GPT Image 2 OpenAI vs Vidu Q2 ShengShu Technology

GPT Image 2

• + Excellent photographic realism and soft, natural lighting.
• + Correct placement and scale of all objects relative to each other.
• + Highly detailed textures on the book cover and the wooden table.

• − The plant is slightly more behind the cube than visible through it compared to Model B.

Vidu Q2

• + Strong adherence to the spatial requirements of the prompt.
• + Interesting play of light and shadow on the table surface.
• + Captures the refraction of the plant through the glass well.

• − The glass cube has illogical reflections, such as a second blue sphere appearing on its base.
• − The book has some minor distortion on the spine text and edges.

GPT Image 2

• + Excellent handling of motion blur on passing cars as requested.
• + Authentic street photography composition with 'imperfect framing' that feels natural.
• + Superior skin textures and realistic environmental details like the toolbox and Japanese signage.

• − The light rain is very subtle and barely visible.
• − Minor logical issue with the man's seat appearing to hover over a paint bucket.

Vidu Q2

• + Stronger emphasis on wet pavement reflections.
• + Dynamic close-up framing that highlights the textures of the bike.

• − Total failure on motion blur for the car in the background.
• − Significant anatomical and structural issues with the bike (the chain and frame are nonsensical).
• − Over-sharpened, high-contrast look that contradicts the 'no stylization' instruction.

GPT Image 2

• + Exceptional photorealistic skin texture and lifelike eyes.
• + Masterful handling of lighting, with soft torchlight glints and realistic metal weathering.
• + Superior hair detailing, including the small beads and intricate braiding requested.

• − The 'battle-worn' aspect is slightly more subtle than expected, appearing more like surface dirt than heavy scarring.

Vidu Q2

• + Strong adherence to the armor engraving and leather strap details.
• + Effective use of bokeh sparks in the background to create atmosphere.
• + Clearly visible facial scarring that aligns with the 'battle-worn' prompt.

• − The skin and hair have a slightly plastic, CGI-like finish compared to Model A.
• − Lighting on the face feels a bit flat and less integrated with the background environment.

GPT Image 2

• + Perfect text rendering with coherent menu items and descriptions
• + Professional, realistic food photography that matches the labels
• + Excellent layout with clear sections for Appetizers, Pizza, and Mains as requested

• − Slightly busy header with the logo and side text, though typical for the industry

Vidu Q2

• + Clean pastel color palette with vibrant accents
• + Solid attempt at a grid-based layout

• − Text is completely illegible gibberish
• − Food images are messy and don't match the pizza category
• − Failed to properly delineate the requested sections logically

GPT Image 2

• + Excellent photorealistic texture on the meat patty, bun, and vegetables.
• + Perfect execution of the 'fiery, glowing effect' for all requested text.
• + Dynamic composition with realistic sauce splashes and ember particles.

• − The '6' in the price has an unusual, slightly stylized shape compared to standard fonts.

Vidu Q2

• + Clean layout with the title centered at the top.
• + Good rendering of fresh lettuce and tomato slices.
• + Includes all requested text elements in the specified hierarchy.

• − The currency symbol is incorrect, showing a blurred hybrid mark instead of the Euro sign.
• − The fiery background feels more like a flat texture than a dynamic environment.
• − Lower level of detail in the food textures compared to the competitor.

GPT Image 2

• + Excellent text accuracy, spelling every word correctly including the complex third item.
• + Perfectly captures the authentic texture of dry chalk on a blackboard.
• + Maintains consistent handwriting style across all lines while following layout instructions.

• − The cursive for the title is relatively simple rather than highly 'elegant' cursive.

Vidu Q2

• + Dynamic chalk texture with varied thickness and pressure.
• + Bold, legible title design with good contrast.

• − Significant spelling errors throughout the menu items such as 'Truffe Musshoom' and 'Octopd'.
• − The text for the third item becomes gibberish ('Browd Botter Chotoouttey').
• − Mismatched price for the first item compared to the prompt ($34 vs $24).

GPT Image 2

• + Excellent adherence to the complex crossed-leg pose from Image 1.
• + High character consistency, successfully transferring the face, hair, sunglasses, and scarf from Image 2.
• + Matches the atmospheric lighting and soft shadows of the original scene perfectly.

• − The right hand (upper) has an extra finger and some anatomical blurring at the fingertips.
• − The left foot on the box has slightly distorted toe anatomy.

Vidu Q2

• + Great facial similarity to the character in Image 2.
• + Captures the scarf and clothing style effectively while maintaining the sharp lighting of the yellow background.
• + The high-resolution rendering makes the textures of the clothes and skin very clear.

• − The left hand (lower) has significant anatomical issues, appearing with six fingers and an unnatural palm shape.
• − The pose is slightly less accurate to Image 1's leg positioning, losing the 'tightness' of the wrap.
• − The fingers on the right hand (upper) are strangely elongated and distorted.

GPT Image 2

• + Excellent adherence to the specific prompt instruction to have the horse on top
• + High realism and detail in the texture of the space suit and horse fur
• + Convincing cinematic lighting and lunar environment background

• − The horse's front legs are morphing into the harness/reins in a confusing way

Vidu Q2

• + Beautiful vibrant colors and cosmic aesthetic
• + Dynamic and graceful horse pose
• + High visual appeal in the nebula and galaxy background

• − Completely failed the negative constraint to have the horse on top
• − Followed the traditional 'astronaut on horse' trope instead of the surreal instruction

GPT Image 2

• + Excellently preserves the base person's face, hair, and vitiligo patterns without distortion.
• + Accurately replicates the scarf pattern and coat style from Image 2.
• + Maintains high source preservation for the background and facial expression.

• − The lighting on the person's face feels slightly flattened compared to the original image.
• − Missed the sunglasses and the specific gold ring accessory from Image 2.

Vidu Q2

• + Successfully includes the sunglasses and gold watch/ring accessories from Image 2.
• + Creates a high-contrast, sharp image that matches the fashion aesthetic.

• − Failed to keep the person's face unchanged, significantly altering the facial structure and adding a mustache.
• − Introduced severe artifacts where the vitiligo pattern is mistakenly applied on top of the coat's fabric.
• − Altered the person's hair and modified the background significantly.

GPT Image 2

• + Excellent photorealism with shallow depth of field and realistic lighting.
• + Very high detail on the capybara's fur and the apparel textures.
• + Captures the 'bored' expression of the passenger perfectly.

• − The composition is quite tight, making it a bit harder to see the full taxi interior.

Vidu Q2

• + Clearer view of the overall taxi interior and the street scene through the windows.
• + Accurate adherence to all major prompt elements, including the steering wheel placement.
• + Great use of vibrant city lights to establish the New York night atmosphere.

• − Lighting on the capybara's head is a bit flat and looks slightly composited.
• − Passenger's hands and phone interaction have minor anatomical warping.

GPT Image 2

• + Perfect text rendering for all requested details, including date and location.
• + High-quality gothic aesthetic with intricate details like thorns, spiderwebs, and lanterns.
• + Creative integration of 'The Arches' and a NYC-style skyline into the background.

• − None notable; the image follows all prompt instructions perfectly.

Vidu Q2

• + Strong parchment effect and central jack-o-lantern glowing effect.
• + Includes requested bats and twisted tree branches.

• − Severe spelling errors in the title, banner text, and event details.
• − Incorrect date (2025 instead of 2026) and messy character rendering (30.70).
• − Lower overall visual quality with some muddy textures in the background.

GPT Image 2

• + Excellent PBR materials with realistic textures on the fish and rice.
• + Sophisticated 3D diorama base with detailed modeling and lighting.
• + Flawless text rendering and icon placement.

• − The scene is slightly more complex than 'minimal' with the added stone lantern and foliage.

Vidu Q2

• + Successfully captures the requested 'cartoon' aesthetic with soft, rounded shapes.
• + Adheres to the diorama and plate requirement.
• + Correct text content and placement.

• − The sushi details are slightly messy and less defined compared to A.
• − The flag icon is floating awkwardly on a pole from the letter 'N'.
• − Lower overall resolution and clarity in textures.

GPT Image 2

• + Excellent fur texture and realistic lighting that catches individual hairs.
• + Dynamic and coherent composition with all animals interacting naturally.
• + Perfectly captures individual animal species features like the fennec-like ears of the fox and the golden retriever's expressive face.

• − The kitten's paw reaching up has a slightly clumped claw structure.

Vidu Q2

• + Includes a wide variety of colorful butterflies and a very lush flower field.
• + Bright, energetic color palette that fits the 'joyful' prompt requirement.

• − The fox looks more like a small dog or a fox-puppy hybrid rather than a distinct fox kit.
• − Duplicated a golden retriever puppy, which results in five animals instead of the four requested.
• − Anatomical issues with the rabbit's body shape and the tails of the puppies.

GPT Image 2

• + Excellent typography with perfect spelling of 'Caffè Florian' and 'Est. 1720'.
• + Superb vector emblem style with sophisticated line work and shading.
• + High-quality subtle paper texture and well-balanced composition.

• − The detail level is quite high for a 'minimalist' request, bordering on ornate.

Vidu Q2

• + Features the requested cloche and steam elements.
• + Uses the correct warm brown and cream color palette.

• − Failed significantly on text rendering with multiple misspellings like 'Farmiin' and 'Fopli20'.
• − The steam creates a strange, inconsistent visual inside the dome.
• − The composition is bottom-heavy and lacks the professional finish of a vector logo.

GPT Image 2

• + Excellent typography and nearly perfect spelling throughout the graphic.
• + Strict adherence to the requested NASA-inspired color palette and layout.
• + High-quality, detailed icons that accurately reflect the specific mission stages requested.

• − The style leans more toward a detailed illustration than a strictly flat vector style.
• − Small graphical glitch on the landing module legs in step 6.

Vidu Q2

• + Successfully captures a clean 'flat vector' aesthetic.
• + Included the requested palette of light gray, blue, and red.

• − Extreme spelling errors making the text unreadable (e.g., 'ALFONCH', 'Laup 1').
• − Failed to follow the requested 6-step chronological sequence accurately.
• − Generic icons lack the specific Saturn V and mission-specific details requested.