FLUX.2 [klein] 9B Black Forest Labs vs GPT Image 1.5 OpenAI

FLUX.2 [klein] 9B

• + Excellent photorealism in the wood texture and window lighting
• + Natural-looking refractions within the thick glass
• + High resolution with cinematic shallow depth of field

• − The glass container is more of a hollow vase/vessel than a perfect cube
• − The blue sphere appears to be floating unnaturally instead of resting on the bottom

GPT Image 1.5

• + Perfect adherence to the geometry of a cube
• + Accurate placement of the sphere resting on the internal floor
• + Clear visibility of the plant through the glass panels

• − Lighting is a bit flat compared to Model A
• − The glass edges look slightly artificial/digital

FLUX.2 [klein] 9B

• + Excellent preservation of the original car's design, proportions, and lighting.
• + Successful implementation of California coastline scenery in the background.
• + Good sense of motion with wheel blur and road lines.

• − The subject is barely visible, with only the top of his hair showing behind the headrest.
• − Does not effectively integrate the specific person from the source image.

GPT Image 1.5

• + Excellent subject preservation, accurately featuring the man's face, hair, and scarf from the source image.
• + Dynamic composition that creates a connection with the viewer.
• + High level of detail in the car interior and background coastline.

• − The car's proportions are slightly distorted compared to the source, appearing slightly compressed.
• − The lighting on the man does not perfectly match the bright, high-sun environment of the background.

FLUX.2 [klein] 9B

• + Excellent skin texture and realistic hand details
• + Effective use of shallow depth of field and street reflections
• + Great subject focused interaction with the bike

• − The white car in the background lacks the requested motion blur
• − Framing feels a bit too clean and centered for a candid prompt

GPT Image 1.5

• + Better execution of motion blur on the passing vehicle
• + More realistic 'candid' and 'imperfect' framing
• + Excellent wet pavement reflections

• − The man's hand/tool interaction with the wheel is anatomically muddled
• − The skin texture is slightly more smoothed compared to Model A

FLUX.2 [klein] 9B

• + Exceptional intricate engraving on the plate armor
• + Clear and accurate rendering of the beaded braids
• + Very strong adherence to the strap and cloth texture requirements

• − The lighting feels a bit more studio-like and less atmospheric compared to Model B
• − Symmetry in the facial features and braids feels slightly artificial

GPT Image 1.5

• + Beautiful cinematic lighting with realistic warm reflections on the metal
• + Excellent battle-worn aesthetic with realistic dirt and grittiness
• + Very lifelike eyes and expressive facial features

• − The braids are a bit messy and the beads are less distinct than in Model A
• − The engraving on the armor is less sharp and detailed

FLUX.2 [klein] 9B

• + Includes all requested text categories
• + Clean white background with professional shadows

• − Non-existent language and garbled text throughout the body
• − Repetitive food photos featuring four similar pizzas instead of varied dishes
• − Overlapping text in the header

GPT Image 1.5

• + Excellent text legibility and coherent menu items
• + High alignment between dish names and corresponding high-quality photos
• + Very professional and clean multi-column layout

• − Crop is slightly tight on the bottom edge
• − The 'grid' for photos is asymmetrical compared to the text section

FLUX.2 [klein] 9B

• + Excellent typography rendering and clean starburst icon.
• + High photorealistic quality with clear, appetizing textures on the bun and meat.
• + Modern professional composition suitable for a high-end food advertisement.

• − The 'exploded' effect is very conservative, with layers barely separated.
• − The background is more static and less 'fiery' compared to the other model.

GPT Image 1.5

• + Captures the 'exploded' request perfectly with wide separation of all ingredients.
• + Intense, high-energy background with dynamic embers and smoke.
• + Good text integration with a fiery glowing effect on the main title.

• − The 'LIMITED TIME ONLY' text is placed at the very bottom and slightly cut off by the frame.
• − Image has a slightly over-processed or 'deep fried' HDR look that reduces realism.

FLUX.2 [klein] 9B

• + Excellent chalk texture on the board surface
• + Highly accurate text rendering of the complex prompt
• + Strong aesthetic with realistic cursive and layout

• − Small spelling error in 'fress' instead of 'fresh'
• − Composition feels slightly crowded with the smudges

GPT Image 1.5

• + Clean layout with consistent handwriting
• + Accurate spelling throughout the entire text
• + Realistic wooden frame integration

• − Texture appears more like a digital overlay than physical chalk
• − The cursive title is less elegant compared to the competing model

FLUX.2 [klein] 9B

• + Near-perfect adherence to the complex anatomical pose including head tilt and arm angle.
• + Highly accurate recreation of character accessories like the scarf pattern and sunglasses.
• + Excellent lighting consistency with the yellow studio environment.

• − One hand is clenched into a fist, differing from the reference pose's open fingers.
• − Minor skin tone inconsistency between the face and feet.

GPT Image 1.5

• + Faithful reproduction of the character's facial features and specific expression.
• + Good preservation of the red box prop from the source environment.

• − Fails the core pose requirement by keeping the torso and head upright.
• − Proportions are slightly distorted where the legs cross.

FLUX.2 [klein] 9B

• + Excellent high-detail rendering of the space background and planets
• + Clean composition with a clear 'astral' aesthetic
• + Precise rendering of the astronaut's face and suit details

• − Failed the core positional prompt: 'horse on top'
• − Includes distracting hallucinated text at the bottom
• − The horse's Anatomy lacks appropriate space gear, breaking immersion

GPT Image 1.5

• + Stronger cinematic lighting and dynamic movement
• + Detailed lunar surface and space debris integration
• + Realistic horse harness and textures matching the environment

• − Failed the core positional prompt: 'horse on top'
• − Composition feels slightly cluttered with too many floating elements
• − Small artifacts in the horse's mane and dust clouds

FLUX.2 [klein] 9B

• + Successfully preserved the subject's face, hair, and vitiligo patterns
• + Accurately replicated the coat, scarf, and jeans from the reference image
• + Included the gold jewelry and watch as requested

• − Lighting on the coat is slightly too bright compared to the beach environment

GPT Image 1.5

• + Perfectly replicated the texture and drape of the plaid scarf
• + Maintained the background and wooden structure accurately

• − Failed to include the subject's face, which was a core instruction
• − Missing the gold jewelry shown in the base outfit

FLUX.2 [klein] 9B

• + Excellent photorealism with sharp textures on the capybara and the woman.
• + Clean composition with a clear view of both subjects.
• + The expression on the woman perfectly matches the 'bored' instruction.

• − The woman is sitting in the passenger seat rather than the back seat.
• − The text on the hat contains gibberish ('NEW TALA').

GPT Image 1.5

• + Correctly places the woman in the back seat as requested in the prompt.
• + The text on the taxi cap is legible and accurate.
• + High-quality realistic lighting and atmospheric bokeh in the background.

• − The capybara's paws look a bit like bird talons or strange human hands rather than natural capybara feet.
• − The perspective through the windshield is slightly confused compared to the car interior.

FLUX.2 [klein] 9B

• + Excellent typography with clean, highly readable font choices
• + Crisp cinematic lighting on the jack-o-lantern and trees
• + Perfect layout with well-balanced negative space

• − The art style is a bit more 'modern digital' than 'vintage gothic'

GPT Image 1.5

• + Authentic vintage gothic aesthetic with a distressed parchment texture
• + Stronger thematic atmosphere with the graveyard and castle background
• + Ornate border design that feels tactile and illustrative

• − Text is slightly less sharp and harder to read against the busy background
• − The layout feels a bit cramped compared to Model A

FLUX.2 [klein] 9B

• + Excellent texture and realistic hair volume
• + Near-perfect preservation of the original face, glasses, and background
• + Seamless blending at the temples where the hair meets the existing beard

• − The hair covers a bit more of the forehead than might be expected for a classic hairline

GPT Image 1.5

• + Strong prompt adherence for a 'thick head of hair'
• + Good color matching between the new hair and existing beard

• − Slightly alters the facial features, making the eyes and nose look different from the source
• − The hairline connection to the glasses and temples is a bit blurry
• − The overall lighting on the face feels slightly flatter compared to the original

FLUX.2 [klein] 9B

• + Excellent 3D cartoon aesthetic with soft, pillowy textures.
• + Perfectly follows the request for a minimalist diorama style.

• − The flag icon is incorrect, resembling the flag of Yemen rather than Japan.
• − The sushi roll on the right has some structural clipping/warping issues.

GPT Image 1.5

• + High-fidelity PBR materials with realistic textures on the wood and ceramics.
• + Accurate Japanese flag icon.
• + Very clean and bold text typography.

• − Included many extra items (soy sauce, tea pot, chopsticks) despite the request for 'minimal garnish'.
• − The scene feels a bit crowded relative to the 'minimalist' prompt.

FLUX.2 [klein] 9B

• + Excellent caricature style with exaggerated facial features
• + Cleverly integrates dogs into the hockey arena crowd
• + Maintains a cohesive cartoon aesthetic throughout

• − The 'hockey' element is relegated to the background rather than being central to the character
• − Hand holding the microphone is poorly rendered with an extra digit/merged finger

GPT Image 1.5

• + Strong resemblance to the subject in the source image
• + Excellent integration of all prompt elements (desk, mic, dogs, and a literal hockey game)
• + Humorous and creative detail of a dog wearing a hockey helmet

• − The 'caricature' style is less exaggerated and borders more on a digital painting
• − Minor background artifacts in the hockey player's leg area

FLUX.2 [klein] 9B

• + Excellent clarity and vibrant flower details
• + Crisp backlighting and god rays implementation
• + Highly expressive and clean character rendering

• − Completely missed the 'baby bunny' requirement
• − Anatomy of the fox's front legs is slightly awkward

GPT Image 1.5

• + Successfully included all four requested animals
• + Beautiful 'tumbling together' composition that feels more organic
• + Excellent soft fur texture and dew sparkle effects

• − Fox's paws are rendered as dark, indistinct blobs
• − The butterfly near the kitten is missing a body/head

FLUX.2 [klein] 9B

• + Excellent watercolor aesthetic that perfectly nails the 'hand-painted' texture request.
• + Near-perfect preservation of the character poses, clothing patterns, and spatial layout.
• + Captures the Studio Ghibli character design style with clean line art and expressive eyes.

• − Has largely replaced the background urban street with a generic countryside/meadow setting.
• − The woman on the right has lost her distinctive angry expression, looking neutral instead.
• − The man's hand has some minor anatomical awkwardness.

GPT Image 1.5

• + Successfully captures the requested dreamy, warm lighting and nostalgic glow.
• + Preserves the urban street background from the source image better than the competitor.
• + Maintains the distinct facial expressions, particularly the jealousy of the woman on the right.

• − The color palette is overly saturated and orange, drifting away from the requested 'soft pastels'.
• − Texture looks more like a digital filter than a hand-painted illustration.
• − The blur on the character in the foreground is a bit distracting.

FLUX.2 [klein] 9B

• + Excellent preservation of the source subject's facial features and clothing.
• + The flying leaves are crisp and integrate well into the environment.
• + Maintains the exact scale and background of the original image.

• − The wind effect on the hair is slightly less dramatic than Model B.
• − A few leaves appear slightly flat against the subject.

GPT Image 1.5

• + Stronger, more natural-looking wind effect on the hair.
• + Good variety in leaf colors and shapes to suggest movement.
• + Preserves the overall layout and dog's appearance perfectly.

• − Subtle change to the woman's face, specifically making the jawline slightly more angular.
• − One leaf appears to be growing out of the dog's head.

FLUX.2 [klein] 9B

• + Excellent typography and spelling accuracy for both the brand name and the specific accents.
• + Clean, vector-style execution that aligns perfectly with modern logo standards.
• + Follows the light background requirement with subtle paper texture.

• − The steam element is a bit simplified/stylized compared to a natural plume.
• − The banner placement cuts into the cloche unlike a traditional emblem.

GPT Image 1.5

• + Beautiful use of vintage illustrative textures and shading on the cloche.
• + Elegant typography with a classic artisan feel.
• + Good use of the banner at the base of the design.

• − Failed the negative space/background requirement by providing a black background instead of a light one.
• − The steam looks slightly detached from the cloche lid.
• − Slightly less legible 'Caffè' text compared to Model A.

FLUX.2 [klein] 9B

• + Clean vector aesthetic with consistent high-contrast shapes.
• + Includes creative silhouette avatars for the astronauts.
• + Good use of the requested color palette.

• − Severe spelling errors in almost every label (e.g., 'EARDHT', 'TRANSLURAL', 'LANDINING').
• − Logical flow of icons is confusing and non-linear.

GPT Image 1.5

• + Excellent layout that follows a clear, logical step-by-step sequence.
• + Perfect spelling on all technical labels and mission steps.
• + Strict adherence to the NASA-inspired color palette and flat-vector style.

• − Iconography is a bit crowded within the individual panels.
• − The 'Translunar' icon is slightly repetitive of the orbit tiles.