FLUX.1 Kontext [dev] Black Forest Labs vs Wan 2.5 (Preview) Alibaba

FLUX.1 Kontext [dev]

• + Excellent adherence to the 'soft window light from the left' instruction
• + High clarity and realistic glass reflections
• + Clean, modern composition

• − The plant is more 'next to' the cube than 'behind' it as requested
• − The red book looks a bit like a plastic prop

Wan 2.5 (Preview)

• + Perfect placement of the plant behind the cube, visible through the glass
• + Realistic texture on the red book with worn edges
• + Strong sense of depth and atmospheric lighting

• − The base of the sphere has a slight clipping artifact with the bottom of the cube
• − There are small white dust artifacts scattered across the image

FLUX.1 Kontext [dev]

• + Handles raindrops on lens/visible rain well
• + Good center-focused composition for a portrait

• − The subject is standing with the bike, not repairing it
• − Anatomically incorrect hand positioning on the handlebars
• − Lack of 'imperfect framing' or 'motion blur' requested

Wan 2.5 (Preview)

• + Accurately depicts the 'repairing' action with tools and posture
• + Captures excellent natural skin texture and elderly features
• + Excellent reflections and shallow depth of field

• − Lacks obvious motion blur from cars
• − Composition is a bit clean for 'imperfect framing'

FLUX.1 Kontext [dev]

• + Ornate engraved plate armor is highly detailed with complex filigree.
• + Dramatic warm lighting creates a strong cinematic mood.
• + Effective shallow depth of field with soft bokeh.

• − Failed to include the requested braided hair with beads.
• − The skin appears somewhat smooth/plastic rather than displaying realistic dirt and scars.
• − The facial expression feels slightly static compared to the battle-worn prompt.

Wan 2.5 (Preview)

• + Excellent adherence to specific details like braided hair with beads and visible dirt/scars.
• + High-quality texture on the leather straps, tattered cloth, and mail.
• + Superior lifelike eyes that convey a weary, battle-worn emotion.

• − The torch in the background is a bit distracting and less 'soft bokeh' than the other sparks.
• − Engraving on the metal is elegant but slightly less intricate than Model A.

FLUX.1 Kontext [dev]

• + Features a bold, high-contrast editorial layout.
• + Photographs of food have realistic textures and appetizing lighting.
• + Uses a modern, asymmetric balancing of text and imagery.

• − Text is largely illegible and nonsensical.
• − The food imagery doesnt clearly map to the section titles provided.

Wan 2.5 (Preview)

• + Excellent adherence to the requested sections (Appetizers, Pizza, Mains).
• + Features a clean, professional grid that looks like a real menu.
• + Clear bold sans-serif fonts used for category headers.

• − Food photos are repetitive and look a bit artificial/plastic.
• − The red and green divider lines are a bit generic.

FLUX.1 Kontext [dev]

• + Excellent typography rendering for the primary title
• + High-quality realistic texture on the charcoal and flames

• − Failed to provide the 'exploded view' requested in the prompt
• − Minor spelling error in 'LNHLY'
• − Starburst price tag is a flat graphic rather than fiery and glowing

Wan 2.5 (Preview)

• + Perfectly executed the 'exploded' burger request with deconstructed components
• + Text integration follows the fiery/glowing aesthetic much better
• + Dynamic composition with a strong sense of motion

• − Smaller text is slightly less crisp than Model A
• − The 'melting' effect on the main title might be too intense for some advertising contexts

FLUX.1 Kontext [dev]

• + Features a clean, centered layout
• + Accurately represents a chalkboard texture
• + Completes the third menu item more naturally based on the partial prompt

• − Several spelling errors including 'Mashroom', 'Risoktso', and 'Octpus'
• − Severe repetition artifacts in the text like 'with with' and duplicated prices
• − Date text at the top is messy and illegible

Wan 2.5 (Preview)

• + Excellent spelling accuracy across all menu items and the date
• + Highly realistic chalk smudge effects and varied handwriting style
• + Followed instructions for elegant cursive-style font in the title

• − Omitted 'Herbs' from the second menu item
• − Line breaks in the third item are slightly awkward

FLUX.1 Kontext [dev]

• + Successfully followed the specific instruction for the horse to be on top of the astronaut
• + High anatomical clarity in the astronaut's face and the horse's head
• + Clean, cinematic lighting that feels surreal and focused

• − The horse's hind legs appear somewhat detached from the composition
• − Relatively simple background compared to the other model

Wan 2.5 (Preview)

• + Very vibrant and detailed space background with nebulae and planetary textures
• + Great sense of motion and dynamic composition

• − Completely failed the negative constraint/instruction for the horse's position
• − Produced a cliche image instead of the requested surreal reversal
• − The horse's back right leg is anatomically messy and blends into the atmosphere

FLUX.1 Kontext [dev]

• + Excellent photorealistic lighting and skin textures for the human passenger
• + Clear and accurate rendering of the capybara's fur and expression
• + Stronger sense of 'night' atmosphere with realistic urban lighting through the window

• − The capybara only has one hand on the steering wheel, failing the prompt's specific instruction
• − The taxi interior looks more like a modern personal car than a classic New York taxi

Wan 2.5 (Preview)

• + Perfect adherence to the instruction of having both front paws on the steering wheel
• + Captures the iconic New York taxi aesthetic effectively, including the roof sign and dashboard meter
• + The 'bored' expression of the passenger is very well-executed

• − The capybara's face looks slightly less realistic and more like a mask/person in a suit
• − Visible water droplets on the windshield are a nice touch but weren't requested

FLUX.1 Kontext [dev]

• + Accurate date and time fulfillment
• + Consistent gothic border style
• + Clean central jack-o-lantern illustration

• − Major text hallucinations in the banner and location details
• − Lacks the requested parchment and moody sky elements
• − Graphic design feels flat and less cinematic

Wan 2.5 (Preview)

• + Excellent text rendering with zero spelling errors across all fields
• + Rich cinematic lighting with a moody sky and parchment texture
• + Detailed spiderweb and thorn border as requested

• − The jack-o-lantern has some minor fire-particle artifacts that break the vintage feel slightly

FLUX.1 Kontext [dev]

• + Successfully added thick, full hair with realistic texture
• + Maintained the overall composition and jacket details

• − Significantly altered the facial features, making the person look younger and different from the source
• − Changed the glasses style and color from the original source image

Wan 2.5 (Preview)

• + Excellent source preservation, keeping the face and glasses identical to the original
• + Realistic hair texture and naturally integrated hairline
• + Maintained the original lighting and color balance

• − The hair volume is slightly less 'full' compared to the other model, though still meets the prompt

FLUX.1 Kontext [dev]

• + Excellent typography rendering for both words.
• + Clean, solid background as requested.
• + Strong minimalist aesthetic.

• − The 'flag icon' is incorrect and abstract.
• − The 3D model is over-simplified, looking more like a plastic toy than a 'refined texture' sushi.

Wan 2.5 (Preview)

• + Beautiful 3D render with high clarity and soft lighting.
• + Correct Japanese flag icon inclusion.
• + Better texture definition on the rice and salmon while maintaining the 'cartoon' feel.

• − The 'J' in 'JAPAN' has an unusual styling compared to the other letters.
• − Slightly less 'isometric' than Model A, opting for a standard 3D perspective.

FLUX.1 Kontext [dev]

• + Maintains the selfie perspective and clothing from the source image
• + Successfully incorporates the 'tv show anchor' profession via the background screen

• − Completely misses the 'hockey' requirement of the prompt
• − Text rendering for 'JOB' is random and the text on the dog tag is gibberish
• − The art style is more of a comic illustration than a caricature

Wan 2.5 (Preview)

• + Includes all requested elements: tv anchor desk, dogs, and hockey gear
• + Captures an excellent caricature likeness with exaggerated facial features and a 'big head' style
• + Creative integration of hockey, showing it on the monitor and including a stick

• − Loses the selfie/close-up composition of the original source image
• − Small artifacts in the dog's jersey and microphone details

FLUX.1 Kontext [dev]

• + Excellent sense of motion and dynamic posing
• + Clean and realistic lighting with natural-looking depth of field
• + Highly appealing and consistent character designs

• − Failed to include the fox kit and the rabbit
• − Replaced requested species with two extra kittens

Wan 2.5 (Preview)

• + Perfect prompt adherence, including all four specific animals requested
• + Beautiful inclusion of god rays and dew sparkles as described
• + Rich variety of wildflowers improving the composition

• − The fox kit's eyes have unnatural, slightly creepy glowing artifacts
• − Floating water droplets look a bit like digital artifacts rather than natural dew

FLUX.1 Kontext [dev]

• + Excellent preservation of the original body poses and clothing details.
• + Clean anime-style cel-shading and sharp line art.
• + Accurately recreates the character expressions from the meme.

• − Lean more toward generic modern anime than the specific hand-painted Ghibli aesthetic.
• − Colors are a bit too saturated for the 'soft pastel' request.
• − The background remains quite static and lacks the requested 'dreamy' texture.

Wan 2.5 (Preview)

• + Strong adherence to the Studio Ghibli aesthetic with soft lighting and a nostalgic mood.
• + Excellent use of pastel colors and painterly textures throughout the image.
• + Maintains the source composition perfectly while adding thematic elements like floating leaves.

• − The man's facial expression is slightly more somber than the original 'distracted' look.
• − The character in the red dress has slightly less defined hair compared to Model A.

FLUX.1 Kontext [dev]

• + Excellent preservation of the source image's lighting and character appearance.
• + Subtle and natural-looking hair motion that aligns with the wind direction.
• + Maintains the original facial details and dog's appearance perfectly.

Wan 2.5 (Preview)

• + Strong adherence to the instruction for 'hair blowing in the wind' with more pronounced volume.
• + Added more leaves across the frame to enhance the sense of motion.
• + Good preservation of the overall composition and background elements.

• − The added leaves look like static green stickers without motion blur or realistic integration.
• − The hair flow is a bit more chaotic compared to the source.
• − Noticeable softening of details on the dog's face compared to the original.

FLUX.1 Kontext [dev]

• + Very clean vector minimalism
• + Perfectly legible and accurate text rendering
• + Strict adherence to the two-tone color palette

• − Missed the request for a banner element
• − The steam icon is very abstract and barely recognizable
• − The cloche dome looks more like a building or cupcake wrapper

Wan 2.5 (Preview)

• + Successfully included the requested banner element
• + Beautiful textured background that hits the vintage vibe
• + Clearer representation of a cloche dome and steam

• − Minor text artifacts in the word 'Caffè'
• − Slightly less 'minimalist' than Model A due to gradients and multiple strokes

FLUX.1 Kontext [dev]

• + Stronger adherence to the flat-vector, minimalist iconography request.
• + Captures the requested muted NASA-inspired color palette well.

• − Significant spelling errors in every label including the main title.
• − The layout is cluttered and the steps are not in a logical chronological order.

Wan 2.5 (Preview)

• + Excellent visual clarity and high-quality vector illustrations.
• + Perfect adherence to the sequence of steps with logical flow and legible text.
• + Creatively included the crew names with appropriate iconography.

• − One of the astronauts appears to be wearing a helmet that looks more like a military helmet than a space gear.
• − The style is slightly more detailed than 'flat-vector', leaning into a semi-realistic illustration style.