HiDream AI's 17B parameter text-to-image model using sparse diffusion transformer with mixture of experts, achieving state-of-the-art image generation quality with strong prompt following
Settled by community votes across 16 shared challenges, with an AI judge weighing in on each.
HiDream I1 Full
#60 of 62 in Text-to-Image
Wan 2.6
#28 of 62 in Text-to-Image
Where the votes landed
HiDream I1 Full
0%
win rate
Ties
0%
Wan 2.6
0%
win rate
Challenge by challenge
The strongest take from each model on every shared challenge, with the AI judge's read.
Geometric Composition
Text-to-Image“A glass cube on a wooden table. Inside the cube is a small blue sphere. On top of the cube sits a red book. A green plant is behind the cube, partially visible through the glass. Soft window light from the left.”
AI Judge Analysis
HiDream I1 Full
- + Excellent handling of glass thickness and internal reflections
- + The sphere appears to be floating inside the cube, creating a more artistic look
- + Accurate soft window lighting consistent with the prompt
- − The plant is very blurry and lacks definition compared to the foreground
Wan 2.6
- + High level of texture detail on the red book and wooden table
- + The blue sphere has realistic caustic light reflections on the bottom of the glass
- + Very clear and sharp composition throughout
- − The top edge of the glass cube under the book looks slightly distorted
- − The plant is positioned more to the side than directly behind the cube
Verdict: Both models followed the prompt perfectly. HiDream I1 Full produces a very clean, atmospheric image with superior glass physics, while Wan 2.6 offers more gritty, realistic textures in the wood and book as well as impressive light caustics. Wan 2.6 is the slightly stronger choice for its overall sharpness and tactile details.
Candid Street Photography
Text-to-Image“A candid street photo of an elderly Japanese man repairing a red bicycle in light rain, reflections on wet pavement, shallow depth of field, 50mm lens, natural skin texture, imperfect framing, motion blur from passing cars, cinematic but realistic, no stylization.”
AI Judge Analysis
HiDream I1 Full
- + Excellent handling of wet pavement reflections
- + Follows the shallow depth of field requirement well
- − Anatomical issues where the man's leg appears to pass through the bicycle frame
- − The bicycle geometry is nonsensical and broken
- − Misses the motion blur requirement for passing cars
Wan 2.6
- + Successfully captures motion blur from passing cars
- + High level of detail in skin texture and raindrops
- + Realistic 'candid' composition with tools on the ground
- − The man appears to be floating slightly above the pavement
- − Small anatomical artifacts on the fingers
Verdict: Wan 2.6 is the superior image as it adheres to more of the complex prompt requirements, specifically including the motion blur and a convincing 'candid' feel. HiDream I1 Full suffers from significant architectural errors in the bicycle and anatomical issues where the subject and object intersect incorrectly.
Fantasy Warrior
Text-to-Image“Close portrait of a battle-worn paladin in ornate engraved plate armor, hair braided with small beads, faint scars and dirt on the skin, warm torchlight reflecting off metal, shallow depth of field, bokeh sparks, lifelike eyes, highly detailed texture on leather straps and cloth underlayer.”
AI Judge Analysis
HiDream I1 Full
- + Strong symmetrical composition and intense facial expression
- + Includes detailed beads in the hair as requested
- + Ornate engravings on the plate armor are clearly defined
- − The scars look like surface-level scratches rather than skin-deep wounds
- − Lighting feels slightly artificial and flat across the face
Wan 2.6
- + Exceptional realism in skin texture, dirt, and lifelike eyes
- + Beautiful lighting and bokeh sparks that create a cinematic atmosphere
- + Excellent detail on the worn leather straps and frayed cloth underlayer
- − The character appears slightly younger than a typical 'battle-worn' paladin archetype
- − The braid beads are a bit oversized compared to the prompt's request for 'small beads'
Verdict: Wan 2.6 provides a significantly more lifelike and cinematic image with superior texture work on the skin, dirt, and leather materials. While HiDream I1 Full captures the 'battle-worn' expression well, it lacks the depth and atmospheric integration of the torchlight and bokeh found in the Wan 2.6 output.
Modern Clean Menu
Text-to-Image“Modern minimalist restaurant menu design, white background with colorful food photos in grid, sections for appetizers/pizza/mains, bold sans-serif fonts, vibrant accents, clean professional layout for casual dining.”
AI Judge Analysis
HiDream I1 Full
- + Strong heading legibility and clean font hierarchy.
- + Accurate high-quality food photography with consistent lighting.
- + Strict adherence to the white background and minimalist aesthetic.
- − Repetitive food photos (all appear to be variations of the same pizza).
- − The section headers are somewhat crowded together in the top half.
Wan 2.6
- + Excellent variety in food photos including salads and different pizza types.
- + Professional use of vibrant color accents in the corners and borders.
- + Sophisticated grid layout that maximizes the visual appeal of a menu.
- − Minor spelling errors in main headings (e.g., 'RESTAUR MENU').
- − Slightly messy text rendering in the finer price and description lines.
Verdict: Wan 2.6 is the superior choice because it captures the 'casual dining' energy and 'vibrant accents' requested in the prompt through a diverse grid of various food items. While HiDream I1 Full produces cleaner text, it fails the 'variety' aspect of the prompt by showing essentially the same pizza in every photo, making it feel less like a functional menu.
Chalkboard Menu
Text-to-Image“Handwritten-style chalkboard menu in a cozy café, all text rendered in the exact same realistic chalk handwriting style with natural variations in letter size, slight slant, and chalk texture — no printed or digital fonts anywhere on the board. Title at the top in elegant cursive chalk handwriting: ‘TODAY’S SPECIALS – APRIL 30, 2026’. Below it, three menu items also in the same handwritten chalk style: ‘Truffle Mushroom Risotto – $24’, ‘Grilled Octopus with Lemon & Herbs – $28’, ‘Brown Butter Chocolate Chip Cookies – $9’. At the very bottom, smaller text in the identical handwritten chalk style (slightly smaller but still clearly legible with the same handwriting characteristics): ‘All items made fresh daily • Ask about our gluten-free options’. Warm ambient café lighting, visible chalk dust and smudges, realistic handwriting imperfections, no clean printed text anywhere.”
AI Judge Analysis
HiDream I1 Full
- + Clean layout and centered composition.
- + Includes a nicely drawn chalk illustration of a coffee cup.
- − Failed to follow the text prompt for specific menu items.
- − Text consists mostly of nonsensical gibberish or misspelled words.
- − Font looks like a digital script rather than authentic chalk handwriting.
Wan 2.6
- + Excellent adherence to the prompt, rendering all requested text accurately including prices.
- + Highly realistic chalk texture with smudges and natural variations.
- + Authentic hand-lettered aesthetic that matches the 'cozy café' atmosphere.
- − The perspective is slightly angled, making the bottom text a bit smaller.
- − Light flare in the top left corner slightly obscures the first letter.
Verdict: Wan 2.6 is the clear winner as it perfectly rendered the specific menu items and prices requested, whereas HiDream I1 Full produced a board full of gibberish. Wan 2.6 also captured a much more realistic chalk texture and handwriting style compared to the digital-looking font in HiDream I1 Full.
The Reversed Rodeo
Text-to-Image“Horse riding astronaut in space — horse on top, not vice versa. Surreal, highly detailed, cinematic.”
AI Judge Analysis
HiDream I1 Full
- + Excellent sense of motion with the horse in full gallop
- + Striking composition with the curve of the planet mirroring the horse's movement
- + High clarity and clean rendering of the space suit
- − Anatomy of the rear legs is slightly messy near the hooves
- − Failed the specific negative constraint; the astronaut is on top of the horse
Wan 2.6
- + Beautiful cinematic lighting and vibrant nebulae in the background
- + Extremely detailed texture on the horse's coat and mane
- + Great rendering of the visor with light reflections
- − Failed the specific negative constraint; the astronaut is riding the horse conventionally
- − Minor artifacts where the reigns intersect with the astronaut's hand
Verdict: Both models failed to follow the logical reversal requested in the prompt (horse on top, not vice-versa), instead opting for the standard 'astronaut riding a horse' trope. However, Wan 2.6 is the superior image due to its more complex lighting, superior background details, and richer textures compared to HiDream I1 Full.
The Capybara Taxi Driver
Text-to-Image“Photorealistic scene inside a yellow New York taxi at night. A capybara is driving, wearing a yellow taxi driver cap and a dark jacket. It has a calm, professional expression and both front paws on the steering wheel. In the back seat sits a human businesswoman in a coat, looking at her phone with a completely normal, bored expression (as if this is just another normal ride). Through the windows you can see the streets of Manhattan at night with blurred lights. Realistic taxi interior, photorealistic, detailed fur and fabric, 35mm lens, night lighting with reflections, shallow depth of field.”
AI Judge Analysis
HiDream I1 Full
- + Excellent handling of the capybara's anatomy and fur texture
- + Clear and readable 'TAXI' text on the hat
- + Superior composition with a more intimate interior perspective
- − The passenger appears to be sitting in the front seat or mid-cabin due to the camera angle
- − The capybara's hands look slightly more primate-like than capybara-like
Wan 2.6
- + Successfully positions the passenger in the back seat as requested
- + Adds realistic atmosphere with rain on the windshield and vibrant city lighting
- + Highly detailed capybara facial features
- − The driver's hat looks more like a police officer's cap than a taxi driver's hat
- − The steering wheel placement is slightly awkward in relation to the capybara's body
Verdict: Both models followed the prompt well, but HiDream I1 Full produced a cleaner, more professional-looking image with better graphic clarity on the apparel. Wan 2.6 adhered better to the spatial request of placing the passenger specifically in the back seat and provided a more atmospheric scene, but failed on the specific detail of the 'taxi driver cap'.
The Halloween Invitation
Text-to-Image“Vintage gothic Halloween party invitation. Dark parchment poster, spooky border with webs and thorns, central glowing jack-o-lantern, bats, twisted trees, moody night sky. Add elegant gothic title text saying "Halloween Party Invitation", a small scroll banner saying "You are invited to a night of frights", and event details at the bottom: Date: 30.10.2026 Time: 7pm Location: The Arches, NYC Spooky but polished, cinematic lighting, square format.”
AI Judge Analysis
HiDream I1 Full
- + Features a clean, centered graphic design layout.
- + Follows the request for a dark parchment style poster.
- + The Jack-O-Lantern has high contrast and clean edges.
- − Fails significantly on text rendering, producing gibberish for the date and location.
- − Missing the requested thorns in the border.
- − The 'Halloween Party' text is incomplete, missing the word 'Invitation'.
Wan 2.6
- + Excellent text rendering, accurately including all requested details down to the address.
- + Strong cinematic lighting with realistic shadows and a complex 'thorns and webs' border.
- + Better adherence to the 'vintage gothic' and 'twisted trees' prompt elements.
- − The parchment texture is less like a 'poster' and more like a frame around a scene.
- − The banner text is legible but slightly small.
Verdict: Wan 2.6 is the clear winner as it successfully rendered every piece of text requested, including the specific date and location which HiDream I1 Full failed to capture. Wan 2.6 also provided a more atmospheric and cinematic interpretation of the 'gothic' and 'twisted trees' request, whereas HiDream I1 Full felt like a simpler clip-art style graphic with significant typos.
Bald man challenge
Image Editing“Give the person a full, thick head of natural hair with realistic texture, density, and a natural hairline. Preserve facial features and lighting.”
AI Judge Analysis
HiDream I1 Full
- + Attempts to add some hair length in the back.
- − Completely fails to preserve the source image, changing both the face and the background.
- − Adds nonsensical artifacts like a strange black object across the face.
- − Does not fulfill the prompt of a full, thick head of hair convincingly.
Wan 2.6
- + Perfectly follows the instruction to add a full, thick head of hair with realistic texture.
- + Excellent preservation of the original face, clothing, and background environment.
- + The hairline and hair integration with the glasses and ears are highly realistic.
- − The hair volume is slightly disproportionate to the original skull shape but remains aesthetically pleasing.
Verdict: Wan 2.6 is the clear winner as it successfully adds a realistic, thick head of hair while perfectly preserving the identity of the person and the background of the source image. HiDream I1 Full failed the challenge completely, essentially generating a new image with a different person, different background, and bizarre structural artifacts.
Isometric Miniature Diorama Scenes
Text-to-Image“Create a clear, 45° top-down isometric miniature 3D cartoon scene of Japan's signature dish: sushi, with soft refined textures, realistic PBR materials, gentle lighting, on a small raised diorama base with minimal garnish and plate. Solid light blue background. At top-center: 'JAPAN' in large bold text, 'SUSHI' below it, small flag icon. Perfectly centered, ultra-clean, high-clarity, square format.”
AI Judge Analysis
HiDream I1 Full
- + Excellent high-clarity 3D rendering with soft, refined textures.
- + Strong isometric perspective that fills the frame effectively.
- + Text is bold and well-rendered, although it misses the flag icon.
- − Missed the request for a small flag icon.
- − The 'JAPAN' text is slightly clipped at the top edges.
Wan 2.6
- + Perfect adherence to all prompt elements, including the small flag icon.
- + Excellent layout with a clearly defined raised diorama base and plate.
- + High-quality text rendering and clean minimalist aesthetic.
- − The scale of the sushi is a bit small within the overall composition compared to the text.
- − The shadow on the base plate is slightly harsh compared to the 'gentle lighting' request.
Verdict: Wan 2.6 is the winner as it followed all instructions perfectly, including the specific request for a small flag icon which HiDream I1 Full omitted. While HiDream I1 Full has more vibrant 3D textures, Wan 2.6 captures the 'miniature diorama' and 'graphic design' elements of the prompt more faithfully.
Over-the-top cartoon caricature
Editing“Create a caricature of me and my job. Make it exaggerated and humorous, incorporating my profession as a tv show anchor and my love for dogs and hockey.”
AI Judge Analysis
HiDream I1 Full
- + Maintains high facial similarity to the original source image.
- + Captures the 'tv show anchor' profession through a studio-like background and headset.
- + Excellent rendering of the denim texture in a caricature style.
- − Fails to incorporate the 'hockey' element of the request.
- − The dog's placement on the shoulder is a bit awkward and lacks interaction.
Wan 2.6
- + Successfully incorporates all elements: tv anchor, dogs, and hockey.
- + The caricature style is very vibrant and expressive.
- + Great composition that tells a story with the pug holding the hockey stick.
- − Lower facial similarity to the source image compared to the other model.
- − The transition from human hands to the hockey stick has some anatomical clipping issues.
Verdict: HiDream I1 Full does an excellent job of preserving the subject's likeness in a caricature style, but it completely misses the hockey requirement. Wan 2.6 fulfills the entire prompt much more creatively by building a full scene with dogs, hockey gear, and a news desk, despite the face looking less like the original person. Wan 2.6 is the winner for better prompt adherence and a more humorous interpretation of the request.
Adorable Baby Animals in Sunny Meadow
Text-to-Image“Hyper-photorealistic scene of fluffy baby animals—a golden retriever puppy, tabby kitten, baby bunny, and red fox kit—with big expressive eyes and ultra-detailed soft fur, playfully chasing butterflies and tumbling together in a lush wildflower meadow, warm golden sunrise light with god rays and dew sparkles, joyful wholesome vibe, 8K masterpiece.”
AI Judge Analysis
HiDream I1 Full
- + Features very cute, expressive eyes that align with the joyful vibe
- + Clean fur texture and vibrant colors
- + Good lighting that creates a soft, warm atmosphere
- − Failed to include the baby bunny requested in the prompt
- − Generated two kittens instead of one kitten and one bunny
- − Static posing misses the 'tumbling' and 'chasing' action requested
Wan 2.6
- + Successfully included all four specific animals: puppy, kitten, bunny, and fox
- + Excellent capture of motion and action, showing the animals tumbling and playing
- + Beautiful implementation of god rays and dew sparkles with high realism
- − The fox's front right leg/paw anatomy is slightly awkward during the stride
- − Some of the floating seeds/dust particles look a bit cluttered near the sun
Verdict: Wan 2.6 is the clear winner as it followed the complex prompt requirements perfectly, including all requested animal species and depicting the 'tumbling' movement. HiDream I1 Full failed on prompt adherence by omitting the bunny and providing two kittens instead, and its composition feels much more like a static pet portrait than an action scene.
Studio Ghibli Anime Style
Editing“Transform this photo into a Studio Ghibli–inspired illustration. Use soft pastel colors, hand-painted textures, gentle lighting, dreamy backgrounds, and a warm, nostalgic mood”
AI Judge Analysis
HiDream I1 Full
- + Strong Studio Ghibli character design aesthetic
- + Vibrant colors and high-quality digital line work
- + Captures the 'nostalgic' and 'dreamy' mood well
- − Low source preservation as it completely changes characters and clothing patterns
- − Loses the specific 'distracted boyfriend' meme expressions and interaction
- − Compositional elements like the background and number of people are altered
Wan 2.6
- + Excellent source preservation, keeping the exact pose, clothing, and expressions of the original meme
- + Beautiful watercolor textures and soft pastel color palette
- + Perfectly balances the Ghibli style with the source material's identity
- − The faces lean slightly more toward traditional shoujo manga than pure Ghibli
- − Added 'sparkle' artifacts might be distracting for some
Verdict: Wan 2.6 is the clear winner for this image editing task because it perfectly preserves the iconic composition and expressions of the 'distracted boyfriend' meme while applying a high-quality watercolor aesthetic. HiDream I1 Full creates a beautiful illustration, but it fails as an edit by fundamentally changing the characters' identities and the specific narrative moment of the original photo.
Golden Hour Stroll
Image Editing“Add dynamic motion to this photo: make hair blow in the wind, add leaves flying, energetic and lively feel.”
AI Judge Analysis
HiDream I1 Full
- + Excellent sense of motion and energy with the character's pose and flying leaves.
- + Captures a lively autumn vibe successfully.
- − Fails as an image edit because the subject's face, clothing, and environment are completely changed.
- − The original dog is completely missing from the scene.
Wan 2.6
- + Perfect source preservation, maintaining the identity of the woman and the dog exactly.
- + Accurately adds the requested wind-blown hair and flying leaves while keeping the background identical.
- + High visual consistency with the original image's lighting and tone.
- − The flying leaves look a bit like static overlays rather than being fully integrated with motion blur.
Verdict: Model B (Wan 2.6) successfully performed the image edit, preserving the subject, the dog, and the background while precisely following the instructions to add wind and leaves. Model A (HiDream I1 Full) fundamentally failed as an edit by generating a completely new image of a different person and removing the dog entirely.
Vintage Cafe Logo
Text-to-Image“Vintage minimalist restaurant logo for "Caffè Florian", retro cloche dome with steam and "Est. 1720" banner, classic typography, warm brown and cream tones, subtle texture on light background, vector emblem style.”
AI Judge Analysis
HiDream I1 Full
- + Clean vector aesthetic
- + Accurate rendering of the banner and date
- − Failed to include the primary brand name 'Caffè Florian'
- − The composition feels a bit empty at the bottom
Wan 2.6
- + Successfully included all requested text including 'Caffè Florian'
- + Excellent vintage paper texture on the background
- + Dynamic and professional logo composition
- − The 'Est. 1720' text is slightly small and less legible than Model A
Verdict: Wan 2.6 is the clear winner as it followed all prompt instructions, specifically including the brand name 'Caffè Florian' which HiDream I1 Full omitted entirely. Wan 2.6 also better captured the requested 'subtle texture' and 'vintage' feel through a more sophisticated layout and background treatment.
Apollo 11: Journey to Tranquility
Text-to-Image“Create a clean, modern vector infographic poster about the Apollo 11 mission. NASA-inspired palette (navy, white, muted red, light gray). Flat-vector style, crisp lines, consistent iconography, subtle gradients only. Steps (stop at landing): 1. Launch (Saturn Vicon) 2. Earth Orbit (Earth + orbit ring icon) 3. Translunar (trajectory arc icon) 4. Lunar Orbit (Moon + orbit ring icon) 5. Descent (lunar module descending icon) 6. Landing (lunar module on the surface icon) Small supporting elements (minimal text): • Crew strip: three silhouette icons with only last names: Armstrong, Aldrin, Collins. • Landing site marker: Moon pin labeled "Tranquility" only. Layout constraints: generous margins, large readable labels, clean background with subtle stars. Vector-only, print-poster look, high resolution.”
AI Judge Analysis
HiDream I1 Full
- + Strong adherence to the flat-vector infographic style.
- + Captures the requested navy, white, and red NASA color palette.
- + Includes multiple specific icons like the Saturn V and Earth with orbit rings.
- − Failed to include exactly 6 steps, only showing 4.
- − Text rendering contains errors including typos like 'UNG' and 'UNLY'.
- − Iconography is somewhat jumbled rather than following a logical sequence.
Wan 2.6
- + Correctly spells the names of the Apollo 11 crew members.
- + Maintains a clean navy and white aesthetic.
- − Completely failed to follow the 'Steps' requirement of the prompt.
- − Lacks the requested vector graphics and icons.
- − Composition is very sparse and does not resemble an infographic poster.
Verdict: HiDream I1 Full followed the core layout and stylistic instructions, creating a recognizable vector infographic with appropriate iconography, even if it missed some steps and had text glitches. Wan 2.6 failed to generate an infographic at all, producing a very simple starfield background with crew names. HiDream I1 Full is the clear winner for attempting the complex multi-part prompt.
Explore each model
Alibaba's multimodal generation model from the Wan AI suite, supporting text-to-video, image-to-video, reference-to-video with audio, and text-to-image, in both Chinese and English