsensical layout and missing stages.
• − Text rendering is poor with significant misspellings and gibberish characters.
• − Visual icons are repetitive and do not accurately represent the requested mission phases.