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

FLUX.1 Kontext [dev]

• + Excellent adherence to lighting instructions with a clear soft window light from the left.
• + Highly realistic textures on the wood grain and glass reflections.
• + Vibrant and saturated colors.

• − The sphere appears more like a large marble than a 'small sphere'.
• − The glass cube has an inconsistent open-bottomed structure that doesn't fully enclose the sphere.

GPT Image 2

• + Perfectly captures the scale of a 'small' blue sphere relative to the cube.
• + Strong composition with clear layering of the foreground objects and the plant behind.
• + The glass cube has a more realistic and defined structural geometry.

• − The blue sphere has a slightly flat, matte texture compared to the highly detailed surrounding materials.
• − The 'soft window light' is less pronounced than in the competitor's image.

FLUX.1 Kontext [dev]

• + Excellent depiction of raindrops and realistic wet pavement reflections
• + High quality rendering of the subject's face and clothing
• + Cinematic color grading and lighting

• − Failed the core action: the man is sitting on/straddling the bike rather than repairing it
• − Composition is very centered and posed, missing the requested 'imperfect framing' and 'candid' feel
• − Background cars are static and sharp rather than having motion blur

GPT Image 2

• + Perfect adherence to the 'repairing' action and 'candid' street photography style
• + Accurately captured motion blur on passing vehicles and imperfect framing
• + Highly realistic skin textures and details, including a toolbox and realistic bicycle wear

• − The light rain is very subtle and less visible than in Model A
• − A slight blurring artifact on the subject's right hand involved in the repair

FLUX.1 Kontext [dev]

• + Ornate engraved plate armor is highly defined and clear
• + Dramatic warm torchlight lighting with effective bokeh sparks
• + High contrast and sharp facial features

• − Missed the request for braided hair with beads
• − Skin appears relatively clean with very minimal 'battle-worn' texture compared to the prompt
• − Leather and cloth textures are somewhat obscured by shadows

GPT Image 2

• + Excellent adherence to the 'braided hair with small beads' requirement
• + Superb skin texture showing dirt, freckles, and a realistic battle-worn look
• + Highly detailed rendering of worn metal, leather straps, and cloth underlayers

• − Lighting feels more like natural daylight with a small backlight rather than warm torchlight
• − The bokeh sparks are very subtle compared to Model A

FLUX.1 Kontext [dev]

• + Excellent high-resolution food photography that feels authentic
• + Strong modern grid layout that maximizes visual impact of the food

• − Text consists of gibberish words and incoherent characters
• − Layout lacks the functional structure of a usable menu

GPT Image 2

• + Perfectly legible English text with logical descriptions and pricing
• + Highly organized sections for Appetizers, Pizza, and Mains as requested
• + Professional use of iconography, branding, and social media footers

• − The food photos look slightly more like stock 3D renders compared to Model A
• − Layout is quite dense, bordering on a flyer rather than a minimalist menu

FLUX.1 Kontext [dev]

• + Clean and legible text rendering
• + Vibrant colors and high-contrast lighting

• − Failed to create an 'exploded' burger, showing multiple whole burgers instead
• − Used the wrong currency symbol (£ instead of €)
• − The composition feels static rather than dynamic

GPT Image 2

• + Perfect adherence to the 'exploded burger' layout
• + Energetic composition with great sense of motion and detail
• + Beautifully integrated fiery text effects and the correct currency symbol

• − The 'LIMITED TIME ONLY' text is slightly less crisp than the other elements
• − The top bun is slightly distorted in perspective

FLUX.1 Kontext [dev]

• + Features a clear, well-framed chalkboard with high contrast.
• + Good imitation of chalk texture on the 'Today Specials' header.

• − Numerous spelling errors including 'Mashroom', 'Risoktso', 'Octpus', and garbled date text.
• − Multiple repetitions of words like 'with with' and duplicated price symbols.
• − The text style looks like a digital marker font rather than natural chalk.

GPT Image 2

• + Excellent adherence to the requested cursive elegant header style.
• + Perfect spelling and grammar for all menu items and the date.
• + Highly realistic chalk texture with natural-looking dust and imperfections on the board.

• − The ambient lighting is slightly dim, making the bottom text a bit harder to read compared to the top.

FLUX.1 Kontext [dev]

• + Excellent anatomical rendering of both the human face and the horse.
• + Clean, cinematic lighting and high resolution.

• − Failed the primary instruction to have the horse riding the astronaut; the horse is simply floating behind him.
• − The composition feels static and less surreal than requested.

GPT Image 2

• + Perfect adherence to the complex prompt instruction of 'horse on top'.
• + High level of surrealism with a saddle placed on the astronaut's back.
• + Strong detailed texture on the moon's surface and the spacesuit.

• − The astronaut's hands have too many fingers (6-7 visible on each hand).
• − The horse's front legs and harness attachment are anatomically confusing.

FLUX.1 Kontext [dev]

• + Excellent fur texture rendering
• + Clear, high-quality lighting inside the cabin
• + Realistic human features

• − The capybara only has one paw on the wheel while the other is on its lap
• − The capybara's face looks slightly more like a general rodent/groundhog than a distinct capybara

GPT Image 2

• + Perfect adherence to the pose prompt with both paws on the steering wheel
• + Very accurate capybara facial anatomy
• + Authentic New York taxi driver hat design

• − Slightly lower background detail clarity
• − The lighting is a bit muddy compared to Model A

FLUX.1 Kontext [dev]

• + Strong, glowing central jack-o-lantern.
• + Border of thorns is clearly defined.

• − Several spelling errors in the scroll banner and location text.
• − Lacks the 'vintage parchment' texture requested, appearing more like a modern digital graphic.
• − The background is mostly flat black rather than a detailed night sky.

GPT Image 2

• + Perfect text rendering for all requested titles, banners, and event details.
• + Excellent gothic atmosphere with rich textures, parchment effects, and cinematic lighting.
• + High level of detail in the border, moon, and background scenery.

• − The pumpkin's internal glow is slightly less intense than Model A's.

FLUX.1 Kontext [dev]

• + Clean, minimalist 2D/3D hybrid aesthetic
• + Accurate text rendering for 'JAPAN' and 'SUSHI'
• + Strict adherence to the 'minimal garnish' instruction

• − The flag icon is unrecognizable and abstract
• − The sushi model is overly simplified, lacking realistic PBR textures requested

GPT Image 2

• + Excellent miniature 3D diorama execution with high detail
• + Correct Japanese flag icon and well-styled 3D typography
• + Superior material rendering with texture on the fish and wood

• − Ignored the 'minimal garnish' instruction, including a complex scene
• − The text layout is slightly crowded at the top

FLUX.1 Kontext [dev]

• + Strong bokeh effect and warm lighting.
• + Action-oriented poses showing joyful movement.

• − Failed to include the bunny and red fox kit, showing only three animals (dog and two kittens/cats).
• − The 'tabby' kitten is depicted as white and orange rather than distinct tabby markings.
• − Anatomical issues with the animals' paws and the floating butterfly wing in the background.

GPT Image 2

• + Excellent prompt adherence, including all four specific animals: puppy, tabby kitten, bunny, and fox kit.
• + Realistic fur textures and clear, expressive eyes in alignment with the prompt.
• + Superior lighting effects featuring clear god rays and dew-like sparkles.

• − The fox kit’s leg and paw anatomy is slightly distorted during the run.

FLUX.1 Kontext [dev]

• + Perfectly follows the minimalist and vector emblem style instruction
• + Clean, high-contrast typography
• + Accurate spelling and accented character rendering

• − Missed the 'banner' requirement for the 'Est. 1720' text
• − The steam effect is very abstract and less recognizable

GPT Image 2

• + Excellent adherence to the 'vintage' and 'banner' requirements
• + Beautifully textured background and sophisticated engraving style
• + Accurate spelling with elegant serif typography

• − Less 'minimalist' than requested, leaning more into high-detail ornamental design

FLUX.1 Kontext [dev]

• + Successfully captures a flat, minimalist vector style
• + Adheres to the muted red and navy color palette
• + Uses interesting abstract iconography

• − Significant text corruption and spelling errors (e.g., 'APOLO')
• − Icons are too abstract and do not clearly represent the requested mission steps
• − The composition feels cluttered and lacks a logical flow

GPT Image 2

• + Excellent layout with a clear chronological timeline
• + High text legibility and accurate spelling of names and stages
• + Includes detailed, recognizable icons for the Saturn V and Lunar Module

• − Stylistically more detailed/illustrative than the requested 'flat-vector' style
• − The 'Earth Orbit' illustration shows the command module relative to the planet in an oversized scale