Unified multimodal model for text-to-image generation, instruction-guided image editing, personalized generation, and virtual try-on
Settled by community votes across 13 shared challenges, with an AI judge weighing in on each.
OmniGen v2
#56 of 62 in Text-to-Image
Recraft V4
#11 of 62 in Text-to-Image
Where the votes landed
OmniGen v2
0%
win rate
Ties
0%
Recraft V4
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
OmniGen v2
- + Excellent clean aesthetic with high resolution.
- + Accurate placement of the red book and blue sphere.
- + Convincing glass reflections and realistic soft lighting.
- − The glass cube appears more like an open container or box due to the thick top rim.
- − The plant is very blurred in the background, reducing the 'visible through the glass' effect.
Recraft V4
- + Strong adherence to the 'partially visible through the glass' prompt element.
- + Highly realistic textures on the wooden table and the aged book pages.
- + Correct lighting direction from the left window.
- − The blue sphere appears to be floating in the center of the cube rather than resting on the bottom.
- − The cube has some structural inconsistencies in the glass thickness at the corners.
Verdict: Both models followed the prompt closely, but OmniGen v2 produced a cleaner, more aesthetically pleasing commercial-style photograph. Recraft V4 captured the specific detail of the plant visible through the glass more effectively and provided richer textures, although the floating sphere feels less natural than the placement in OmniGen v2.
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
OmniGen v2
- + Excellent color vibrance and clear, sharp subject focus.
- + Strong symmetrical reflections on the wet pavement.
- + Bicycle structure is clean and reasonably logical despite some gear floating.
- − Fails to include motion blur on passing cars as requested.
- − The man is standing still rather than actively 'repairing' the bike.
- − Composition feels too clean and centered for a requested 'imperfect framing' candid shot.
Recraft V4
- + Highly accurate adherence to the 'candid' street photography style with imperfect framing.
- + Successfully captures motion blur from passing cars and a sense of activity.
- + Realistic skin texture and posture that perfectly matches the 'repairing' prompt.
- − The actual bike structure is physically impossible, missing a frame connection to the front wheel.
- − The face/head area has some slight digital artifacts/blending issues with the hat.
Verdict: Recraft V4 captured the specific atmosphere of the prompt much better, delivering on the motion blur, candid framing, and the action of repairing. While OmniGen v2 produced a cleaner, more aesthetically 'perfect' image, it missed several technical prompt requirements and felt more like a posed portrait than a candid street photo.
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
OmniGen v2
- + Strong bokeh effect with vivid torchlight reflections
- + Clean and beautifully rendered face and hair
- + Excellent engraving details on the shoulder pauldron
- − Character looks pristine rather than battle-worn
- − Scars are depicted as simple dots like freckles
- − Lack of visible detail on the cloth underlayer
Recraft V4
- + Exceptional realism in skin texture, dirt, and realistic scarring
- + Highly accurate interpretation of battle-worn aesthetic
- + Intricate detail on leather straps, buckles, and woven cloth underlayer
- − The bokeh sparks are slightly less prominent than Model A
- − Composition is a bit tighter, cropping the top of the hair slightly
Verdict: Recraft V4 significantly outperforms OmniGen v2 in meeting the specific descriptive elements of the prompt, particularly the 'battle-worn' aesthetic and the fine textures of the leather and cloth. While OmniGen v2 produced a clean and aesthetically pleasing fantasy portrait, Recraft V4 captured the gritty realism and intricate detail requested, from the authentic skin blemishes to the weathered metal.
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
OmniGen v2
- + Strong use of vibrant accent colors and colorful background tiles.
- + Artistic food photography with good lighting and composition.
- + Follows the two-page spread layout typical of many menus.
- − Significant spelling errors in every heading (e.g., 'RESTAURATED', 'PIZZZZAN').
- − Text in the body is nonsensical greeking or garbled characters.
- − The grid layout is somewhat inconsistent across the two pages.
Recraft V4
- + Flawless text rendering with perfect spelling and pricing.
- + Very clean, professional minimalist layout with a clear grid.
- + Perfect adherence to the requested categories: Appetizers, Pizza, and Mains.
- − The design is very safe and lacks the 'vibrant accents' requested in the prompt beyond the orange pricing.
- − Shadows under the food items are a bit uniform/rendered.
Verdict: Recraft V4 is the clear winner as it produces a functional, professional menu with perfectly legible text and a logical layout. While OmniGen v2 has more artistic food photos and more creative use of color, the catastrophic spelling errors and illegible body text make it unusable for its intended purpose.
Magic Burger Explosion: Fiery Photorealism Challenge
Text-to-Image“Ad for 'Magic Burger'. Dynamic, exploded burger with all components (bun, patty, cheese, lettuce, tomato, sauce) suspended in mid-air. Emphasize photorealistic detail and a sense of motion. Dark, fiery background with glowing embers. Integrate text: 'MAGIC BURGER' as a prominent title, 'LIMITED TIME ONLY' as a secondary message, and '€6.99' in a starburst, all rendered with a fiery, glowing effect.”
AI Judge Analysis
OmniGen v2
- + Excellent typography rendering with clean, clear text.
- + Bright, high-impact colors suitable for a fast-food advertisement.
- − Failed the main prompt requirement for an 'exploded' burger, showing a fully assembled one instead.
- − The background is generic fire rather than a dark, atmospheric field with embers.
Recraft V4
- + Perfectly captured the 'exploded' view with suspended components as requested.
- + Highly photorealistic textures on the bun, meat, and vegetables.
- + Successfully integrated all text requirements including the specific Euro currency symbol.
- − The composition is a bit tight, with the bottom bun feeling slightly cut off by the frame edge.
- − The 'limited time only' text is smaller and slightly less legible than expected for an ad.
Verdict: Recraft V4 is the clear winner as it followed all complex instructions, particularly the 'exploded burger' requirement and the specific currency symbol, whereas OmniGen v2 produced a standard assembled burger. Recraft V4 also achieved a much higher level of photorealism and a more dynamic sense of motion and depth.
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
OmniGen v2
- + Very realistic chalk texture and thickness
- + Followed instructions for natural variations in letter size and slant
- − Major spelling errors like 'SPECALS' and illegible garbled text in the menu items
- − Failed to render the title in elegant cursive as requested
Recraft V4
- + Perfect text accuracy for the title and all three menu items
- + Excellent composition showing the cafe environment with illustrative chalk sketches
- + Consistent and legible handwriting style
- − The text looks slightly too uniform, almost like a digital font rather than natural chalk handwriting
- − Did not implement the 'elegant cursive' specifically for the title
Verdict: Recraft V4 is the clear winner because it correctly spells all requested menu items and numbers, whereas OmniGen v2 produced significant spelling errors and nonsensical garbled characters. Additionally, Recraft V4 provided a much better composition by placing the board in a realistic café setting with charming chalkboard illustrations.
The Reversed Rodeo
Text-to-Image“Horse riding astronaut in space — horse on top, not vice versa. Surreal, highly detailed, cinematic.”
AI Judge Analysis
OmniGen v2
- + Clean, vibrant colors and lighting.
- + Correctly identifies the core subject elements (astronaut and horse).
- − Fails the specific inverse-riding instruction (horse on top).
- − The horse's feet are cut off at the bottom of the frame.
- − The composition is very static and flat compared to the cinematic request.
Recraft V4
- + Excellent cinematic lighting and textures.
- + Dynamic composition with successful integration of space elements like asteroids.
- + High level of detail in the astronaut suit and horse's coat.
- − Fails the specific inverse-riding instruction (horse on top).
- − The horse's front leg anatomy is slightly distorted.
Verdict: Both models failed the negative constraint to have the 'horse on top' of the astronaut, defaulting to a standard rider configuration. However, Recraft V4 produced a vastly superior image in terms of cinematic quality, detail, and environmental depth compared to the flat, clip-art style of OmniGen v2.
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
OmniGen v2
- + Excellent photographic lighting on the capybara's fur
- + Good character expressions for both subjects
- − Failed the spatial layout by placing the passenger in the front seat
- − Disturbing anatomical error where the capybara has human hands
Recraft V4
- + Perfect adherence to character placement with the passenger in the back seat
- + Highly realistic cinematic film grain and rainy atmosphere
- + Accurate rendering of capybara paws on the steering wheel
- − The capybara's head is slightly large for the cabin space
Verdict: Recraft V4 is the clear winner as it followed all spatial instructions, correctly placing the passenger in the back seat and giving the driver animal paws rather than human hands. OmniGen v2 failed the primary prompt requirements by putting the passenger in the front seat and using human hands on the capybara, creating an uncanny effect.
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
OmniGen v2
- + Strong parchment effect and framing
- + Excellent main title typography
- + Vibrant cinematic lighting on the central pumpkin
- − Numerous text errors and garbled characters in the secondary text
- − Multiple conflicting event details at the bottom
- − Missing the thorns requested for the border
Recraft V4
- + Flawless text rendering for all requested details
- + Highly detailed and atmospheric artistic style
- + Includes all requested elements including webs, thorns, and bats
- − Parchment texture is less obvious as a physical poster than Model A
- − The 'scroll banner' is more of a modern ribbon tag than a traditional scroll
Verdict: Recraft V4 is the clear winner as it successfully rendered every piece of text and specific detail from the prompt with perfect accuracy. While OmniGen v2 has a nice graphic design layout, it failed significantly on text coherence and accuracy of the event details.
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
OmniGen v2
- + Excellent typography with clean bold text and 3D shadowing.
- + Vibrant cartoon aesthetic with smooth, refined textures.
- + Clear diorama base that fits the isometric miniature style.
- − The flag icon is generic and does not represent the Japanese flag.
- − The sushi design is biologically confusing, appearing as Nigiri with shrimp tails but filled like Maki.
Recraft V4
- + Accurately depicts the Japanese flag icon as requested.
- + Realistic PBR materials, especially on the salmon and ice base, are highly impressive.
- + Correctly renders the requested text with elegant spacing.
- − The text is thin rather than 'large bold' as specified.
- − The ice base is slightly messy compared to the 'ultra-clean' requirement.
Verdict: Recraft V4 followed the specific details of the prompt much better, correctly including the Japanese flag and providing a more realistic PBR material interpretation. OmniGen v2 has superior typography and a cleaner diorama base, but failed on the specific flag icon and provided nonsensical sushi anatomy.
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
OmniGen v2
- + Vibrant and warm lighting with clear god rays.
- + Very clean, simplified composition suitable for a greeting card aesthetic.
- − Failed to include all requested animals, missing either the bunny or the fox as the third animal is an ambiguous hybrid.
- − The style is highly stylized and cartoonish rather than the requested hyper-photorealistic style.
- − Characters are static and sitting rather than playfully chasing and tumbling.
Recraft V4
- + Successfully included all four requested animals: golden retriever, tabby kitten, bunny, and fox kit.
- + Captured the requested action of chasing and tumbling with dynamic poses.
- + Achieved a high level of photorealism with intricate fur textures, dew sparkles, and realistic lighting.
- − Some butterflies in the background appear as small, indistinct shapes.
- − The composition is quite busy with many elements competing for attention.
Verdict: Recraft V4 followed the prompt instructions near-perfectly, providing all four specific animals in a dynamic, photorealistic action scene. In contrast, OmniGen v2 failed the animal count and delivered a highly stylized, static image that ignored the 'hyper-photorealistic' and 'chasing' requirements of the prompt.
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
OmniGen v2
- + Clean vector-style execution
- + Clear and legible 'Est. 1720' text
- + Effective minimalist cloche design
- − Misspelled name as 'CAFFFLORIN'
- − Logo lacks the sophistication expected for a vintage luxury cafe
Recraft V4
- + Perfect text rendering with correct spelling and accent
- + Excellent vintage typography and artistic shading
- + Sophisticated composition that fits the requested style
- − Did not place 'Est. 1720' on a banner as requested
- − Background is white rather than the requested cream tone
Verdict: Recraft V4 is the clear winner due to its superior typography and correct spelling of the brand name, including the accent on 'Caffè'. While OmniGen v2 achieved a cleaner vector look and followed the banner instruction, the misspelling 'CAFFFLORIN' makes it unusable as a logo.
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
OmniGen v2
- + Successfully uses the requested navy and red color palette.
- + Achieves a clean, balanced layout with professional-looking spacers.
- − Incorrectly labeled the mission as 'Apolo 17' instead of 'Apollo 11'.
- − Text is largely gibberish with significant spelling errors.
- − The icons do not clearly represent the specific requested stages like a Saturn V or Lunar Module.
Recraft V4
- + Excellent text rendering with correct spelling for the mission name, stages, and crew names.
- + Strictly followed all six specific iconography steps requested in the prompt.
- + Superior flat-vector style that looks like a real infographic.
- − The layout is slightly scattered with icons not perfectly aligned on a grid.
- − Minimal use of the requested red color compared to Model A.
Verdict: Recraft V4 is the clear winner as it correctly followed every step of the detailed infographic prompt, providing specific and accurate icons for the Saturn V and Lunar Module while maintaining perfect text legibility. OmniGen v2 failed on basic facts, incorrectly identifying the mission as Apollo 17 and producing illegible text.
Explore each model
Recraft's latest text-to-image generation model with high-quality output, supporting various aspect ratios and custom color palettes