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 14 shared challenges, with an AI judge weighing in on each.
HiDream I1 Full
#60 of 62 in Text-to-Image
OmniGen v2
#57 of 62 in Text-to-Image
Where the votes landed
HiDream I1 Full
100.0%
win rate
Ties
0.0%
OmniGen v2
0.0%
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 reflections and refractions of the background plant
- + Realistic lighting and shadow integration with the wooden table
- + Accurate textures on the book and sphere
- − The sphere is floating unnaturally in the center of the cube
- − The glass cube has multiple conflicting reflections of the sphere
OmniGen v2
- + Natural placement of the sphere resting on the bottom of the cube
- + Clean, high-resolution aesthetic with vibrant colors
- + Captures all prompt elements accurately including the background plant
- − The glass lacks realistic thickness and refraction on the bottom edge
- − Shadows on the table are less complex than the other model
Verdict: Both models followed the prompt perfectly, including the spatial relationships of the objects. HiDream I1 Full offers more realistic optical physics and lighting, but OmniGen v2 provides a more grounded composition where the sphere sits naturally on the surface rather than floating.
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 adherence to the 'candid' street photography style with an urban background
- + Rich, realistic wet pavement reflections and convincing lighting
- + Captures an authentic 'repairing' posture and expression
- − Anatomical issues with the man's legs and feet blending into the bike frame
- − Severe physical anomalies in the bicycle structure (missing front fork/wheel connection)
OmniGen v2
- + Much cleaner bicycle anatomy and character rendering
- + Effective representation of light rain particles
- + Good color contrast between the red bike and green foliage
- − Missed the 'urban street' feel, looks more like a park or residential road
- − The man is standing still holding the bike rather than 'repairing' it
- − Cars in the background lack the requested motion blur and look static
Verdict: HiDream I1 Full captured the requested cinematic atmosphere, shallow depth of field, and urban 'candid' feel much better than OmniGen v2, which looks more like a staged AI portrait. However, HiDream I1 Full suffered from significant structural errors in the bicycle and the man's lower body, whereas OmniGen v2 produced a technically cleaner but less prompt-accurate image.
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
- + Excellent details on the ornate engraving and leather straps.
- + Captures the 'battle-worn' aesthetic effectively with distinct scars and grit.
- + Striking lifelike eyes with realistic lighting reflections.
- − The scars appear slightly over-stylized/drawn on rather than natural skin texture.
OmniGen v2
- + Beautiful warm lighting and bokeh effect.
- + Clean, high-quality facial rendering and smooth skin transitions.
- + Accurate interpretation of the hair beads and braided style.
- − The subject lacks the 'battle-worn' appearance, looking too pristine and clean.
- − The armor textures are somewhat soft and lack the specified 'engraved plate' complexity.
Verdict: HiDream I1 Full is the clear winner for its superior attention to the specific keywords like 'battle-worn', 'engraved plate', and 'leather straps', creating a much more detailed and gritty atmosphere. While OmniGen v2 produces a high-quality portrait, it misses the mark on character personality, presented a subject that looks more like a cosplayer than a weathered paladin.
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
- + Legible and clean sans-serif typography.
- + Logical, structured layout that resembles a real menu page.
- + Consistent food photography style.
- − Shows only pizza images despite different section headers.
- − Minimalist design is perhaps a bit too plain.
OmniGen v2
- + Vibrant color accents and varied food items as requested.
- + Creative grid layout with interesting graphic elements.
- + Diverse food photography including soup and different pizzas.
- − Poor typography with significant spelling errors like 'RESTAURATED MENTS'.
- − The text blocks are illegible squiggles rather than characters.
- − Layout feels cluttered and less professional for a 'clean' aesthetic.
Verdict: HiDream I1 Full produces a much more functional and professional menu design with legible text and a coherent structure, although it fails to vary the food types. OmniGen v2 captures the 'vibrant accents' and 'grid' prompt requirements better but fails significantly on typography and basic spelling, rendering it unusable as a design template.
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
- + Excellent chalk texture and realistic handwriting aesthetic
- + Clean and professional composition with a nice illustration
- + High legible title with perfect date rendering
- − Completely failed to follow the requested menu items
- − Text in the body of the menu is nonsensical gibberish
OmniGen v2
- + Successfully attempted the specific menu items requested in the prompt
- + Good implementation of natural variations in letter size and slant
- + All handwriting feels authentically messy like real chalk
- − Contains spelling errors like 'SPECALS' and 'glite free'
- − Text rendering is somewhat jumbled with overlapping letters
- − The frame and board texture look less realistic compared to the other model
Verdict: HiDream I1 Full produces a much more visually pleasing and realistic-looking chalkboard, but it ignores the specific menu content requested by the user. OmniGen v2 successfully includes the requested dishes (Truffle Mushroom Risotto, Grilled Octopus) but suffers from poor spelling and messy text overlaps. OmniGen v2 is the preferred model for this specific task as it follows the semantic instructions of the prompt while HiDream I1 Full relies on generic placeholder text.
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 cinematic lighting and atmosphere with a detailed planetary backdrop.
- + Dynamic action pose conveying movement and speed.
- + High level of texture detail on both the space suit and the horse's mane.
- − Failed the negative constraint; the astronaut is on top of the horse, not vice versa.
OmniGen v2
- + Clean, illustrative style with clear subject-background separation.
- + Logical placement of the astronaut on the horse's back.
- − Failed the negative constraint; the astronaut is on top of the horse, not vice versa.
- − The image feels static and lacks the cinematic depth requested.
- − Anatomical issues with the horse's legs and the saddle placement.
Verdict: Both models failed the specific instruction to place the horse on top of the astronaut (a common logic-trap in prompt engineering), interpreting it instead as a standard astronaut riding a horse. HiDream I1 Full is the clear winner as it provides a much more cinematic, detailed, and visually compelling image with dynamic lighting, whereas OmniGen v2 produced a flat, static image with several anatomical errors.
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 adherence to the 'front paws on steering wheel' instruction
- + High-quality fur texture and realistic passenger skin tones
- + Clear and legible 'TAXI' text on the cap
- − The passenger's hand holding the phone has some anatomical distortion
OmniGen v2
- + Good bokeh effect for the city lights in the background
- + Captures the bored expression of the passenger well
- − Failed the anatomy of the driver by giving the capybara human hands
- − The capybara's head is awkwardly composited onto the body
- − Both the passenger and driver appear to be sitting in the front area of the car
Verdict: HiDream I1 Full is the clear winner as it successfully rendered animal paws on the steering wheel, whereas OmniGen v2 gave the capybara human hands. HiDream I1 Full also achieved a more realistic integration of the animal head and the clothing, while OmniGen v2 suffered from significant perspective and layout issues.
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
- + Clean layout with clear gothic motifs like cobwebs and trees
- + Includes the scroll banner as requested in the prompt
- + High visual clarity and great jack-o'-lantern rendering
- − Text accuracy is very poor, with jumbled letters for the date and location
- − Missing the word 'Invitation' in the main title
OmniGen v2
- + Successfully renders the full title 'Halloween Party Invitation'
- + Includes much more accurate event details including the date and location
- + Captures a stronger cinematic atmosphere with the moon glow behind the pumpkin
- − Banner text is essentially gibberish
- − The 'thorns' in the border are less distinct than in Model A
Verdict: OmniGen v2 is the superior choice because it successfully followed the complex text instructions for the date, time, and location, which are critical for an invitation. While HiDream I1 Full has a slightly cleaner aesthetic, its failure to render legible event details makes it non-functional as a template.
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 3D modeling with realistic PBR textures for the rice and fish.
- + Solid diorama base design that feels cohesive with the isometric style.
- + Clean and professional typography for the text elements.
- − Missed the small flag icon requested in the prompt.
- − The text is slightly off-center to the left.
OmniGen v2
- + Included the requested flag icon above the text.
- + Well-balanced composition that is perfectly centered.
- + Good 45-degree isometric perspective as requested.
- − The sushi looks less realistic and more plasticky compared to the other model.
- − The flag icon provided does not represent the Japanese flag incorrectly.
- − The text has some minor artifacts on the drop shadow.
Verdict: HiDream I1 Full produces a much higher quality 3D render with superior textures and lighting, making the sushi look appetizing and professional. While OmniGen v2 followed more specific instructions like the flag icon and centering, the actual quality of the 3D assets and text rendering in HiDream I1 Full is significantly better.
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
- + Stronger adherence to the 'fox' request with more realistic features
- + Detailed texture in the fur and surrounding foliage
- + Excellent use of lighting, depth of field, and god rays
- − Missing the baby bunny requested in the prompt
- − Includes two kittens instead of one kitten and one bunny
OmniGen v2
- + Bright, vibrant colors that capture the 'joyous' vibe
- + Captures the sense of a golden sunrise effectively
- − Missing both the baby bunny and the fox; the right-most animal looks like a stylized hybrid
- − Visual style is more illustrative/AI-cartoonish than 'hyper-photorealistic'
- − Anatomical issues, particularly with the paws and the floating butterfly on the left
Verdict: HiDream I1 Full produces a much higher quality image with better textures and a more realistic aesthetic that aligns with the 'hyper-photorealistic' prompt, although it fails to include the bunny. OmniGen v2 fails significantly on prompt adherence by missing two of the four specified animals and results in a flat, cartoonish style with several anatomical artifacts.
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
- + Captures the Studio Ghibli art style more authentically in terms of linework and character design.
- + Includes nice details like the straw hat and floral patterns that fit the aesthetic.
- − Completely changes the core dynamic and expressions of the 'Distracted Boyfriend' meme.
- − Fails to preserve the specific clothing and accessories of the source image.
OmniGen v2
- + Excellent preservation of the source image's layout, clothing patterns, and pose.
- + Successfully translates the specific facial expressions (the look of betrayal/shock) into an anime style.
- + Cleverly retains the background bus and urban structure.
- − The art style is closer to generic modern anime than the specific hand-painted Ghibli aesthetic requested.
- − Lines are a bit too clean and lack the hand-painted texture mentioned in the prompt.
Verdict: While HiDream I1 Full captures the Ghibli 'look' more closely in terms of artistic rendering, it completely fails the task of editing the provided image, essentially generating a new scene. OmniGen v2 is the winner because it successfully transforms the meme while preserving all key identifying elements including the poses, specific clothing patterns, and the humorous narrative of the source.
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
- + Effective hair-blow effect
- + Natural-looking flying leaves
- − Completely lost the dog from the original image
- − Substantial changes to the subject's face/identity
- − Altered the background and layout significantly
OmniGen v2
- + Perfectly preserved the original subject, dog, and background
- + Successful hair-blow effect
- + High visual clarity and cleanliness
- − Leaves look a bit artificial and sparse
- − Movement is less 'energetic' compared to the other model
Verdict: OmniGen v2 is the clear winner as it successfully applied the requested edits while maintaining the integrity of the source image, including the dog and the woman's face. HiDream I1 Full failed the core editing task of source preservation by removing the dog entirely and regenerating a new person.
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
- + Strong minimalist vector aesthetic
- + Clean and professional graphic design
- + Perfect rendering of the banner and date text
- − Completely missing the primary name typography (Caffè Florian)
- − Literal interpretation puts a cup inside a cloche which looks slightly cluttered
OmniGen v2
- + Successfully included the restaurant name typography
- + Excellent minimalist line art for the cloche dome
- + Balanced and centered composition
- − Misspelled the name as 'CAFFFLORIN'
- − Logo structure is slightly more generic compared to the first
Verdict: HiDream I1 Full produces a much higher quality vector emblem with professional aesthetics, but failed to include the primary name requested. OmniGen v2 attempted all text elements but suffered from spelling errors and a slightly less refined 'vintage' texture. HiDream I1 Full is the winner for its superior visual polish and adherence to the 'vintage minimalist' style, despite the missing text.
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
- + Successfully captures the requested navy, white, and red color palette
- + Illustrations of the Saturn V and rocket plumes are aesthetically pleasing and fit the vector style
- + Main title text 'APOLLO 11' is clear and well-aligned
- − Failed to follow the requested 6-step logical sequence of an infographic
- − Text labels underneath icons are nonsensical or literally repeat prompt instructions (e.g., '+ ICON')
- − Layout is cluttered with overlapping elements that do not represent a clear timeline
OmniGen v2
- + Attempts a grid-based infographic layout which feels more organized
- + Consistent iconography across the six slots provided
- + Closer to a 'flat-vector' style with minimal complexity
- − Significant spelling errors in the main title ('APOLO 17' instead of 'Apollo 11') and across all labels
- − The icons do not accurately represent the specific requests (e.g., Saturn V icon, Lunar Module icon)
- − Source agency acronym is incorrect ('NSA' instead of 'NASA')
Verdict: HiDream I1 Full produces much better individual illustrations and cleaner main text, but it fails to create a coherent infographic timeline. OmniGen v2 attempts a better infographic structure, but provides numerous spelling errors, incorrect mission numbering, and generic icons. HiDream I1 Full is the winner for its superior visual quality and adherence to the artistic style, despite the nonsensical secondary labels.
Explore each model
Unified multimodal model for text-to-image generation, instruction-guided image editing, personalized generation, and virtual try-on