GPT Image 1.5 OpenAI vs Vidu Q2 ShengShu Technology

GPT Image 1.5

• + Excellent photographic realism and natural soft lighting
• + Accurately represents the transparency of the glass cube
• + Clean and simple composition that highlights all prompt elements

• − The sphere is slightly larger than 'small' would imply
• − The top of the cube appears to be open or missing under the book

Vidu Q2

• + High level of detail in the wooden table texture and plant leaves
• + Good handling of complex shadows and caustics
• + The sphere is a smaller, more 'small' size as requested

• − The book is hovering awkwardly above the cube with a noticeable gap
• − Perspective on the cube's base feels slightly distorted

GPT Image 1.5

• + Excellent preservation of the man's facial features and specific hairstyle
• + The lighting on the man and inside the car realistically matches the outdoor coastal environment
• + The cinematic perspective feels more like a passenger taking a photo.

• − The car is a right-hand drive model in California
• − The steering wheel and man's hand look slightly unnatural and low-detail

Vidu Q2

• + Successfully shows almost the entire car from the source image
• + High-quality rendering of the California coastline and winding road
• + Correct placement for a driver in a US-based setting (left-hand drive)

• − The man's facial details and distinctive hair are noticeably changed from the source
• − The car's proportions feel slightly stretched compared to the reference
• − Poor hand/steering wheel integration

GPT Image 1.5

• + Excellent depiction of raining conditions with visible droplets and wet textures.
• + Coherent and logical bicycle anatomy including the chain and cassette.
• + Authentic skin textures and believable 'candid' crouching posture.

• − Missing the requested motion blur on the passing car.
• − The composition feels a bit too centered and clean for 'imperfect framing'.

Vidu Q2

• + Effective shallow depth of field and beautiful ground reflections.
• + Strong adherence to the 'imperfect framing' prompt with a tight, cropped composition.

• − Significant anatomical errors including three hands and a disjointed bicycle frame.
• − Poorly rendered bicycle mechanics with the chain appearing as a disconnected wire.
• − Face lacks the same level of natural skin texture found in Model A.

GPT Image 1.5

• + Excellent depiction of cinematic torchlight and bokeh sparks
• + Highly realistic skin texture with lifelike dirt and scarring
• + Exceptional detail on the engraved armor and weathered leather straps

• − The braiding with beads is a bit messy and lacks clear definition in some strands

Vidu Q2

• + Clear and distinct hair braids with visible beads
• + Strong, clean composition with balanced lighting
• + Good interpretation of the cloth underlayer texture

• − The skin and armor look slightly too clean for the 'battle-worn' description
• − The lighting feels more studio-like rather than atmospheric torchlight
• − Less detailed texture on the leather straps compared to Image A

GPT Image 1.5

• + Perfect text rendering with zero spelling errors and logically categorized sections.
• + High-quality food photography that looks professional and appetizing.
• + Clean, modern layout that adheres strictly to the minimalist casual dining prompt.

• − Slightly standard or safe aesthetic that lacks a unique artistic flair.

Vidu Q2

• + Vibrant color palette and energetic layout that feels youthful.
• + Creative use of icons and graphic accents.

• − Unreadable gibberish text throughout the entire design.
• − Food images are cluttered and lack the clarity of professional food photography.
• − Disorganized layout that fails to clearly separate the requested categories.

GPT Image 1.5

• + Excellent adherence to all text requirements including the specific currency symbol.
• + Superior texture rendering on the patty and bun for a photorealistic food-photography look.
• + Highly dynamic composition with embers and flying debris that enhances the 'exploded' theme.

• − The composition is a bit crowded near the top text.

Vidu Q2

• + Clean layout with well-defined fire elements in the background.
• + Good use of the 'fiery, glowing effect' on the main title text.

• − Failed to render the Euro symbol correctly, using a stylized hash instead.
• − The food textures appear somewhat plastic or artificial compared to the other model.
• − The explosion of ingredients feels less 'dynamic' and more like a simple vertical hover.

GPT Image 1.5

• + Excellent text rendering with perfect spelling across all items.
• + Highly realistic chalk texture with dusty smudges and authentic line weights.
• + Followed the truncated prompt for the third item by logically completing it as 'Brown Butter Chocolate Chip Cookies'.

• − The composition is a bit tight at the top and bottom edges.

Vidu Q2

• + Dynamic chalk lettering with high contrast and artistic flair.
• + Good background bokeh that suggests a café setting.

• − Extremely poor spelling and legibility, with several words becoming gibberish.
• − Inconsistent prices and symbol rendering.
• − Messy composition with overlapping text and chaotic lines at the bottom.

GPT Image 1.5

• + Excellent preservation of facial identity and accessories like the sunglasses and scarf.
• + Correctly matched the red stool and yellow background environment.
• + Smooth integration of the clothing details onto the new pose.

• − The character is wearing pants which contradicts the source pose containing bare legs.
• − The anatomy of the right arm and hand is slightly stiff compared to the reference.

Vidu Q2

• + Successfully captured the bare legs from the reference pose while maintaining the character's upper body.
• + Highly accurate recreation of the dynamic hand and finger positions from Image 1.
• + Excellent lighting and shadow work on the yellow background.

• − The facial features are slightly distorted and lose some of the specific identity of the man in Image 2.
• − Structural issues with the right hand showing an extra finger/atypical anatomy.

GPT Image 1.5

• + Excellent cinematic lighting and texture on the horse's coat.
• + Detailed space environment with diverse elements like planets and a lunar lander.
• + High degree of realism in the mechanical parts of the spacesuit.

• − The prompt specifically asked for 'horse on top', which this model failed to interpret, placing the astronaut on top instead.
• − The composition feels a bit crowded with many competing background elements.

Vidu Q2

• + Successfully captured a more surreal aesthetic as requested in the prompt.
• + Vibrant color palette that blends the horse and the cosmic background artistically.
• + Cleaner focus on the central subject without distracting background props.

• − Failed the negative constraint/instruction 'horse on top, not vice versa'.
• − Anatomically awkward front legs on the horse.

GPT Image 1.5

• + Excellent preservation of the original wooden structure on the left
• + Realistic fabric texture and drape of the coat and scarf
• + Successfully adapted the pose from Image 2 onto the character's lower half

• − Failed to preserve the head and face of the person from Image 1
• − Did not include the sunglasses from Image 2
• − The vitigilo patterns on the chin and hands do not match the source Image 1

Vidu Q2

• + Keeps the person's exact face, hair, and vitiligo patterns visible
• + Includes the sunglasses and jewelry from Image 2
• + Maintains the background beach environment accurately

• − The coat is rendered with strange sand-like stains on the sleeve that weren't requested
• − The pose is slightly stiff compared to the reference images
• − Minor distortion on the hand and watch detail

GPT Image 1.5

• + Excellent photorealism with shallow depth of field typical of high-end photography
• + The capybara's paws are placed correctly on the steering wheel with realistic texture
• + The lighting inside the cabin feels organic and well-integrated

• − The composition is a bit tight, cutting off much of the vehicle interior
• − The passenger's face is slightly blurry

Vidu Q2

• + Dynamic composition showing more of the car's interior and the city environment
• + Perfect adherence to the bored expression requested for the passenger
• + Clear and sharp details throughout the entire frame

• − The capybara's paws look more like long human fingers, which is anatomically incorrect
• − The lighting feels slightly more artificial or 'rendered' compared to Image A

GPT Image 1.5

• + Perfect text rendering for all requested details, including date and location.
• + Atmospheric and cohesive vintage gothic aesthetic with a dark parchment feel.
• + Strong composition where every element like the thorns, webs, and moon feels integrated.

• − The dark, moody lighting makes the 'Location' text at the bottom slightly harder to read against the dark background.

Vidu Q2

• + Bright, clear parchment texture that provides high contrast for the central items.
• + Good variety in the thickness of the thorny border.

• − Extreme spelling errors in almost every text field, including 'Intovztion' and 'invieed'.
• − Hallucinated incorrect date information and failed to separate time from location.
• − Art style feels more like a digital illustration than a 'vintage' gothic poster.

GPT Image 1.5

• + Excellent texture matching the beard
• + Highly realistic hairline and integration with sideburns
• + Near-perfect preservation of facial structures and background details

• − Slightly changes the bridge of the glasses where it meets the nose

Vidu Q2

• + Successfully adds a full head of hair
• + Maintains original glasses and facial features well

• − The hair texture appears overly smooth and lacks the coarse realism of the existing beard
• − The hairline looks artificial and slightly floating
• − Slight distortions in the background lighting around the hair edges

GPT Image 1.5

• + Excellent text rendering and layout.
• + High-detail PBR materials with realistic wood and ceramic textures.
• + Perfect adherence to the 45-degree isometric diorama request.

• − The scene is slightly less 'minimal' than requested due to the teapot and soy sauce bottle.

Vidu Q2

• + Matches the 'cartoon' aesthetic well with soft colors.
• + Clean, minimal composition.
• + Includes the flag icon as requested.

• − Text rendering is slightly less crisp than the competitor.
• − The 3D modeling of the sushi is somewhat mushy and lacks the clarity of high-end PBR materials.
• − The diorama base is very simple compared to the detailed scene in the other image.

GPT Image 1.5

• + Expertly captures the news anchor aesthetic with studio lighting, cameras, and a ticker.
• + Strong caricature style with exaggerated features that still maintain the subject's likeness.
• + Very high visual quality and creative integration of the dog wearing a hockey helmet.

• − Changes the subject's outfit from the original denim to a red dress.
• − The hockey player in the background feels a bit disconnected from the main scene compared to the other elements.

Vidu Q2

• + Excellent source preservation, maintaining the denim jacket and black shirt from the original image.
• + Creative integration of hockey elements, such as the puck on the desk and the rink backdrop.
• + Includes clear 'News' text and a professional microphone.

• − The hands are poorly rendered with extra or distorted fingers.
• − The caricature of the face is less distinctive than model A, leaning more towards a generic cartoon style.

GPT Image 1.5

• + Excellent fur texture rendering and realistic dew sparkles on the grass.
• + Emotional resonance through expressive, high-detail eyes and adorable poses.
• + Superior lighting with beautiful god rays and a consistent golden hour glow.

• − The fox's paw in the bottom right corner is slightly malformed or merged into the grass.

Vidu Q2

• + Stronger sense of 'chasing' and 'tumbling' movement throughout the scene.
• + Effective use of vertical space with butterflies spread across the entire frame.

• − Includes two golden retriever puppies instead of the requested one.
• − Noticeable artifacts such as the kitten missing hind legs and the rabbit having an extra tail/leg.
• − Lower photorealistic quality with a more digital, over-saturated illustrative feel.

GPT Image 1.5

• + Excellent adherence to the 'soft pastel' and 'warm, nostalgic mood' requirements.
• + Successfully captures the dreamy, painterly background texture associated with Ghibli films.
• + Translates facial expressions into a charming anime style that fits the theme.

• − The excessive glow and high saturation pull it slightly toward a generic shoujo manga style rather than pure Ghibli realism.
• − The foreground character is quite blurry, which loses some structural detail.

Vidu Q2

• + Stronger preservation of the original source image's geometry and composition.
• + Clean line work and color palette that feels very close to Studio Ghibli's character cel animation style.
• + Preserves subtle details like the plaid pattern on the shirt accurately while stylizing them.

• − The background is a bit more generic and less 'dreamy' compared to Model A.
• − Lighting is relatively flat compared to the 'gentle lighting' requested in the prompt.

GPT Image 1.5

• + Excellent depiction of wind-blown hair with high hair-strand detail.
• + Adds a large quantity of leaves varied in size and motion blur.
• + Maintains original identity and facial features perfectly.

• − The leaf placement looks a bit cluttered across the subject's body.
• − Motion blur on the leaves is inconsistent.

Vidu Q2

• + Natural integration of leaves with a better sense of depth.
• + Very high fidelity preservation of the original source image colors and lighting.
• + Clean hair movement that feels realistic without being overly chaotic.

• − The hair on the subject's right side (viewer's left) appears slightly more static than the other side.
• − Fewer leaves compared to Model A, making the effect slightly more subtle.

GPT Image 1.5

• + Excellent text rendering with perfect spelling of 'Caffè Florian' and 'Est. 1720'.
• + Strong vector-style illustration with sophisticated shading and texture.
• + Correct application of the requested banner element and classic typography.

• − Failed the request for a 'light background' by using a solid black background.
• − The steam element is a bit thick and stylized compared to the rest of the realism in the cloche.

Vidu Q2

• + Followed the background instruction by using a light, textured cream background.
• + The color palette adheres well to the 'warm brown and cream' request.

• − Significant spelling errors in every piece of text including 'Farmiin' and 'Esttt'.
• − The composition is cluttered with redundant text lines.
• − The cloche handle and steam lines are poorly defined and lack professional finish.

GPT Image 1.5

• + Excellent text rendering with no spelling errors.
• + Perfect adherence to all six requested steps and iconography.
• + Clean, professional flat-vector aesthetic with a cohesive color palette.

• − The rocket in the first panel is partially cut off at the top.
• − Composition is slightly crowded toward the bottom panels.

Vidu Q2

• + Successfully captured the requested color palette.
• + Good use of white space for a modern clean look.
• + Accurate icons for the Earth and Moon orbit steps.

• − Severe spelling and gibberish text throughout the entire image.
• − Inconsistent number of astronauts (five instead of three).
• − The 'Saturn V' icon is a generic cartoon rocket shuttle, not a heavy lift launch vehicle.