FLUX.1 Kontext [pro] Black Forest Labs vs GPT Image 2 OpenAI

FLUX.1 Kontext [pro]

• + Excellent photographic realism with natural lens blur (bokeh)
• + Clean, precise geometry for the glass cube
• + Accurate representation of the requested lighting from the left

• − The blue sphere has a fuzzy, felt-like texture that maybe wasn't intended
• − The sphere appears to be floating slightly above the bottom surface of the cube

GPT Image 2

• + Strong prompt adherence for all spatial relationships
• + Highly realistic glass refraction and thickness
• + Vibrant colors and sharp textures on the book and sphere

• − The red book is slightly wider than the cube, creating a minor balance oddity
• − The green plant is a bit more crowded in the frame compared to the first model

FLUX.1 Kontext [pro]

• + Excellent depiction of raining atmosphere and wet textures.
• + Highly realistic skin details and naturalistic lighting.
• + Strong bokeh effect that isolates the subject well.

• − The man appears to be riding or leaning on the bike rather than 'repairing' it.
• − Rain streaks look a bit like a static filter overlay in some areas.

GPT Image 2

• + Accurately depicts the 'repairing' action with tools visible.
• + Captures the motion blur of passing cars effectively.
• + The framing feels more like a spontaneous 'candid' shot as requested.

• − Significant anatomical errors with the hands (extra fingers/merged digits).
• − Wet pavement reflections are present but less cinematic than Model A.
• − The bike's handlebars and seat geometry are slightly warped.

FLUX.1 Kontext [pro]

• + Excellent skin texture and realistic facial features
• + Ornate engraving on the armor is clear and prominent
• + Strong lighting coherence with the torchlight effect

• − Missed the request for beads in the hair braids
• − The dirt and scars are very faint, making the character look perhaps too clean for 'battle-worn'

GPT Image 2

• + Strict adherence to all prompt elements including hair beads and dirt/grime
• + Beautifully detailed engraving and weathered texture on the armor
• + Dynamic lighting and effective use of bokeh

• − Slightly less lifelike eye rendering compared to model A
• − Composition feels slightly tighter on the shoulder than the face

FLUX.1 Kontext [pro]

• + Features very clean, minimalist layout consistent with 'modern' design trends.
• + Accurately represents the requested sections (Appetizers, Pizza, Mains).
• + High photographic quality for the food items shown.

• − Text consists of nonsensical, garbled characters.
• − Food images do not always align with the section titles (e.g., a salad-like dish under Pizza and a pizza under Mains).

GPT Image 2

• + Exceptional text rendering with coherent, legible dish names and descriptions.
• + Perfectly executes the grid layout with a high variety of colorful, professional food photos.
• + Strong branding elements and icons that enhance the professional casual dining theme.

• − Slightly more cluttered than typical 'minimalism' due to the high density of photos.

FLUX.1 Kontext [pro]

• + Excellent photorealistic texture on the bun and melting cheese.
• + Clean and legible typography for the main title.
• + High aesthetic quality in the 'magical' ember lighting.

• − Failed to include the starburst element for the price.
• − The price text is redundant, appearing twice.
• − The burger explosion is very static and lacks the requested sense of motion.

GPT Image 2

• + Successfully followed all prompt instructions, including the starburst and specific text placements.
• + Dynamic 'exploded' composition with ingredients tilted and sauce splashing for a high sense of motion.
• + Impressive fiery texture applied directly to the font as requested.

• − The 'LIMITED TIME ONLY' text is slightly less polished than the other text elements.
• − Some minor lighting inconsistencies between the burger and the background flames.

FLUX.1 Kontext [pro]

• + Excellent chalk texture and realistic graining on the letters
• + Perfectly legible text including the small footer note

• − Failed to render the title in the requested 'elegant cursive' style
• − The handwriting looks somewhat digital and uniform compared to natural chalk

GPT Image 2

• + Successfully rendered the title in elegant cursive as requested
• + High level of prompt adherence for all menu items and prices
• + Enhanced cozy café atmosphere with better environmental detail and lighting

• − None notable

FLUX.1 Kontext [pro]

• + Excellent cinematic lighting and composition
• + High level of detail in the astronaut suit and horse anatomy
• + Creative surrealism by including a second small astronaut riding the horse

• − The astronaut has horse hooves instead of boots
• − The physics of the reins and the way the horse is balanced on the astronaut are a bit incoherent

GPT Image 2

• + Perfectly literal adherence to the 'horse on top' spatial instruction
• + High texture quality on the lunar surface and spacesuit
• + Clear and legible NASA-style branding

• − The composition is very centered and less cinematic than the competitor
• − The astronaut's hands have too many fingers (6-7 per hand)

FLUX.1 Kontext [pro]

• + Excellent fur texture and lighting on the capybara.
• + The capybara's paw placement on the steering wheel is very natural.

• − The passenger is on a phone call rather than 'looking at her phone' as requested.
• − The 'TAXI' text on the hat is partially obscured or oddly cropped.

GPT Image 2

• + Perfect adherence to the passenger's action of looking at her phone.
• + The wide-angle composition better captures the 'New York at night' atmosphere through multiple windows.
• + The cap and jacket look more like authentic driver uniforms.

• − The capybara's paws look slightly more like human fingers in gloves than animal paws.
• − Minor distortion on the car door frame in the foreground.

FLUX.1 Kontext [pro]

• + Perfect text rendering for the main titles and banner
• + Clean, high-contrast composition that is easy to read
• + Strong cinematic lighting on the jack-o-lantern

• − Included a line of hallucinated nonsensical text ('Your: Vorkleat...')
• − Background is a bit sparse compared to the 'parchment' request

GPT Image 2

• + Exceptional vintage parchment aesthetic with intricate thorn and web borders
• + Accurate rendering of all requested text elements without hallucinations
• + Richly detailed background featuring the skyline, arches, and gothic architecture

• − The scrolling on the 'Invitation' text is slightly less clean than model A
• − The jack-o-lantern is a bit darker, making it pop slightly less against the background

FLUX.1 Kontext [pro]

• + Excellent typography style that matches the cartoon aesthetic
• + Clean, minimalist composition that adheres strictly to the simple diorama request
• + Soft, clay-like textures create a cohesive miniature look

• − The flag icon is stylized to the point of being a slightly odd shape
• − The rice grains look more like foam balls than realistic rice

GPT Image 2

• + High visual complexity with detailed PBR materials on the sushi and stone
• + Precise rendering of the Japanese flag icon
• + Stronger sense of a 'miniature 3D scene' with the garden and lantern details

• − The text layout is slightly disjointed with large gaps between elements
• − Ignored the request for 'minimal garnish' by including a large garden and multiple leaves

FLUX.1 Kontext [pro]

• + Excellent soft lighting and dreamy golden hour atmosphere.
• + High level of fur detail and consistent 'cute' aesthetic across all animals.

• − Failed to include a distinct bunny, instead creating two cat-like hybrids in the middle.
• − The animals are sitting statically rather than 'playfully chasing' and 'tumbling' as requested.

GPT Image 2

• + Accurately included all four distinct species: puppy, kitten, bunny, and fox.
• + Captured the requested action of chasing and tumbling, providing a much more dynamic composition.
• + Better implementation of 'god rays' and backlighting on the animals' fur.

• − The fox's front paw has some anatomical warping.
• − The kitten's tail is positioned somewhat awkwardly in the air.

FLUX.1 Kontext [pro]

• + Excellent minimalist execution that feels modern yet vintage
• + Bold and readable classic typography
• + Subtle but effective paper texture on the background

• − Small spelling error in 'EEST' instead of 'EST'
• − The steam element is a bit overly simplified

GPT Image 2

• + Highly detailed engraving style with excellent shading on the cloche
• + Correct spelling of 'EST. 1720'
• + Beautifully nested composition within a decorative frame

• − May be considered too complex for a 'minimalist' request
• − The 'F' in Florian is slightly disconnected in its script style

FLUX.1 Kontext [pro]

• + Features a more whimsical, artistic vector style.
• + Correctly captures the requested navy and muted red palette.

• − Fails to follow the numbered sequence and specific steps requested.
• − Produces nonsensical labels and inaccuracies, such as placing Saturn near the rocket and mislabeling the Moon as Earth.
• − Text rendering is poor and illegible in several places.

GPT Image 2

• + Strictly follows all 6 requested steps with perfect text rendering.
• + Expertly executes the clean, modern infographic layout with clear iconography.
• + Adds creative details like the crew names and landing site which enhance the poster theme.

• − The illustrations are slightly more Detailed/3D than a 'flat-vector' style would typically dictate.
• − The Saturn V rocket silhouette in step 1 is a bit squat compared to the real vehicle.