FLUX.2 [flex] Black Forest Labs vs GPT Image 1.5 OpenAI

FLUX.2 [flex]

• + Clean, minimalist composition with a high-quality photographic feel.
• + Excellent rendering of the glass cube's beveled edges and physical interactions with the book.
• + Subtle and realistic soft lighting from the left window.

• − The plant is quite blurred, making its visibility through the glass less distinct than requested.
• − The blue sphere has a matte texture that lacks the reflective properties typically found in such a scene.

GPT Image 1.5

• + Strong adherence to the requirement of seeing the plant through the glass.
• + The blue sphere features more realistic reflections, including the window light.
• + Good texture on the red book cover and the wooden table grain.

• − The light reflection on the table looks slightly repetitive or artificial.
• − The bottom of the glass cube has a mirrored base which wasn't requested and slightly confuses the geometry.

FLUX.2 [flex]

• + Successfully integrates the exact car and man into a California coastline setting.
• + Preserves the full silhouette and design details of the Rolls Royce Phantom Drophead Coupé.
• + Maintains excellent visual coherence with realistic motion blur on the wheels and road.

• − The man is slightly small in the composition compared to the car's scale.
• − The character's face loses some detail due to the distance of the shot.

GPT Image 1.5

• + Excellent character preservation, capturing the man's face, hair, and clothing accurately.
• + Stronger focus on the 'driving' aspect with a clear view of the interior and hands on the wheel.
• + The coastline background is highly scenic and matches the requested location well.

• − Crops out the majority of the car, which was a primary subject of the source images.
• − The perspective of the car interior feels slightly claustrophobic compared to the open-top source.

FLUX.2 [flex]

• + Excellent adherence to 'motion blur from passing cars' with long-exposure style headlights.
• + Very realistic skin texture and lighting on the subject.
• + Strong cinematic composition with a shallow depth of field reflecting a 50mm lens.

• − The bicycle frame geometry is slightly physically impossible near the pedals.
• − The man's hands are a bit muddled in the interaction with the chain.

GPT Image 1.5

• + Captures a more authentic 'candid' street photography feel with tools and accessories.
• + Excellent texture on the wet pavement and clothing.
• + The bicycle design is more coherent and realistic for a commuter bike.

• − Missing the 'motion blur from passing cars' requested in the prompt.
• − The car in the background is static rather than blurred by movement.

FLUX.2 [flex]

• + Symmetrical and clear armor engraving and leather strap details.
• + Excellent representation of the bead requirement in the braided hair.
• + Balanced composition with effective use of bokeh sparks.

• − The scars look like surface-level digital paint rather than integrated skin texture.
• − The lighting on the face is a bit flat compared to the dramatic armor highlights.

GPT Image 1.5

• + Exceptional skin texture with realistic pores, dirt, and integrated scarring.
• + Stunning interplay of warm torchlight and cool shadows on the skin and metal.
• + Highly lifelike eyes with complex reflections.

• − Some minor geometric inconsistencies in the armor engravings on the left side.
• − The straps/buckles are less defined than in the competing image.

FLUX.2 [flex]

• + Strict adherence to the requested grid layout for photos.
• + Balanced use of colors corresponding to the specific sections.
• + Professional mock-up presentation with realistic shadows.

• − Text is largely gibberish and unreadable.
• − The food photos in the grid don't always match the section they are under (e.g., pizza and steaks under the 'Appetizers' header).

GPT Image 1.5

• + Perfectly legible and logical text with realistic menu items and descriptions.
• + Strong photographic quality with food that looks appetizing and relevant to the sections.
• + Good use of bold sans-serif fonts as requested.

• − Layout is more of a horizontal split than a distinct grid as requested.
• − The 'Appetizers' section has pizza photos adjacent to it, which is slightly confusing.

FLUX.2 [flex]

• + Extremely clean and legible typography with a realistic flame effect.
• + The 'exploded' effect feels more controlled and allows for a clearer view of individual high-quality textures.
• + Excellent lighting on the food items that matches the fiery background.

• − The composition feels slightly more like a static stack than a dynamic explosion.

GPT Image 1.5

• + Stronger sense of motion and 'exploded' energy with flying debris and angled components.
• + Very high level of detail on the grilled texture of the patty and the toasted bun.
• + Great integration of the starburst effect into the fiery scene.

• − The text is a bit cramped at the top and bottom of the frame.
• − The overall image is very busy, which slightly reduces the focus on the product.

FLUX.2 [flex]

• + Excellent chalk texture with realistic smudges and dust on the board.
• + Perfect spelling and completion of the text prompt including the footer.
• + Strong photographic composition with a realistic wooden frame and depth of field.

• − The font style feels slightly more uniform and less 'handwritten' than requested.
• − The lighting reflection is a bit intense at the top center of the board.

GPT Image 1.5

• + Very realistic chalk handwriting style with authentic variations in letter height and slant.
• + Excellent adherence to the 'slanted' and 'natural variation' part of the prompt.
• + Perfect transcription of the requested text including the added menu item details.

• − The overall image resolution and sharpness are lower than the competitor.
• − The framing is very tight, losing the 'cozy café' environmental context.

FLUX.2 [flex]

• + Excellent character replication including the sunglasses, scarf patterns, and facial hair.
• + Good environment matching with the yellow lighting and red ottoman.
• + Accurately places the character in a crouched, dynamic position on the ottoman.

• − The pose is significantly altered from the source, missing the specific 'crossed-leg' balance seen in Image 1.
• − The raised arm is bent and pointing rather than the straight, elegant extension in the reference.

GPT Image 1.5

• + Successfully replicates the specific crossed-leg pose and body lean from the reference image.
• + Maintains the character's clothing details including the scarf tassels and sunglasses.
• + High visual quality and seamless integration of the character into the yellow studio environment.

• − The lower hand pose is slightly generic compared to the delicate finger position in Image 1.
• − The top edge of the image is cut off slightly compared to the reference frame.

FLUX.2 [flex]

• + Excellent adherence to the specific prompt instruction of the horse being on top.
• + Beautiful cinematic lighting and color palette with a clear surreal aesthetic.
• + High anatomical quality for both the horse and the astronaut's suit.

• − The connection point between the horse's front legs and the astronaut's hands is a bit physically ambiguous.
• − Small artifacts present in the asteroid field.

GPT Image 1.5

• + High level of surface detail on the lunar ground and space lander.
• + Clear, sharp focus across the entire composition.

• − Completely failed the negative constraint to have the horse on top.
• − Commonplace, literal interpretation rather than the requested surreal subversion.
• − The dust effect looks a bit disconnected from the vacuum of space context.

FLUX.2 [flex]

• + Excellent lighting and high-definition texture on the capybara's fur.
• + Dynamic composition with a wide-angle view that captures more of the NYC city lights.
• + Realistic rendering of the capybara's paws gripping the steering wheel.

• − The transition between the capybara's head/neck and the human-looking hands is slightly uncanny.
• − The capybara's head is turned away from the road, making the pose feel less natural.

GPT Image 1.5

• + Perfect framing through the windshield creates a cinematic and immersive 'inside the taxi' feel.
• + The capybara's expression is very professional and forward-facing as a driver should be.
• + Successfully captures the specific requested 'bored' expression of the passenger.

• − The passenger's hair and facial features are slightly blurry/out of focus compared to the foreground.
• − The taxi cap logo is a bit generic compared to the detailed badge in Model A.

FLUX.2 [flex]

• + Excellent typography with clean, legible gothic fonts
• + High-quality central jack-o-lantern rendering with effective glowing light
• + Well-defined torn parchment border effect

• − The composition feels slightly empty in the mid-ground
• − The border design is a bit repetitive

GPT Image 1.5

• + Atmospheric and rich vintage texture that feels like real weathered parchment
• + Intricate border with high detail on the thorns and spiderwebs
• + Stronger thematic coherence between the background elements and the foreground

• − Text rendering is slightly less sharp than Model A
• − The 'thorns' in the border are very busy, making the edges look a bit messy

FLUX.2 [flex]

• + Excellent preservation of original identity and facial details.
• + Natural-looking buzz-cut texture that matches the existing beard density.
• + Flawless integration of the hairline and ears.

• − The hair is somewhat thin/short, bordering on a buzz cut rather than a 'full, thick head' as requested.

GPT Image 1.5

• + Successfully provides a much thicker, fuller head of hair.
• + Matches the hair texture well to the existing beard texture.
• + Preserves the overall environment and composition.

• − Slightly alters the facial structure, making the man look somewhat younger or like a different person.
• − The hairline integration on the forehead is a bit harsh.

FLUX.2 [flex]

• + Excellent adherence to the 'cartoon' and 'soft texture' requirement
• + Clean minimalist composition with a perfect 45-degree isometric angle
• + Strong text clarity and accurate flag icon

• − Texture of the sushi toppings is slightly too plastic-looking for PBR materials

GPT Image 1.5

• + Highly impressive realistic PBR materials, especially on the wood and ceramic
• + Detailed interpretation of a diorama with a 3D base
• + Excellent text rendering and layout

• − Ignored the 'minimal garnish' instruction by adding excessive accessories like a teapot and soy sauce
• − Less 'cartoon' in style than requested, leaning more towards hyper-realism

FLUX.2 [flex]

• + Successfully incorporates all elements including hockey, dogs, and the news desk.
• + Maintains the exact denim outfit from the source image.
• + Consistently applies a clean, 2D cartoon caricature style across the whole image.

• − The 'hockey' element is a bit subtle, represented primarily by the dogs' jerseys and one stick.
• − The text on the news desk ('THEIANAES') is nonsensical.

GPT Image 1.5

• + Excellent caricature of the subject's face while maintaining a strong resemblance to the source.
• + High level of humorous exaggeration with the hockey-playing dog and news ticker.
• + Clearly legible and relevant text in the graphic overlay.

• − Changed the subject's outfit from the denim shirt in the source to a red dress.
• − The hand holding the microphone has anatomical issues (too many fingers/blurry).

FLUX.2 [flex]

• + Excellent composition with clear space between animals, allowing their full bodies and movements to be seen.
• + Superior lighting and atmosphere, featuring beautiful god rays and a distinct sunrise glow.
• + Includes all four requested animals with distinct, accurate features for each species.

• − The kitten and puppy appear to be floating or jumping rather than 'tumbling' as specified.
• − The butterflies look slightly pasted on rather than fully integrated into the lighting of the scene.

GPT Image 1.5

• + Captures the 'tumbling together' part of the prompt much more effectively with a cozy, grouped interaction.
• + Exceptional fur texture and detail, particularly on the kitten’s paws and the fox's face.
• + The warm, golden light feels very integrated into the scene's dew and sparkles.

• − The fox is missing its lower body/hind legs, looking like a disembodied head and torso.
• − The composition is very crowded, making it harder to distinguish the individual forms of the four animals.

FLUX.2 [flex]

• + Excellent preservation of the original pose and character positions.
• + Captures the iconic line-heavy Studio Ghibli 'Only Yesterday' or 'Princess Kaguya' style aesthetic.
• + Maintains the specific plaid pattern of the man's shirt very accurately.

• − The woman on the right misses the angry/shocked expression from the original.
• − The color palette is a bit washed out compared to typical Ghibli vibrant landscapes.

GPT Image 1.5

• + Successfully captures the warm, nostalgic, and dreamy lighting requested.
• + Preserves the emotional context of the scene, including the girlfriend's angry expression.
• + Beautiful watercolor-inspired texturing and soft pastel color palette.

• − The woman in the foreground has significantly longer and more voluminous hair than the source image.
• − More of a general shoujo-manga aesthetic rather than specifically Studio Ghibli.

FLUX.2 [flex]

• + Excellent preservation of the source image's identity and background.
• + The hair motion looks natural and maintains the woman's facial structure.
• + The dog's tail is subtly modified to suggest wagging movement.

• − The flying leaves look like static green shapes pasted over the foreground rather than part of the environment.
• − The motion feel is slightly less energetic compared to model_b.

GPT Image 1.5

• + Stronger sense of 'dynamic motion' with more abundant and varied leaf textures.
• + Great hair dynamics that create a wind-swept appearance.
• + Good color integration of the leaves with the autumn-toned lighting.

• − Noticeable anatomical distortion around the woman's right shoulder and neckline where the hair meets the jacket.
• − The dog's left front leg is morphed and looks unnatural compared to the source.

FLUX.2 [flex]

• + Perfect adherence to the light background request.
• + Excellent minimalist vector style that feels like a clean modern-vintage logo.
• + Very clean and readable typography for both the main title and the banner text.

• − The texture is extremely subtle, almost invisible compared to the prompt's request.

GPT Image 1.5

• + Excellent use of vintage textures and shading on the cloche dome.
• + Accurate recreation of all requested text elements, including the accent on 'Caffè'.
• + Stronger 'vintage' character in the illustration style.

• − Completely ignored the request for a light background, providing a black one instead.
• − Composition feels a bit more cluttered and less 'minimalist' than Model A.

FLUX.2 [flex]

• + Excellent adherence to the clean, modern vector aesthetic.
• + Very high text clarity and accuracy across all labels.
• + Elegant use of negative space and layout for a professional infographic feel.

• − Failed to include the 6th step (Landing).
• − The 'Translunar' iconography is a bit disconnected from the other elements.

GPT Image 1.5

• + Successfully included all 6 requested steps from Launch to Landing.
• + Strong consistency in the flat-vector illustration style.
• + Creative inclusion of the crew silhouettes and the 'Tranquility' landing site marker.

• − Image is cropped at the top, cutting off the main title.
• − The 'Translunar' arc is somewhat cluttered compared to the other clean sections.