Head to head: Kling Image vs Google: Nano Banana 2 (Gemini 3.1 Flash Image)
Kling Image vs Google: Nano Banana 2 (Gemini 3.1 Flash Image)
By Ryan Merket · Published
Kling Image and Google’s Nano Banana 2 face prompts that test text, counting, spatial relationships, and dense scene details. This comparison is limited to the saved task sample.
Google’s Nano Banana 2 (Gemini 3.1 Flash Image) takes the saved verdict, with an aggregate score of **66.2** to Kling Image’s **56.8**. It won six tasks to Kling’s one, with one tie—a decisive editorial result for this sample, not a statistical conclusion. Nano Banana 2’s advantage came from delivering more of the requested details and arrangements: it handled the inverted umbrella and locksmith scene, the Art Deco tram-stop brief, the tailor-window details, and the room’s furniture layout especially well. It also won the text and attribute-binding tasks, though the text comparison had a trade-off: Kling made the headline more prominent, while Nano Banana 2 rendered the punctuation more precisely. Kling’s one task win was exact counting: both models showed seven espresso cups, but Kling’s cups were more consistently defined. The night-market repair scene was a tie; Kling made several activities easier to read, while Nano Banana 2 brought stronger signage and market detail. This assessment applies only to the saved task sample; it is not a claim about general capability beyond these prompts.
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 A gives the headline clear prominence and centers all three legible lines on a deep navy, textured background, while Model B’s surrounding white mockup and large empty space make it feel less like the requested focused poster. Model B renders the time separator more precisely, but its headline is less dominant and the overall layout is weaker. (Second judge pass, order swapped — scores are the average of both: Model B renders all three lines crisply with the requested punctuation, centered | Anthropic: Claude Fable 5.1: Both render all three lines correctly, but Mode…
Attribute binding
A photorealistic tabletop scene: a matte green cube to the LEFT of a glossy red sphere, a small blue cylinder BEHIND both, and a single yellow rubber duck sitting ON TOP of the cube. Colors must stay on the correct objects. Soft studio light, 16:9.
OpenAI: GPT-6 Luna: Model A more closely matches the requested soft studio tabletop scene, with all four objects and their spatial relationships clearly represented. Model B is also polished, but its natural indoor setting departs from the studio-lighting direction and its blue cylinder appears quite dark. (Second judge pass, order swapped — scores are the average of both: Model B more convincingly matches the requested object colors and materials, with a clearly glossy red sphere and matte green cube; the duck and | Anthropic: Claude Fable 5.1: Model B satisfies every binding and spatial rel…
Night Market Repair Crew
A lively multi-subject urban street scene at the pop-up night market on Guilder Lane, where a four-person scooter repair crew works beside a steaming dumpling cart while a violin busker, a flower vendor with three towering buckets of marigolds, two children trading glow stickers, and a courier balancing six parcel tubes all interact in one coherent space; strings of mismatched lanterns overhead, wet cobblestones reflecting crimson and jade light, clear foreground-midground-background separation, dynamic but well-balanced 16:9 composition.
OpenAI: GPT-6 Luna: Model A brings nearly all the requested activities into a coherent, legible scene, including the steaming dumpling cart, scooter repair, violinist, marigolds, children, and a courier carrying parcel tubes. Model B has strong Guilder Lane signage and lively market detail, but its busier composition obscures some interactions and does not clearly show the parcel tubes or a four-person repair crew. (Second judge pass, order swapped — scores are the average of both: Model A more clearly includes the | Anthropic: Claude Fable 5.1: Model B nails far more of the prompt's specific…
Neon Umbrella Vector
A clean stylized vector illustration of a tired bicycle locksmith named Ivo closing his tiny curbside kiosk under a sudden violet rainstorm on Bramble Quay, with one mint-green umbrella flipped inside out, scattered brass keys glinting on the wet pavement, and a sleepy orange cat curled in the shop window; moody but gentle late-evening atmosphere, simplified geometric shapes, crisp outlines, limited palette of plum, teal, amber, and charcoal, subtle reflections, balanced 16:9 street-corner composition.
OpenAI: GPT-6 Luna: Model A has especially crisp geometric vector styling and includes the bicycle, but its umbrella does not read as clearly inside out. Model B better captures the locksmith kiosk, the tired shopkeeper closing up, and the unmistakably inverted mint umbrella, with the cat, scattered keys, and rainy quay also present. (Second judge pass, order swapped — scores are the average of both: Model B more clearly delivers the flipped inside-out mint umbrella, scattered glinting keys, locksmith kiosk, and sl | Anthropic: Claude Fable 5.1: Model B delivers a richer, more coherent scene…
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 clearly separated, nearly identical espresso cups in a straight row on white marble, with a clean overhead view and soft lighting. Model B also has seven cups in a row, but the espresso surfaces look less consistent and more visibly distorted. (Second judge pass, order swapped — scores are the average of both: Both images show exactly seven clearly separate espresso cups in a straight overhead row on white marble. Model A gives the cups slightly stronger definition an | Anthropic: Claude Fable 5.1: Both show seven espresso cups in a row, but Mod…
Art Deco Tram Stop
A faithful Art Deco poster-style image of the midnight tram stop at Larkspur Circle, featuring a streamlined silver tram numbered 27, two elegantly dressed commuters beneath a fan-shaped shelter, towering stepped facades with sunburst ornament, bold symmetry, metallic gold and deep sapphire accents, hard-edged shadows, and theatrical uplighting after rain, composed like a refined 1930s travel advertisement in 16:9.
OpenAI: GPT-6 Luna: Model B more fully realizes the refined 1930s travel-poster brief, with two commuters, a clearly numbered tram, a fan-shaped shelter, rain reflections, and crisp Art Deco typography. Model A has strong gold-and-sapphire styling and symmetry, but its bottom lettering is cropped and it shows three commuters rather than two. (Second judge pass, order swapped — scores are the average of both: Model B more completely delivers the requested refined 1930s travel-ad aesthetic, with two commuters beneath | Anthropic: Claude Fable 5.1: Model A delivers bold symmetry, a sunburst, a f…
Tailor Window Textures
A hyper-detailed close street view of the window display at Nettle & Twine, a narrow urban tailor shop at dawn, showing sharply rendered herringbone wool jackets, loose measuring tapes, hand-stitched buttonholes, condensation beads on the glass, chipped painted wood trim, brushed brass signage, and a half-finished cuff on a mannequin arm inside; natural cool morning light raking across every surface to emphasize crisp fabric weave and material detail, realistic composition.
OpenAI: GPT-6 Luna: Model B includes more of the requested tailoring details, especially the shop name, measuring tapes, chipped trim, and a mannequin arm with an unfinished cuff, while retaining convincing street-window realism. Model A has a stronger close-up of the herringbone fabric and condensation, but omits several specific details and crops the signage. (Second judge pass, order swapped — scores are the average of both: Model B includes the full tailor-shop window scene, with legible Nettle & Twine signage, | Anthropic: Claude Fable 5.1: Model B fulfills nearly every prompt element—co…
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-6 Luna: Model B more clearly places the bed against the left wall, a desk directly beneath the back-wall window, a centered round rug, and a floor lamp in the front-right corner. Model A includes the requested elements but the desk is less recognizable, while Model B also has richer, more polished flat-vector detail and a coherent composition. (Second judge pass, order swapped — scores are the average of both: Model B clearly places all requested furnishings and presents them in a polished, consistent f | Anthropic: Claude Fable 5.1: Model B places the bed cleanly against the left…
Matchup powered by OpenRouter.