DALL-E 3 OpenAI vs FLUX.1 Kontext [dev] Black Forest Labs

DALL-E 3

• + High artistic detail with intricate textures on the wood and sphere.
• + Excellent lighting and atmosphere.
• + Correct placement of the plant behind the cube.

• − Failed the primary spatial instruction by putting the book inside the cube and the sphere on top of the book.
• − Conceptualized the cube with a bulky wooden frame rather than a simple glass cube.

FLUX.1 Kontext [dev]

• + Perfect adherence to spatial instructions including the book on top and sphere inside.
• + Realistic textures and reflections on the glass and wooden surface.
• + Correct interpretation of 'soft window light from the left' including the visible window.

• − The sphere is slightly large for the described 'small sphere' proportion.
• − The cube has a mirror-like bottom surface not specifically requested.

DALL-E 3

• + Captures the 'repairing' action perfectly with the man crouching and using tools.
• + Excellent adherence to the 'imperfect framing' and 'shallow depth of field' with foreground bokeh elements.
• + Very realistic wet pavement reflections and cinematic atmospheric lighting.

• − The man's skin texture on his back looks slightly unnatural/overly aged.
• − Anatomy of the feet is a bit distorted.

FLUX.1 Kontext [dev]

• + High clarity and sharp rendering of the subject's face.
• + Explicitly shows raindrops throughout the frame.

• − Completely failed the 'repairing' prompt; the man is standing with/mounting the bike.
• − Composition is too centered and clean, missing the 'imperfect' and 'candid' street photography aesthetic.
• − Lacks the requested 'motion blur from passing cars'.

DALL-E 3

• + Excellent close-up composition highlighting high-fidelity skin and armor textures.
• + Strong adherence to the 'engraved plate armor' and 'warm torchlight' prompts.
• + Superior rendering of fine details like the beard hair and the fabric grain on the collar.

• − Failed to include the requested 'braided hair with small beads'.
• − The scars appear a bit stylized/digital rather than naturally aged skin damage.

FLUX.1 Kontext [dev]

• + Captures a more intense, brooding emotion suitable for a battle-worn character.
• + Features subtle braiding in the hair, moving closer to that specific prompt requirement.
• + Good use of bokeh sparks in the background.

• − The armor engraving is less 'ornate' and looks more like cast molding compared to Image A.
• − The lighting is somewhat flat and lacks the dramatic warm reflections on the metal requested.
• − The texture on the leather straps is less distinct and detailed than the competitor.

DALL-E 3

• + Provides a complete multi-page layout view suitable for a presentation.
• + Accurately includes several sections labeled for pizza, appetizers, and mains.
• + Uses vibrant color accents that enhance the professional layout.

• − Text rendering is very poor and distorted, even for large headings.
• − The grid layout feels a bit cluttered compared to a professional menu.

FLUX.1 Kontext [dev]

• + Excellent typography with clean, bold sans-serif fonts that look modern.
• + High-quality, appetizing food photography with vibrant colors.
• + Strong grid composition that follows contemporary design trends.

• − Text is largely gibberish with several misspelling artifacts.
• − Fails to clearly distinguish between 'appetizers', 'pizza', and 'mains' as requested.

DALL-E 3

• + Excellent 'exploded' effect with all components suspended separately.
• + Superior photorealistic detail in the food textures.
• + Dynamic and creative lighting that enhances the sense of motion.

• − Multiple spelling errors in the text including 'MAGIC BURGR' and 'Limiited'.
• − The price is not in a starburst as requested.

FLUX.1 Kontext [dev]

• + Perfect text rendering for 'MAGIC BURGER' and '€6.99'.
• + Includes the starburst element for the price as requested.
• + Clean composition with legible typography.

• − Failed to provide an 'exploded' view; the burger is mostly assembled and static.
• − Spelling error in 'ONLY' which appears as 'LNHLY'.
• − The visual style is more illustrative and less photorealistic than requested.

DALL-E 3

• + Excellent chalk texture throughout the image
• + Atmospheric lighting and café environment details
• + Artistic flourishes and illustrations enhance the board aesthetic

• − Significant spelling errors on almost every line
• − Incorrect prices for items
• − The layout is cluttered and difficult to read as a functional menu

FLUX.1 Kontext [dev]

• + High legibility for the handwriting
• + Followed the menu items more accurately than the competitor
• + Clean composition that resembles a real café chalkboard

• − Repetitive words in text such as 'Mushroom Mashroom' and 'with with'
• − The date is garbled and unreadable
• − The chalk texture is very clean and lacks the natural dustiness of real chalk

DALL-E 3

• + Excellent cinematic atmosphere and background detail
• + Stunning use of light and color in the nebulae and clouds
• + High level of polish on the astronaut and horse silhouettes

• − Failed the negative constraint to have the horse on top of the astronaut
• − Standard interpretation of the prompt without surreal reversal

FLUX.1 Kontext [dev]

• + Successfully followed the difficult spatial instruction of horse 'on top'
• + Creative and literal interpretation of the surreal request
• + Great facial detail inside the helmet

• − The composition is somewhat static and less cinematic than the competitor
• − Background is relatively plain compared to the vibrant space scene in Image A

DALL-E 3

• + Excellent detailed rendering of capybara fur and whiskers.
• + Cinematic lighting consistent with a night scene in a city.
• + Clear taxi driver aesthetic with the uniformed cap and dashboard details.

• − Completely failed to include the human businesswoman in the back seat.
• − The cap is black/navy rather than the requested yellow.

FLUX.1 Kontext [dev]

• + Successfully included all prompt elements including the businesswoman and the cell phone.
• + The yellow cap color matches the prompt perfectly.
• + Superior composition that allows the viewer to see both the driver and the passenger clearly.

• − The capybara's face is slightly less 'photorealistic' and more anthropomorphized than Model A.
• − One of the front paws is resting on its lap rather than both being on the steering wheel.

DALL-E 3

• + Excellent gothic aesthetic with intricate 3D-like textures and layered composition.
• + Highly cinematic lighting that makes the elements feel integrated and atmospheric.
• + Rich detail in the border elements including webs, thorns, and twisted trees.

• − Text rendering is almost entirely illegible gibberish, failing the technical prompt requirements.
• − Layout is overly cluttered, making it difficult to read as a functional invitation.

FLUX.1 Kontext [dev]

• + Most text is legible and follows the prompt instructions for specific dates and locations.
• + Clean, high-contrast layout that functions well as a graphic design piece.
• + Follows the specific request for a banner scroll and thorns border accurately.

• − The 'dark parchment' texture requested in the prompt is missing in favor of a plain black background.
• − The scroll text contains minor spelling errors and artifacts despite the main headings being clear.

DALL-E 3

• + Excellent execution of the isometric 3D diorama style.
• + High level of detail with multiple sushi types and realistic lighting.
• + Clean and visually appealing composition.

• − Completely ignored the request for large bold text at the top-center.
• − Failed to place 'JAPAN' and 'SUSHI' text in the image.

FLUX.1 Kontext [dev]

• + Successfully followed text instructions for 'JAPAN' and 'SUSHI'.
• + Accurate 45-degree isometric perspective.
• + Clean, minimalist interpretation of the cartoon scene.

• − The flag icon is abstract and does not represent the Japanese flag well.
• − Very simplistic 3D modeling compared to the level of detail requested in 'refined textures'.
• − The diorama base is extremely basic.

DALL-E 3

• + Excellent adherence to the list of specific animals (puppy, kitten, bunny, fox).
• + Beautiful lighting effects including god rays and dew sparkles as requested.
• + Very high level of detail in the fur texture and background elements.

• − The butterflies have weirdly mammalian 'hamster-like' heads which is anatomically confusing.
• − Style leans more towards digital illustration/CGI than 'hyper-photorealistic'.

FLUX.1 Kontext [dev]

• + Natural, realistic lighting and photographic depth of field.
• + Dynamic poses that accurately capture the 'playfully chasing' and 'tumbling' prompt requirements.
• + Anatomically correct butterflies and more realistic fur textures.

• − Failed to include all requested animals; missing the bunny and the red fox kit.
• − The three animals included look very similar in color and style, reducing the variety requested.

DALL-E 3

• + Excellent execution of the 'vintage vector emblem' style with complex textures
• + Rich, warm color palette that fits the requested theme perfectly
• + High-quality typography and balanced composition

• − Failed to include the specific name 'Caffè Florian', replacing it with generic text
• − The banner does not contain the 'Est. 1720' text as requested, placing it at the bottom instead

FLUX.1 Kontext [dev]

• + Perfect adherence to the specific text requested including 'Caffè Florian'
• + Clean, minimalist design that follows the prompt's layout requirements
• + Accurate depiction of a styled cloche dome and steam icon

• − Lack of the 'banner' element specified in the prompt
• − Almost zero texture, appearing more like a modern flat icon than a vintage emblem
• − The font choice is a bit too modern for a 'vintage' and 'classic' request

DALL-E 3

• + Excellent adherence to the NASA-inspired color palette.
• + Sophisticated, professional layout that feels like a real museum-quality infographic.
• + Highly detailed icons and complex vector-style shading.

• − Generated three variants in one image instead of a single poster.
• − Incorrectly includes Space Shuttle-style icons which are anachronistic to Apollo 11.
• − Text is largely illegible filler content.

FLUX.1 Kontext [dev]

• + Follows the specific 6-step structure requested in the prompt more clearly.
• + Minimalist flat-vector style is consistent throughout the icons.
• + Correctly attempts to label the steps as 'Launch' and 'Descent'.

• − Significant spelling errors ('APOLO', 'DESCNT', 'TRANGULIITY').
• − Text rendering is messy with overlapping characters and gibberish strings.
• − Composition is quite basic and lacks the 'modern infographic' aesthetic.