Head to head: Boogu Image vs Krea 2 Large
Boogu Image vs Krea 2 Large
By Ryan Merket · Published
An automated head-to-head image matchup. Boogu Image scored 59.4, Krea 2 Large scored 58.6.
### Negation Winner: **Krea 2 Large**. OpenAI: GPT-5.6 Sol Pro: Model B cleanly includes the requested chair, books, tea, window, and warm light without forbidden interior decor, though it misses the required 16:9 format. Model A has stronger widescreen composition but appears to include cropped picture-frame elements at the top and right edges, and its book-spine text is visibly garbled. (Second judge pass, order swapped — scores are the average of both: Model B is more polished and avoids lamps and artwork, though its square format and visible outdoor vege | Anthropic: Claude Fable 5.1: Model A delivers the 16:9 format and all r… ### Quiet Fury Winner: **Krea 2 Large**. OpenAI: GPT-5.6 Sol Pro: Model B more convincingly depicts quiet fury, wet unblinking eyes, scratched glass, cyan vending-machine reflection, the eviction notice, and a recognizable jar of lychees within a richly detailed kiosk. Model A is polished and well framed, but the expression reads more sorrowful, the fruit is less clearly identifiable, and the malformed cyan lettering on the cheek is distracting. (Second judge pass, order swapped — scores are the average of both: Model B more completely captures the unmistakabl | Anthropic: Claude Fable 5.1: Model B conveys genuine fury with wet unbl… ### Neon Laundromat Winner: **Krea 2 Large**. OpenAI: GPT-5.6 Sol Pro: Model B delivers the stronger, more polished urban corner composition with convincing fogged windows, neon signage, vending machine, plaid bags, lone cone, and puddle reflections. Model A adheres better to the requested 16:9 format and exactly two washers, but its sparse diorama-like scene and extra cone feel less atmospheric and refined. (Second judge pass, order swapped — scores are the average of both: Model B offers a richer, more polished urban corner scene with strong fogged-glass atmosphe | Anthropic: Claude Fable 5.1: Model A matches the 16:9 ratio and two was… ### Courier Leap Winner: **Krea 2 Large**. OpenAI: GPT-5.6 Sol Pro: Model B captures the central airborne leap, whipping receipts, frozen rain, tilted energy, and tram-light motion far more dramatically, though it misses the requested 16:9 framing and renders the boxes somewhat unnaturally. Model A has cleaner realism, accurate storefront text, and correct framing, but the courier remains seated on the bicycle, substantially weakening the prompt’s defining action. (Second judge pass, order swapped — scores are the average of both: Model B better captures the req | Anthropic: Claude Fable 5.1: Model A delivers the correct 16:9 frame, c… ### Named art style Winner: **Boogu Image**. OpenAI: GPT-5.6 Sol Pro: Model A more fully realizes the requested dramatic cresting wave, fishing boat, dawn light, stylized foam, and muted indigo-and-cream ukiyo-e treatment in a balanced, dynamic composition. Model B is attractive and technically polished, but the oversized empty sky and cropped wave crest reduce visual impact and prompt specificity. (Second judge pass, order swapped — scores are the average of both: Model A presents a more dynamic, legible cresting wave, clearer dawn light, and stronger visual bala | Anthropic: Claude Fable 5.1: Model A delivers a clearly cresting wave w… ### Restricted palette Winner: **Boogu Image**. OpenAI: GPT-5.6 Sol Pro: Model A adheres much more closely to the required 16:9 format, flat treatment, and four-color palette, though the campsite is somewhat undersized. Model B is more detailed and visually engaging, but its square format, prominent sky gradient, and multiple tonal variations directly violate key restrictions. (Second judge pass, order swapped — scores are the average of both: Model A adheres much more closely to the required four-color palette, flat treatment, and 16:9 format, though its campsite is | Anthropic: Claude Fable 5.1: Model A honors the strict four-color palet… ### Spatial layout Winner: **Krea 2 Large**. OpenAI: GPT-5.6 Sol Pro: Model B more clearly and cohesively places the bed along the left wall, the rug centrally, the desk beneath the back-wall window, and the lamp toward the front-right, while maintaining a polished flat-vector isometric style. Model A largely follows the layout but has a less convincing front-right lamp placement, weaker spatial consistency, and a more washed-out presentation. (Second judge pass, order swapped — scores are the average of both: Model B clearly places every requested element, with a | Anthropic: Claude Fable 5.1: Model A places the rug more centrally and… ### Spice Stall Macro Winner: **Tie**. OpenAI: GPT-5.6 Sol Pro: Model A better matches the requested 16:9 diagonal framing, folded newsprint, clearly rendered “Lot 47-C” tag, dense market setting, and sharply resolved spice texture. Model B has convincing macro detail and a more visibly chipped spoon, but its square format, cropped utensil, less legible tag, and distracting foreground bag weaken adherence and composition. (Second judge pass, order swapped — scores are the average of both: Model A better satisfies the requested 16:9 diagonal framing, clearly | Anthropic: Claude Fable 5.1: Model B delivers a far more convincing macr… ### Negation Winner: **Krea 2 Large**. OpenAI: GPT-5.6 Sol Pro: Model A violates the plant prohibition with a prominent potted plant, but otherwise matches the requested objects, lighting, mood, and 16:9 format. Model B also shows extensive plants through the window and misses the required aspect ratio, though it has slightly cleaner technical rendering. (Second judge pass, order swapped — scores are the average of both: Model B better respects the key negation by avoiding an indoor plant, lamp, and artwork, though the outdoor greenery and square framing con | Anthropic: Claude Fable 5.1: Model A directly violates the core negatio… ### Quiet Fury Winner: **Tie**. OpenAI: GPT-5.6 Sol Pro: Model A better conveys restrained, tearful fury while clearly integrating the eviction notice, lychees, cyan vending-machine reflection, torn route map, and cramped kiosk setting into a cohesive narrative composition. Model B has stronger glass scratches and foreground props, but the expression reads as overt, exaggerated anger rather than quiet fury, and the station context is less developed. (Second judge pass, order swapped — scores are the average of both: Model A better captures restrained, | Anthropic: Claude Fable 5.1: Model A delivers torn route maps, a cyan v… ### Neon Laundromat Winner: **Krea 2 Large**. OpenAI: GPT-5.6 Sol Pro: Model B more convincingly realizes the laundromat with two visible machines, fogged glazing, vending machine, plaid duffel bags, a lone cone, and polished neon puddle reflections, while Model A has weaker interior detail and incorrectly includes two cones. Model A better matches the requested 16:9 format, but Model B has substantially stronger composition, aesthetics, and technical execution despite its square aspect ratio. (Second judge pass, order swapped — scores are the average of both: Mode | Anthropic: Claude Fable 5.1: Model A honors the 16:9 format, fogged win… ### Courier Leap Winner: **Boogu Image**. OpenAI: GPT-5.6 Sol Pro: Model A better matches the requested 16:9 cinematic realism, dusk setting, striped puddle, tram streaks, and accurately rendered storefront text. Model B is more dynamically staged and clearly includes a cargo bicycle, but its square format and glossy illustration-like aesthetic depart significantly from the prompt. (Second judge pass, order swapped — scores are the average of both: Model A more faithfully delivers the requested 16:9 cinematic realism, exact storefront text, rainy backstreet, ai | Anthropic: Claude Fable 5.1: Model A delivers the requested 16:9 cinema… ### Named art style Winner: **Boogu Image**. OpenAI: GPT-5.6 Sol Pro: Model B delivers the more dramatic and cohesive cresting-wave composition, with a clearly readable fishing boat, dawn atmosphere, restrained indigo-and-cream palette, and convincing stylized foam. Model A strongly reflects ukiyo-e linework and flat color planes but feels less dynamically composed, and its oversized pale wave reads somewhat like a mountain. (Second judge pass, order swapped — scores are the average of both: Model B more strongly captures the requested monumental cresting wave, vu | Anthropic: Claude Fable 5.1: Model A nails the ukiyo-e idiom with flat… ### Restricted palette Winner: **Boogu Image**. OpenAI: GPT-5.6 Sol Pro: Model A closely follows the required four-color palette, flat-vector style, campsite subject, and 16:9 format, though some shapes are slightly crude. Model B is visually appealing but uses multiple teal shades and a square aspect ratio, substantially weakening prompt adherence. (Second judge pass, order swapped — scores are the average of both: Model A better satisfies the 16:9 format and visibly uses all four requested colors in a balanced campsite scene, though subtle tonal variation weakens t | Anthropic: Claude Fable 5.1: Model A delivers a 16:9 flat-vector scene… ### Spatial layout Winner: **Krea 2 Large**. OpenAI: GPT-5.6 Sol Pro: Model A follows the requested layout very clearly and has a cleaner minimalist vector treatment, but its furniture forms are somewhat rudimentary. Model B also places every key element correctly while offering a more balanced, polished, and visually engaging composition, despite slightly heavier texture and detail. (Second judge pass, order swapped — scores are the average of both: Model B more precisely places the desk directly under the window while clearly including all requested furnishings, | Anthropic: Claude Fable 5.1: Model A places all four elements correctly… ### Spice Stall Macro Winner: **Boogu Image**. OpenAI: GPT-5.6 Sol Pro: Model A better matches the requested 16:9 macro framing, chipped spoon, sharply resolved spice texture, folded newsprint, and scooter-filled creamy bokeh, though the tag appears to separate the digits in “47-C.” Model B renders the tag more accurately and conveys a dense stall, but its square format, distant subject, smoother spice, and cluttered composition weaken prompt adherence and macro impact. (Second judge pass, order swapped — scores are the average of both: Model A better fulfills the 1 | Anthropic: Claude Fable 5.1: Model A delivers the requested 16:9 macro…
Negation
A cozy reading nook with an armchair, a stack of books, and a mug of tea by a window — with absolutely NO plants, NO lamps, and NO artwork or picture frames anywhere in the frame. Warm afternoon light, 16:9.
OpenAI: GPT-5.6 Sol Pro: Model B cleanly includes the requested chair, books, tea, window, and warm light without forbidden interior decor, though it misses the required 16:9 format. Model A has stronger widescreen composition but appears to include cropped picture-frame elements at the top and right edges, and its book-spine text is visibly garbled. (Second judge pass, order swapped — scores are the average of both: Model B is more polished and avoids lamps and artwork, though its square format and visible outdoor vege | Anthropic: Claude Fable 5.1: Model A delivers the 16:9 format and all r…
Quiet Fury
A hyper-detailed painted portrait of a night-shift subway kiosk attendant behind scratched safety glass, unmistakable quiet fury on her face after finding a handwritten eviction notice tucked beside a jar of lychees, jaw tight, nostrils slightly flared, one eyebrow lifted, eyes wet but unblinking, reflection of a vending machine glowing cyan across her cheek, cramped urban station interior with torn route maps and coin rolls in soft focus
OpenAI: GPT-5.6 Sol Pro: Model B more convincingly depicts quiet fury, wet unblinking eyes, scratched glass, cyan vending-machine reflection, the eviction notice, and a recognizable jar of lychees within a richly detailed kiosk. Model A is polished and well framed, but the expression reads more sorrowful, the fruit is less clearly identifiable, and the malformed cyan lettering on the cheek is distracting. (Second judge pass, order swapped — scores are the average of both: Model B more completely captures the unmistakabl | Anthropic: Claude Fable 5.1: Model B conveys genuine fury with wet unbl…
Neon Laundromat
An isometric 3D render of a narrow corner laundromat on Kester Lane at neon night, magenta and acid-green signage reflecting across wet sidewalk tiles, two unattended washing machines glowing through fogged windows, a vending machine buzzing beside a stack of plaid duffel bags, moody puddle reflections, deep midnight-blue sky, a lone orange traffic cone casting a long shadow under mixed neon spill, urban street scene, 16:9
OpenAI: GPT-5.6 Sol Pro: Model B delivers the stronger, more polished urban corner composition with convincing fogged windows, neon signage, vending machine, plaid bags, lone cone, and puddle reflections. Model A adheres better to the requested 16:9 format and exactly two washers, but its sparse diorama-like scene and extra cone feel less atmospheric and refined. (Second judge pass, order swapped — scores are the average of both: Model B offers a richer, more polished urban corner scene with strong fogged-glass atmosphe | Anthropic: Claude Fable 5.1: Model A matches the 16:9 ratio and two was…
Courier Leap
A rain-speckled urban backstreet at dusk, cinematic realism, a bike courier in a teal windbreaker vaulting off a wobbling cargo bicycle to snatch a tumbling stack of violet takeout boxes before they hit the pavement, one sneaker airborne over a striped puddle, coat tails and receipt slips whipping sideways, water droplets frozen in the air, tilted composition and motion streaks from passing tram lights, gritty storefront signs reading "Kite Repair" and "24h Key Lab," high dynamic energy, 16:9
OpenAI: GPT-5.6 Sol Pro: Model B captures the central airborne leap, whipping receipts, frozen rain, tilted energy, and tram-light motion far more dramatically, though it misses the requested 16:9 framing and renders the boxes somewhat unnaturally. Model A has cleaner realism, accurate storefront text, and correct framing, but the courier remains seated on the bicycle, substantially weakening the prompt’s defining action. (Second judge pass, order swapped — scores are the average of both: Model B better captures the req | Anthropic: Claude Fable 5.1: Model A delivers the correct 16:9 frame, c…
Named art style
A ukiyo-e woodblock print of a fishing boat riding a large cresting wave at dawn, faithful to the flat color planes, bold outlines, and stylized foam of the tradition, muted indigo and cream palette.
OpenAI: GPT-5.6 Sol Pro: Model A more fully realizes the requested dramatic cresting wave, fishing boat, dawn light, stylized foam, and muted indigo-and-cream ukiyo-e treatment in a balanced, dynamic composition. Model B is attractive and technically polished, but the oversized empty sky and cropped wave crest reduce visual impact and prompt specificity. (Second judge pass, order swapped — scores are the average of both: Model A presents a more dynamic, legible cresting wave, clearer dawn light, and stronger visual bala | Anthropic: Claude Fable 5.1: Model A delivers a clearly cresting wave w…
Restricted palette
A flat-vector illustration of a mountain campsite at dusk using ONLY four colors — cream, burnt orange, teal, and charcoal — with no gradients and no stray off-palette colors, balanced composition, 16:9.
OpenAI: GPT-5.6 Sol Pro: Model A adheres much more closely to the required 16:9 format, flat treatment, and four-color palette, though the campsite is somewhat undersized. Model B is more detailed and visually engaging, but its square format, prominent sky gradient, and multiple tonal variations directly violate key restrictions. (Second judge pass, order swapped — scores are the average of both: Model A adheres much more closely to the required four-color palette, flat treatment, and 16:9 format, though its campsite is | Anthropic: Claude Fable 5.1: Model A honors the strict four-color palet…
Spatial layout
A clean isometric illustration of a bedroom: a bed against the LEFT wall, a round rug centered on the floor, a desk under the WINDOW on the back wall, and a floor lamp in the FRONT-RIGHT corner. Flat-vector style, consistent perspective.
OpenAI: GPT-5.6 Sol Pro: Model B more clearly and cohesively places the bed along the left wall, the rug centrally, the desk beneath the back-wall window, and the lamp toward the front-right, while maintaining a polished flat-vector isometric style. Model A largely follows the layout but has a less convincing front-right lamp placement, weaker spatial consistency, and a more washed-out presentation. (Second judge pass, order swapped — scores are the average of both: Model B clearly places every requested element, with a | Anthropic: Claude Fable 5.1: Model A places the rug more centrally and…
Spice Stall Macro
Macro close-up food-stall scene in a dense street market, ultra-realistic photography style, a chipped ceramic spoon heaped with electric-orange ajika spice resting on folded newsprint beside a tiny paper tag marked "Lot 47-C," razor-thin depth of field from a 100mm macro lens, individual salt crystals and pepper seeds sharply resolved while the background dissolves into creamy bokeh of scooter taillights and hanging tarps, diagonal framing, 16:9
OpenAI: GPT-5.6 Sol Pro: Model A better matches the requested 16:9 diagonal framing, folded newsprint, clearly rendered “Lot 47-C” tag, dense market setting, and sharply resolved spice texture. Model B has convincing macro detail and a more visibly chipped spoon, but its square format, cropped utensil, less legible tag, and distracting foreground bag weaken adherence and composition. (Second judge pass, order swapped — scores are the average of both: Model A better satisfies the requested 16:9 diagonal framing, clearly | Anthropic: Claude Fable 5.1: Model B delivers a far more convincing macr…
Matchup powered by OpenRouter.