Head to head: Wan vs Google: Nano Banana Pro (Gemini 3 Pro Image)
Wan vs Google: Nano Banana Pro (Gemini 3 Pro Image)
By Ryan Merket · Published
A practical image-generation matchup testing typography, perspective, palette control, and faithful execution of detailed scene briefs. The results turn on a familiar trade-off: polished atmosphere versus precise composition and scale.
Google’s Nano Banana Pro takes the matchup, but this is a lean rather than a runaway: it scores **64.5 to Wan’s 60.1**, wins **five tasks to three**, and carries **limited confidence**. Its edge comes from breadth across distinct visual styles, not from sweeping every kind of prompt. Google stands out on the Art Deco bathhouse, where vivid mosaics and a sunburst grille make the requested poster aesthetic feel deliberate; on the chrome-and-water reflection scene, where the window and red apple read more clearly; and on attribute-heavy or palette-led briefs, including the atrium, rainy teahouse, and four-color campsite. Across these, it more often turns the prompt’s ingredients into a finished, atmospheric image. Wan’s wins expose where Google’s polish can come at the expense of the brief. Wan makes the three-line navy poster larger and easier to read, builds a cleaner one-point-perspective book aisle, and—most importantly—renders the market hall’s whale as the requested skeleton at a believable scale. Google’s market-hall whale is not a skeleton and overwhelms the scene, a concrete miss rather than a stylistic disagreement. **Final call: Google: Nano Banana Pro (Gemini 3 Pro Image) wins on overall range and finish, but Wan is the sharper pick when legibility, disciplined perspective, or literal control of a scene’s scale matters most.**
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 presents the requested deep-navy poster directly, with all three lines crisp, correctly spelled, centered, and clearly prioritized. Model B also renders the wording accurately, but the text block is smaller and the poster is surrounded by substantial mockup whitespace. (Second judge pass, order swapped — scores are the average of both: Model A presents the three requested lines larger and more legibly, with a clean centered layout on a deep navy background. Model B is also accurate and p | Anthropic: Claude Fable 5.1: Both render all three lines correctly spelled a…
art_style_fidelity_bathhouse
An evening street-facing view of the fictional Sun Motif Bathhouse rendered faithfully in Art Deco style, 16:9: stepped facade, sunburst grillework, zigzag mosaics, fluted pilasters, geometric stained glass, sleek chrome signage, and stylized figures entering beneath a marquee; dramatic uplighting, polished symmetry, luxurious 1930s poster-like composition, and ornament that clearly embodies Art Deco rather than generic vintage design.
OpenAI: GPT-6 Luna: Model B more fully realizes the prompt’s luxurious 1930s Art Deco poster aesthetic, with vivid zigzag mosaics, a prominent sunburst grille, geometric glass, and polished neon signage. Model A is symmetrical and on-theme, but its simpler, more photographic presentation and surrounding buildings make the bathhouse feel less like the focused poster-like centerpiece requested. (Second judge pass, order swapped — scores are the average of both: Model B more fully realizes the evening street-facing br | Anthropic: Claude Fable 5.1: Model B delivers a vivid, poster-like Art Deco…
Reflections & glass
A photorealistic close-up of a chrome teapot on a polished dark marble counter next to a glass of water; the window and a red apple on the counter must be correctly reflected in BOTH the chrome and the water's surface, dramatic side light, 16:9.
OpenAI: GPT-6 Luna: Model B has stronger close-up composition and more dramatic, convincing reflections, including the window and a red apple visible in the chrome and near the water. Model A is clean and photorealistic, but the apple reflection in the water is less apparent and the lighting feels less dramatic. (Second judge pass, order swapped — scores are the average of both: Model B more clearly delivers the requested close-up and shows the window and red apple reflected across the chrome and water, with dramat | Anthropic: Claude Fable 5.1: Model B delivers more dramatic side light and s…
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 presents a particularly clear, centered one-point perspective, with the shelves and overhead lights converging cleanly and the books diminishing consistently into the distance. Model B is also visually appealing and well aligned with the prompt, but its deeper, busier composition makes the vanishing point slightly less crisp. (Second judge pass, order swapped — scores are the average of both: Model A more precisely delivers the long, symmetrical one-point-perspective aisle, with shelves | Anthropic: Claude Fable 5.1: Model A delivers an exceptionally clean one-poin…
attribute_binding_atrium
A crisp editorial interior rendering of the lobby atrium of the fictional Veridian Ledger Archive, 16:9, viewed from a slightly elevated front-left angle in clean morning skylight: a brushed copper reception desk, a matte cobalt-blue concrete staircase, four translucent amber resin pendant lamps, a forest-green velvet banquette, a blackened steel directory totem, a pale pink terrazzo floor with white aggregate, and a frosted mint-glass revolving door; every object must keep its own exact color and material with no color swapping, bleeding, or mixed surfaces, in a refined contemporary architectural visualization style.
OpenAI: GPT-6 Luna: Model B includes all the specified elements, notably four amber pendants, and presents them in a polished, spacious composition with strong morning skylight. Model A is also cohesive, but shows only three pendants; both images have imperfect directory text. (Second judge pass, order swapped — scores are the average of both: Model B includes all specified elements, notably four amber pendants, and presents them in a polished, cohesive atrium with clear material and color separation. Model A is al | Anthropic: Claude Fable 5.1: Model B satisfies more of the prompt's specific…
palette_accuracy_teahouse
A serene courtyard teahouse interior in a restrained gouache-style architectural illustration, 16:9, composed symmetrically at dusk with paper lantern glow and rain beyond the shoji screens; restrict the entire image strictly to the named palette Dust Orchard: ochre, terracotta, sage green, muted plum, warm ivory, and charcoal gray, with no other colors anywhere—not in reflections, skin, foliage, ceramics, sky, or shadows.
OpenAI: GPT-6 Luna: Model A has exceptionally clear symmetry and a tightly controlled palette, but Model B better captures the restrained gouache atmosphere, dusk, lantern glow, and rainy exterior. Model B’s people and broader tonal range make its strict palette adherence slightly less certain. (Second judge pass, order swapped — scores are the average of both: Model B more fully delivers the dusk teahouse scene, with glowing paper lanterns, rain visible beyond the shoji, and a restrained palette close to the named | Anthropic: Claude Fable 5.1: Model B delivers the gouache texture, visible r…
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-6 Luna: Model B presents a clear, balanced dusk campsite with a strong mountain silhouette and a cohesive four-color flat-vector palette. Model A includes the requested elements and palette, but its crowded foreground and pale sky make the scene feel less balanced and less distinctly dusk. (Second judge pass, order swapped — scores are the average of both: Model B more clearly conveys a mountain campsite at dusk with a balanced, polished layout and the requested four-color flat-vector treatment. Model A | Anthropic: Claude Fable 5.1: Model A keeps a fairly strict four-color palett…
perspective_scale_market_hall
A wide-angle architectural visualization of the interior of North Quay Market Hall, 16:9, seen from eye level down the central aisle with strong one-point perspective toward a distant clock tower window: towering cast-iron columns, evenly spaced trusses, two levels of balconies, tiny shoppers, medium-sized produce stalls, and one monumental suspended whale skeleton installation crossing the nave; maintain believable relative scale between people, stalls, columns, balconies, and the skeleton, with straight structural lines, consistent vanishing point geometry, and no warped or mis-sized elements.
OpenAI: GPT-6 Luna: Model A closely follows the requested centered perspective and includes the two-level balconies, orderly structure, produce stalls, clock window, and a clearly skeletal whale installation at believable scale. Model B has an atmospheric market-hall composition, but its enormous suspended whale is not a skeleton and overwhelms the scene’s intended scale. (Second judge pass, order swapped — scores are the average of both: Model A includes the requested whale skeleton and presents a clear, symmetric | Anthropic: Claude Fable 5.1: Model A follows the brief more literally (eye-l…
Matchup powered by OpenRouter.