Head to head: Wan vs Krea 3 Edit
Wan vs Krea 3 Edit
By Ryan Merket · Published
A head-to-head across eight image tasks, from expressive documentary scenes to exact counts and clean, legible text. The results turn on how consistently each model follows complicated visual instructions—not just how polished its images look.
Krea 3 Edit takes the matchup, winning five tasks to Wan’s two, with one tie. Its advantage is clearest when a prompt asks for several details to land at once: in the night repair convoy, it brings together the open luggage bay, roadside repair, map-sharing passengers, child in the support vehicle, and radio-using patrol rider. Wan’s convoy looked polished, but missed or obscured key requested interactions. Krea also comes out ahead on the perspective, poster-text, empty-tram-avenue, and salvage-scooter tasks. The poster is a particularly clean example: Krea keeps the exact wording in the requested three centered lines, while Wan breaks the title across two. For the scooters, Krea makes all seven individually countable and gives them numbered inspection tags; Wan’s more prominent branding doesn’t overcome the overlap and less evident count. Wan has real wins, not just consolation points. Its bracelet image better captures a candid tying gesture, with plausible fingers and soft daylight, and its seven espresso-filled cups read more clearly than Krea’s cups, which appear empty. The anxious ferry-captain result is a tie: Wan makes the captain, throttle, and storm details easier to read, while Krea gives the face a stronger anxious expression but loses some clarity and the requested cool-dawn lighting. The aggregate score—66.2 for Krea against 58.2 for Wan—and limited confidence support a clear overall verdict, even though individual prompts are not one-sided. **Krea 3 Edit wins: it is the more reliable choice across varied instructions, especially when a scene depends on multiple specific elements being present and legible.**
Anxious ferry captain
A documentary photojournalism close-up of a middle-aged ferry captain on the rain-speckled bridge of the coastal vessel Marrowfin, gripping a brass throttle with visibly tense hands and wearing a distinctly anxious expression—wide eyes, tight jaw, raised brows, lips slightly parted as if holding breath—while blurred gray waves and a red harbor beacon show through the windows behind him; cool dawn light, natural skin texture, shallow depth of field, cinematic 16:9 composition.
OpenAI: GPT-6 Luna: Model A more clearly conveys a ferry captain through the uniform and cap, with a visibly anxious expression and an unobstructed grip on the brass throttle; the rain, waves, and red beacon also read clearly. Model B has the tighter close-up and strong natural texture, but the throttle and captain identity are less legible, and the lighting is notably darker than the requested cool dawn feel. (Second judge pass, order swapped — scores are the average of both: Model B better matches the requested d | Anthropic: Claude Fable 5.1: Model B nails the anxious expression (raised br…
Perspective & scale
A photorealistic one-point-perspective shot down a long empty library aisle, shelves receding to a single vanishing point, books getting consistently smaller with distance and nothing warped or wrongly sized, warm overhead light, 16:9.
OpenAI: GPT-6 Luna: Model A delivers a clear, warm one-point library aisle, while Model B has especially balanced symmetry, convincing scale recession, and a more polished composition. Both closely match the prompt, though Model B’s long parquet floor and centered distant window strengthen the sense of depth. (Second judge pass, order swapped — scores are the average of both: Model A has the more precise, symmetrical one-point perspective, with shelf lines and overhead lights converging cleanly and the books consis | Anthropic: Claude Fable 5.1: Both deliver a clean one-point perspective with…
Night repair convoy
A coherent multi-subject documentary photojournalism scene at a desert roadside breakdown at night: a turquoise intercity coach with its luggage bay open, a mechanic kneeling by the front wheel changing a hose, the driver holding a flashlight over the engine area, two tired passengers sharing a map on a suitcase, a child asleep across two station-wagon seats visible through a nearby support vehicle’s window, and a highway patrol motorcycle parked at an angle with its rider speaking calmly into a radio; sodium-vapor and headlight illumination, strong visual hierarchy, believable spacing and interaction, cinematic 16:9.
OpenAI: GPT-6 Luna: Model B includes nearly all requested interactions, especially the open luggage bay, roadside repair, map-sharing passengers, child visible in the support vehicle, and patrol rider using a radio. Model A is visually polished but misses or relocates several key details, including the child in the support vehicle and the clear luggage-bay repair. (Second judge pass, order swapped — scores are the average of both: Model B brings together nearly all requested interactions, including the map-sharing | Anthropic: Claude Fable 5.1: Model B delivers a convincing photojournalistic…
Hands & anatomy
A natural candid photo of a person tying a friendship bracelet, both hands clearly visible with anatomically correct fingers and a plausible grip on the threads, soft daylight, shallow depth of field, 16:9.
OpenAI: GPT-6 Luna: Model A better captures the soft daylight and shallow depth of field, with both hands plausibly pinching and guiding the threads. Model B makes the braided bracelet especially clear, but its deeper focus and more posed overhead composition fit the candid-photo prompt less well. (Second judge pass, order swapped — scores are the average of both: Model A more clearly captures a candid moment of tying a colorful friendship bracelet, with both hands plausibly pinching and guiding the threads in soft | Anthropic: Claude Fable 5.1: Model A delivers the requested soft daylight an…
Empty tram avenue
A documentary-style urban scene of Alder Quay Avenue at blue hour, glistening after light rain, with steel tram tracks curving between amber-lit shopfronts and overhead power lines reflected in puddles, a lone timetable kiosk and wet bollards along the sidewalk, and absolutely no trams anywhere in the frame; wide-angle perspective, crisp atmospheric detail, balanced 16:9 composition.
OpenAI: GPT-6 Luna: Model A more completely delivers the requested scene, with a clear timetable kiosk, wet bollards, overhead wires, curving tracks, and strong puddle reflections in a balanced wide-angle composition. Model B is atmospheric and convincing, but its kiosk is less distinctive and Model A’s brighter, crisper detail gives it the edge despite imperfect sign text. (Second judge pass, order swapped — scores are the average of both: Model A more clearly presents the requested avenue, timetable kiosk, wet bo | Anthropic: Claude Fable 5.1: Model B delivers a convincing documentary-style…
Legible multi-line text
A minimalist event poster with three lines of crisp, correctly-spelled text stacked and centered: 'NIGHT MARKET' large on top, 'Fridays · 6–11pm' in the middle, 'Riverside Pier 4' at the bottom, on a deep navy background, clean sans-serif, subtle grain.
OpenAI: GPT-6 Luna: Model B follows the requested three-line centered layout and renders all text correctly, while Model A splits the title across two lines. Model A gives the title stronger emphasis, but Model B’s restrained typography and balanced spacing better match the minimalist poster brief. (Second judge pass, order swapped — scores are the average of both: Model B presents the exact wording in three crisp, centered lines on a deep navy, subtly textured background. Model A is also clean and legible, but bre | Anthropic: Claude Fable 5.1: Both render all three lines correctly spelled a…
Exact counting
A clean overhead flat lay on white marble of EXACTLY seven identical espresso cups arranged in a single straight row, each clearly separate and individually countable, even soft lighting, 16:9.
OpenAI: GPT-6 Luna: Model A shows exactly seven identical, espresso-filled cups in a clean straight row on white marble, while Model B also has seven cups but they appear empty and their handles nearly meet the neighboring cups. Model A’s arrangement is especially clear and polished. (Second judge pass, order swapped — scores are the average of both: Model B and Model A both show exactly seven cups in a straight overhead row on white marble. Model A makes the espresso cups more immediately recognizable with visible | Anthropic: Claude Fable 5.1: Both Model A and Model B deliver exactly seven…
Seven salvage scooters
A highly detailed studio product photograph of exactly seven distinct vintage motor scooters from the fictional brand Virello, lined in a gentle arc on a matte concrete floor, each one a different color and clearly separated so all seven are individually countable, with chrome mirrors, worn leather seats, and small numbered inspection tags visible; softbox lighting, clean shadows, neutral backdrop, ultra-sharp realism, 16:9.
OpenAI: GPT-6 Luna: Model B makes all seven scooters individually countable and prominently includes numbered inspection tags, with a gentle arc and convincing worn details. Model A shows the Virello branding more clearly, but its scooters overlap heavily and the inspection tags and exact count are less evident. (Second judge pass, order swapped — scores are the average of both: Model B shows seven individually numbered scooters in a gentle arc, with visibly worn seats and a convincing salvage character. Model A is | Anthropic: Claude Fable 5.1: Model B delivers exactly seven clearly separate…
Matchup powered by OpenRouter.