GPT Image 1 Mini OpenAI vs GPT Image 1.5 OpenAI

GPT Image 1 Mini

• + Excellent photorealism and textures, especially on the book's paper and cover.
• + Creates a very clean, aesthetically pleasing composition with soft, cinematic lighting.

• − The plant is placed to the side rather than behind the cube as requested.
• − The sphere appears to be floating inside the cube, which feels less physically grounded.

GPT Image 1.5

• + Strictly follows the spatial instruction of placing the plant behind the cube.
• + Handles complex glass reflections and refractions accurately, including the reflection of the sphere on the bottom pane.
• + Better adherence to the 'small blue sphere' prompt as the sphere is more in proportion to the cube.

• − The lighting is a bit harsher and less 'soft' than requested.
• − The textures on the desk and book are slightly less refined compared to the competitor.

GPT Image 1 Mini

• + Excellent preservation of the man's identity, hair, and clothing.
• + The car details, including the white leather and hood style, match the source image well.
• + Beautiful warm lighting and convincing motion blur on the road.

• − The steering wheel looks a bit thick and rubbery compared to the luxury car interior.

GPT Image 1.5

• + Successfully places the man and car in a California coastal setting.
• + Good facial likeness and hair texture compared to the source image.

• − The man is placed in the right-hand seat for a US driving context, making him the passenger, not the driver.
• − The steering wheel is on the left side and appears to be steered by 'ghost' hands or poorly positioned arms.

GPT Image 1 Mini

• + Excellent shallow depth of field creates a cinematic look.
• + Beautiful reflections on the wet pavement.
• + Natural skin texture and realistic expression on the subject.

• − The structural anatomy of the bicycle is slightly nonsensical with overlapping frames and spokes.
• − Lacks the requested motion blur from passing cars.

GPT Image 1.5

• + Successfully includes the requested motion blur from a passing car in the background.
• + Strong storytelling with the addition of tools and a more complex street environment.
• + Very realistic lighting and reflection details on both the ground and the car.

• − The composition feels a bit cramped with the crop cutting off the front wheel.
• − The face has slightly softer detail compared to the first model.

GPT Image 1 Mini

• + Excellent rendering of the ornate engraved plate armor with subtle, realistic weathering.
• + Very realistic skin texture and facial anatomy with lifelike eyes.
• + Effective use of shallow depth of field and soft bokeh, creating a painterly cinematic feel.

• − The hair beads requested in the prompt are barely visible or missing in the braids.
• − The lighting is slightly flat compared to the dramatic torchlight requested.

GPT Image 1.5

• + Highly detailed skin texture showing sweat, dirt, and realistic scarring as requested.
• + Includes clear details of small beads in the hair braids and a visible leather strap texture.
• + More dynamic and vibrant torchlight reflections on the metal armor.

• − The bokeh sparks in the background are somewhat distracting and less natural-looking.
• − The armor engraving is slightly less delicate than in the other model.

GPT Image 1 Mini

• + Excellent typography rendering with crisp, bold sans-serif fonts
• + Perfect adherence to the requested grid layout for photos
• + Clean and minimalist composition with high-contrast lines

• − The menu body is empty of actual dishes or text descriptions
• − Lacks functional details like prices or descriptions requested for a professional layout

GPT Image 1.5

• + Provides a complete, functional menu with dish names, descriptions, and prices
• + Includes vibrant colorful accents and high-quality food photography
• + Effective use of white space and legible typography for a casual dining feel

• − The photo alignment for the 'Mains' section is slightly off compared to the top grids
• − The fonts are less 'bold' and modern than Image A

GPT Image 1 Mini

• + Extremely clean and legible text rendering
• + Clear, organized exploded view of the burger
• + Strong adherence to the 'glowing embers' and fiery text effects

• − Static composition feels less 'dynamic' than model B
• − Background is somewhat sparse compared to the prompt's request for a fiery atmosphere

GPT Image 1.5

• + High energy and dynamic motion with flying particles
• + Impressive photorealistic textures on the patty and vegetables
• + Very rich, fiery background that enhances the magic theme

• − Text is slightly less integrated into the 'glowing' style compared to image A
• − Composition is a bit cluttered, making individual ingredients harder to distinguish

GPT Image 1 Mini

• + Excellent text legibility and spelling accuracy
• + Consistent letter spacing and alignment
• + Realistic chalk grain texture on the lettering

• − The title is in all-caps rather than the requested elegant cursive
• − The handwriting style looks a bit too perfect, bordering on a digital font look
• − Layout is slightly cramped for the bottom items

GPT Image 1.5

• + Successfully rendered the title in elegant cursive script
• + The handwriting has natural variations in tilt and size that look more authentically human
• + Captures a more realistic atmosphere with chalk dust and smudges on the board

• − Slightly less crisp text rendered at the bottom compared to the top
• − The chalk texture is very heavy, which can slightly obscure the finer lines of the text

GPT Image 1 Mini

• + Successfully replicates the yellow background and red ottoman from the source.
• + Captures the general facial hair and sunglasses of the character.

• − Fails the complex leg-crossing pose from Image 1, depicting only one foot on the stool.
• − Significantly alters the character's facial features and hair texture compared to Image 2.
• − Omits the specific floral patterns on the scarf.

GPT Image 1.5

• + Almost perfectly replicates the intricate crossed-leg pose from Image 1.
• + Excellent character consistency, including the facial structure, hair texture, and even the silver bracelet.
• + Accurately recreates the specific floral patterns and fringe on the scarf from the character reference.

• − The fingers on the hand at the top right are slightly distorted.
• − Small anatomy artifact where the left leg meets the stool.

GPT Image 1 Mini

• + Excellent tonal consistency and cinematic lighting.
• + Clean, uncluttered composition that emphasizes the 'surreal' aspect of space.

• − Failed the negative constraint to put the horse on top of the astronaut.
• − Anatomically awkward back legs on the horse.

GPT Image 1.5

• + High level of intricate detail on the spacesuit and lunar module.
• + Dynamic sense of motion with the kicked-up moon dust.

• − Failed the negative constraint to put the horse on top of the astronaut.
• − Composition is quite busy with multiple celestial bodies and textures competing for attention.

GPT Image 1 Mini

• + Excellent full-body composition including shoes
• + Accurately recreates the pattern and texture of the scarf
• + Matches the vitiligo patterns on the hands realistically

• − Facial features and hair significantly changed from Image 1
• − Lighting on the face is much harsher than the source image

GPT Image 1.5

• + Near-perfect preservation of the face, hair, and sand on the cheek
• + Includes the sunglasses and jewelry from Image 2
• + Excellent lighting and shadow integration on the person

• − Crops the original image into a square, losing the lower legs/shoes
• − The scarf pattern is slightly less accurate than Model A

GPT Image 1 Mini

• + Excellent soft lighting and cinematic bokeh in the background
• + Very realistic fur texture and clothing fabric details

• − Only one paw is visible on the steering wheel instead of both as requested
• − The composition is a bit tight on the capybara's face, cropping the background actress significantly

GPT Image 1.5

• + Perfect adherence to the instruction for both paws on the steering wheel
• + Stronger composition showing more of the taxi interior and the passenger
• + Includes clear 'TAXI' text on the cap as a nice touch

• − The passenger's facial features and hands are slightly less defined/more blurred than in Image A

GPT Image 1 Mini

• + Extremely clear and legible text layout.
• + Clean graphic design style with professional spacing.
• + Perfect adherence to the banner and title requirements.

• − Simple, flat texture on the background.
• − Border detail is relatively subtle and dark.

GPT Image 1.5

• + Excellent 'vintage parchment' texture and atmospheric depth.
• + High-quality ornamental border featuring prominent thorns and webs.
• + Includes extra thematic details like the cemetery and moon that enhance the mood.

• − Text layout is a bit more cluttered than Model A.
• − Minor artifact on the trailing edge of the scroll banner.

GPT Image 1 Mini

• + Successfully added a thick head of hair as requested.
• + Maintains the general color scheme of the jacket and background.

• − Significantly alters the person's face, making them look younger and losing the original identity.
• − The hairline and forehead area look digitally smoothed and unnatural.
• − The background is noticeably simplified compared to the texture of the original.

GPT Image 1.5

• + Excellent source preservation, maintaining the exact facial features and identity of the man.
• + Hair texture and curls look realistic and consistent with the beard.
• + The background and jacket details are perfectly preserved from the original image.

• − The transition from the hair to the forehead has a very slight sharpness that could be softer.

GPT Image 1 Mini

• + Excellent adherence to the 'minimal' requirement
• + Clean and professional typography layout
• + Consistent soft, 3D cartoon textures

• − Simple composition might feel sparse compared to a diorama
• − Slight lack of realistic PBR material variety

GPT Image 1.5

• + Strong execution of the 'raised diorama base' with layered materials
• + Highly detailed PBR textures for the wood, ceramic, and liquids
• + Excellent 3D miniature feel with complex modeling

• − Failed the 'minimal garnish' instruction by including a teapot, tea cup, and soy sauce
• − The 'SUSHI' text is smaller and less bold than Image A

GPT Image 1 Mini

• + Excellent caricature style with exaggerated features like the wide smile and large eyes.
• + Captures the denim shirt from the original image for better character continuity.
• + Cleverly includes all requested elements: anchor desk, dog, and hockey equipment.

• − The hand holding the microphone has an anatomical error with merging fingers.
• − The background is relatively flat and less detailed compared to Model B.

GPT Image 1.5

• + Highly immersive scene incorporating a live hockey game on the broadcast screen.
• + Humorous detail including a second dog wearing a hockey helmet.
• + Maintains very high visual quality and complex lighting for a studio environment.

• − The caricature style is slightly less 'exaggerated' in the face than Model A, leaning closer to a standard cartoon portrait.
• − The microphone hand is quite small and lacks detail.

GPT Image 1 Mini

• + Excellent depiction of motion with all four animals 'running' or jumping
• + Very clean composition with distinct, well-defined animals
• + Great adherence to the request for god rays and golden light

• − The fox's front right paw is slightly unnatural in posture
• − The golden retriever's back legs are omitted/hidden in a way that looks slightly disconnected

GPT Image 1.5

• + Expertly captures the 'tumbling together' aspect of the prompt
• + More intricate fur textures and highly expressive facial features
• + Superior lighting effects with beautiful dew sparkles and a visible sunburst

• − The fox has too many visible claws/toes on its right paw
• − The cat's anatomy is slightly distorted due to the 'tumbling' pose

GPT Image 1 Mini

• + Excellent preservation of the original poses and expressions.
• + Strong hand-painted texture consistent with a sketchbook or colored pencil look.
• + Maintains clear focus on the foreground subjects without excessive blurring.

• − The color palette is a bit overly yellow/saturated, losing some of the soft pastel requested.
• − The aesthetic leans more toward Western illustration/colored pencil than specific Studio Ghibli anime style.

GPT Image 1.5

• + Perfectly captures the 'gentle lighting' and 'warm, nostalgic mood' requested.
• + Captures the iconic Studio Ghibli character design style with softer facial features.
• + The background lighting and bokeh effects create a truly dreamy atmosphere.

• − The girl in the foreground is significantly more blurred than in the original image.
• − Slightly less literal texture than Model A, though still feels painterly.

GPT Image 1 Mini

• + Successfully added billowing hair motion
• + Consistent color palette for added leaves

• − Changed the subject's face significantly, losing the likeness of the original image
• − Low source preservation; the background bridge and character features are altered

GPT Image 1.5

• + Excellent source preservation, keeping the original woman's face and dog's features intact
• + High-quality motion effect on the hair that feels dynamic and natural
• + Effective use of motion blur on the falling leaves to enhance the 'windy' feel

• − The leash handle has a slight anatomical distortion compared to the source

GPT Image 1 Mini

• + Excellent typography with clean, consistent letterforms.
• + Perfect adherence to spellings and accents.
• + Balanced, symmetric layout for a vector emblem style.

• − Ignored the 'light background' instruction, opting for black.
• − The stippled texture on the cloche is a bit harsh.

GPT Image 1.5

• + Beautiful use of warm brown and cream tones with depth.
• + Sophisticated banner design and more natural steam illustration.
• + Contains nice subtle textures that fit the vintage theme well.

• − Ignored the 'light background' instruction.
• − Minor typographic inconsistencies in the 'FLORIAN' lettering compared to the clean lines of Model A.

GPT Image 1 Mini

• + Excellent typography and spelling of all required steps.
• + Minimalist, clean vector style that strictly adheres to the 'flat-vector' request.
• + Logical flow of information with numbered steps and clear icons.

• − The 'Translunar' iconography is a bit messy and lacks a clear trajectory arc icon as requested.
• − The layout feels slightly sparse with significant whitespace.

GPT Image 1.5

• + Professional grid-based composition that feels like a complete infographic poster.
• + Includes specific crew names with correct spelling.
• + Strong visual storytelling within each panel.

• − Missed the numbering requested for the steps.
• − Green is included in the color palette, which was not part of the NASA-inspired color constraint.
• − Minor artifacting on some text (e.g., 'Tranquillity' has a slight double-l spelling variance).