FLUX.1 Kontext [dev] Black Forest Labs vs GPT Image 1.5 OpenAI

FLUX.1 Kontext [dev]

• + Excellent photographic realism with a high-end lens feel.
• + Perfect handling of refractive glass and reflections.
• + Strong adherence to the 'light from the left' instruction with realistic shadows.

• − The glass cube is missing a front face, appearing more like a glass stand or open box.

GPT Image 1.5

• + Successfully renders all six sides of the glass cube.
• + Follows all spatial instructions including the sphere inside and plant behind.
• + Good material texture on the wooden table.

• − The plant visibility through the glass is slightly murky.
• − The lighting lacks the depth and contrast found in the competing image.

FLUX.1 Kontext [dev]

• + Excellent color vibrance and clear rainy day atmosphere.
• + Realistic skin texture and facial features during a candid-style shot.

• − The subject is posing with or sitting on the bike rather than repairing it.
• − The bicycle is missing its chain, which is a significant structural error.
• − Lacks the requested motion blur from passing cars; cars appear static.

GPT Image 1.5

• + Strong adherence to the 'repairing' action with a squatting pose and toolkit.
• + Excellent use of shallow depth of field and motion blur on the background car.
• + Perfectly captures the 'imperfect framing' and 'candid' street photography aesthetic.

• − The bike's rear structure and kickstand area are a bit cluttered/messy in detail.

FLUX.1 Kontext [dev]

• + Excellent high-relief engraving on the chestplate
• + Strong cinematic lighting with dramatic shadows
• + Sharp texture on the fabric underlayer

• − Missed the request for braided hair with beads
• − The face appears too clean and lacks the requested dirt and grit
• − Armor looks more like cast bronze than battle-worn plate

GPT Image 1.5

• + Perfect adherence to all prompt details including braids with beads, dirt, and scars
• + Superior texture on leather straps and cloth elements
• + Extremely lifelike eyes and realistic skin micro-details

• − Slightly cluttered composition due to the amount of hair and detail
• − Reflection on the metal is quite bright, bordering on overexposed in small spots

FLUX.1 Kontext [dev]

• + Strong minimalist aesthetic with bold typography
• + High-quality food photography that feels cohesive
• + Unique grid layout that creates a high-end editorial feel

• − Text is nonsensical and includes severe artifacts or spelling errors
• − Doesn't clearly differentiate several of the specific food sections requested
• − Some graphic elements overlap awkwardly with text

GPT Image 1.5

• + Excellent text readability with no spelling errors
• + Accurately includes all requested sections (Appetizers, Pizza, Mains)
• + Logical layout with Pricing and descriptions following real-world menu conventions

• − Composition feels slightly crowded compared to a truly 'minimalist' design
• − Grid of pictures is less integrated into the white space than Model A

FLUX.1 Kontext [dev]

• + Excellent typography rendering for 'MAGIC BURGER'.
• + Good use of the starburst element for the price.
• + High resolution and clean composition.

• − Failed the main request for an 'exploded' burger, showing a fully assembled one instead.
• − Spelling error in 'LIMITED TIME ONLY' (rendered as 'LNHLY').
• − The background coals look a bit generic and static compared to the dynamic request.

GPT Image 1.5

• + Perfect adherence to the 'exploded' burger request with dynamic suspended components.
• + High level of texture detail on the patty, bun, and vegetables.
• + Successfully applied the fiery, glowing effect to both the background and the text.

• − The pricing starburst is a bit jagged and overlaps the burger.
• − Some visual clutter from the excessive ember sparks near the top text.

FLUX.1 Kontext [dev]

• + Clean wood frame composition.
• + Consistent handwriting style across different sections.

• − Significant spelling errors throughout including 'Mashroom', 'Risoktso', and 'Octpus'.
• − Date text at the top is largely illegible and jumbled.

GPT Image 1.5

• + Perfect text rendering with zero spelling errors.
• + Excellent chalk texture and smudging details for high realism.
• + Strict adherence to the cursive style requested for the title.

• − The bottom line of text uses a slightly different, more uniform weight that looks a bit more digital than the main items.

FLUX.1 Kontext [dev]

• + Excellent adherence to the 'horse on top' spatial instruction
• + Clear and clean visual style with high resolution
• + Captures the surreal nature of the prompt well

• − The composition is a bit static compared to a cinematic style
• − The astronaut's suit looks more like a modern jump suit than a technical space suit

GPT Image 1.5

• + Dynamic and cinematic lighting with rich textures
• + High level of detail on the space suit and celestial bodies

• − Complete failure to follow the core instruction of 'horse on top'
• − Anatomical issues with the horse's front legs/hooves
• − Overly busy background creates visual clutter

FLUX.1 Kontext [dev]

• + Excellent fur texture and sharp lighting
• + High resolution with clean, professional composition
• + Human facial expression is very grounded and realistic

• − Failed the instruction to have both paws on the steering wheel
• − The cap is a modern baseball style rather than a traditional taxi driver cap
• − The capybara's face looks slightly anthropomorphized compared to a real animal

GPT Image 1.5

• + Perfectly followed the instruction for both paws on the steering wheel
• + Captures a very authentic 'New York taxi driver' cap and vintage interior aesthetic
• + Features a more realistic animal face that contrasts well with the absurd scenario

• − The passenger in the back is slightly out of focus and has some minor artifacts on her hands/phone
• − The 'TAXI' text on the cap is a bit generic

FLUX.1 Kontext [dev]

• + Excellent large title text readability
• + Clean and simple graphic composition
• + Follows the square format request

• − Significant spelling errors in the scroll banner and location text
• − Lacks the 'vintage parchment' texture requested, appearing more like a flat digital poster
• − Bats look like simple cartoon icons rather than cinematic elements

GPT Image 1.5

• + Perfect text rendering for all requested details, including the scroll and event info
• + Strong adherence to 'vintage gothic parchment' and 'moody night sky' aesthetic
• + Highly detailed thorns and webs border that adds to the cinematic quality

• − The 'Arches' location text is slightly less prominent than other details
• − Composition is a bit crowded with many competing atmospheric elements

FLUX.1 Kontext [dev]

• + Excellent addition of thick, dark hair with defined texture.
• + Preserves the overall composition and lighting of the background.

• − Significantly alters facial features, making the person look like a different individual.
• − The hair texture looks slightly airbrushed/artificial compared to the original beard.

GPT Image 1.5

• + Successfully preserves the original facial structure, glasses, and skin details.
• + Matches the hair texture and color perfectly to the existing beard for a cohesive look.
• + Maintains the grainy, film-like quality of the source image.

• − The hairline transition is slightly visible upon close inspection, but still realistic.

FLUX.1 Kontext [dev]

• + Excellent soft 3D cartoon aesthetic
• + Very clean and minimalist composition
• + Large, legible text

• − The flag icon is a strange geometric shape instead of the Japanese flag
• − The sushi anatomy is slightly confusing with the black band placement

GPT Image 1.5

• + Detailed and accurate Japanese flag icon
• + Beautiful PBR materials with realistic textures
• + Excellent interpretation of a 'miniature diorama' on a base

• − Ignored the 'minimal garnish' instruction by adding a teapot, soy sauce, and multiple items
• − Text is smaller and less bold compared to Model A

FLUX.1 Kontext [dev]

• + Provides a comic-book style caricature that translates physical features well.
• + Includes a small dog and a television set to represent the job.

• − Completely misses the hockey requirement.
• − The text 'JOB' and other elements are unnecessary and low-quality.
• − Poorly rendered hand/paw at the bottom right.

GPT Image 1.5

• + Successfully incorporates all elements: TV anchor, multiple dogs, and hockey (background and helmet).
• + High visual quality with a professional digital caricature aesthetic.
• + Excellent text rendering of 'BREAKING NEWS' and clear facial expressions.

• − Changes the subject's clothing from the original denim to a red dress.
• − The hockey stick held by the small dog is floating/awkwardly placed.

FLUX.1 Kontext [dev]

• + Excellent sense of motion and action/chasing.
• + Clean, high-quality digital aesthetic with well-defined butterflies.
• + Strong lighting and color saturation that fits the 'joyful' theme.

• − Failed to include all four animals (missing the bunny and fox).
• − The animals look more like generic 3D assets than photorealistic fur.
• − Incorrectly generated two cats instead of a kitten and fox.

GPT Image 1.5

• + Successfully included all four requested animals: puppy, kitten, bunny, and fox.
• + Beautifully rendered 'god rays' and dew sparkles as requested in the prompt.
• + Highly detailed fur texture and complex wildflower interaction.

• − The composition is a bit crowded with the animals overlapping awkwardly.
• − The fox's paw in the bottom right corner has some anatomical blurring.
• − The lighting on the puppy's face is slightly flat compared to the dramatic background rays.

FLUX.1 Kontext [dev]

• + Excellent preservation of the original image's composition and poses
• + Very clean line art and clear character expressions
• + Matches the modern anime aesthetic common in 2D digital art

• − Does not capture the painterly Ghibli aesthetic well
• − Colors are too saturated and lack the requested soft pastel quality

GPT Image 1.5

• + Perfectly captures the Studio Ghibli 'hand-painted' texture and soft lighting
• + Uses the requested pastel color palette and warm, nostalgic mood
• + Strong adherence to the 'dreamy background' requirement

• − Slight loss of sharpness in facial features compared to the source
• − The girl in the foreground loses some of her distinctive facial proportions

FLUX.1 Kontext [dev]

• + Excellent source preservation, maintaining the woman's face and the dog's features accurately.
• + Subtle and realistic hair movement that looks natural for a breeze.

• − The 'flying leaves' are barely noticeable and look more like green specks than leaves.
• − Overall energy shift is very minimal compared to the source.

GPT Image 1.5

• + Strong adherence to the request for 'energetic and lively' through dramatic wind and leaf effects.
• + Successfully captures many flying leaves with motion blur, creating a sense of depth.

• − Significant loss of detail in the woman's face and hair texture compared to the source.
• − Slight alteration of the dog's facial features and eyes.

FLUX.1 Kontext [dev]

• + Clean vector aesthetic suitable for a minimalist brand
• + Excellent text legibility and font choice
• + Perfectly centered and balanced composition

• − Missed the 'banner' requirement for the established date
• − The cloche icon looks more like a building dome or a cupcake

GPT Image 1.5

• + Successfully included the 'banner' for the 'Est. 1720' text
• + Very clear and well-illustrated cloche dome
• + Excellent use of the 'subtle texture' and 'vintage' style requested

• − Failed the 'light background' requirement, opting for black
• − The word 'Caffè' is slightly less legible due to the scripted font style

FLUX.1 Kontext [dev]

• + Follows the color palette closely with a navy background.
• + Captures a stylized, abstract vector feel.

• − Text is heavily garbled and contains numerous spelling errors like 'APOLO'.
• − The icons are messy, non-representative of the prompt steps, and difficult to interpret.
• − Failed to follow the logical sequence of the mission steps.

GPT Image 1.5

• + Excellent text rendering with correct spelling for all mission steps and astronaut names.
• + Perfect adherence to the 6-step logical sequence with clear, relevant icons for each.
• + Clean, professional composition that looks like a real infographic poster.

• − Added green to the Earth icon which was not part of the requested NASA-inspired palette.
• − Slightly more complex than 'flat-vector' due to some layered shapes, though still highly effective.