Head to head: Wan vs Google: Nano Banana 2 (Gemini 3.1 Flash Image)
Wan vs Google: Nano Banana 2 (Gemini 3.1 Flash Image)
By Ryan Merket · Published
A broad image-model matchup spanning exact object placement, perspective, anatomy, named styles, and action-heavy scenes. The prompts test not just visual polish, but whether each model makes the requested details immediately legible.
Google: Nano Banana 2 (Gemini 3.1 Flash Image) leads the saved task sample with an aggregate score of **63.4**, against **58.8** for Wan. It won five tasks to Wan’s two, with one tie; the saved editorial strength is a clear result, not a statistical claim. Google’s edge came in scenes where the central action needed to read at a glance: the airborne beach-handball shot, the rain-soaked fox courier’s skid, and the clay regatta crew working at an oarlock. It also made the ukiyo-e fishing scene and bracelet-tying moment more legible and cohesive. Wan’s wins were more exacting layout tests. It better matched the requested isometric bedroom arrangement, and its shuttlecock-drill image came closer to the specified count—though it still appeared to show twelve rather than eleven. The library-aisle perspective was a tie: Wan’s cleaner vanishing point met the brief, while Google’s image had more atmosphere but a less direct composition. **Verdict: Google: Nano Banana 2 (Gemini 3.1 Flash Image) is the saved winner, with a clear edge in this sample’s action-led and style-driven prompts; Wan is the stronger pick here for select spatial constraints. This assessment applies only to the saved task sample.**
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 A closely matches the requested isometric flat-vector style and places the bed, centered round rug, desk beneath the back-wall window, and lamp in the front-right corner. Model B follows the main furniture layout and is polished, but its perspective and detailed rendered style depart from the prompt. (Second judge pass, order swapped — scores are the average of both: Model B places the bed, centered rug, desk beneath the back-wall window, and lamp in the front-right area as requested, with | Anthropic: Claude Fable 5.1: Model B correctly places every element as speci…
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-6 Luna: Model B more clearly depicts fishing boats with people riding the cresting wave, while adding a dawn sun and a convincing ukiyo-e setting. Model A has a strong wave and restrained palette, but the boats are less legible as the central subject; Model B’s decorative lettering is a minor distraction. (Second judge pass, order swapped — scores are the average of both: Model B more clearly shows a fishing boat and crew riding the cresting wave, with a strong ukiyo-e palette, bold outlines, and styliz | Anthropic: Claude Fable 5.1: Model A nails the muted indigo-and-cream palett…
Golden-hour sand sprint
A women’s beach handball team sprinting through wind-ridged black volcanic sand during golden hour, one athlete midair for a spinning shot while two defenders kick up glittering sand, long amber shadows, salt haze over a distant breakwater, cinematic sports photography with crisp detail and warm glowing backlight, balanced wide composition, 16:9
OpenAI: GPT-6 Luna: Model B more clearly captures the airborne handball shot and the two defenders sprinting through kicked-up sand, with a stronger cinematic wide composition and vivid golden-hour light. Model A includes the volcanic ridges and breakwater but the action reads less decisively as a shot. (Second judge pass, order swapped — scores are the average of both: Model B more fully captures the cinematic beach-handball action, with a clear airborne shot, defenders spraying dark sand, and vivid golden backlig | Anthropic: Claude Fable 5.1: Model B reads as genuine beach handball (correc…
Vector velo-fox poster
A clean vector illustration poster of a fox bicycle courier skidding around a rain-slick corner with a package rack full of wildflowers, mood of mischievous determination, limited palette of teal, coral, cream, and charcoal, sharp geometric shapes, bold negative space, simplified city blocks and motion streaks in the background
OpenAI: GPT-6 Luna: Model B more clearly captures the fox courier skidding through rain, with forceful motion, a flower-filled rack, and a dramatic geometric city backdrop. Model A has a polished poster layout and strong palette, but its character and skidding action read less clearly, and the centered design leaves substantial unused space around the poster. (Second judge pass, order swapped — scores are the average of both: Model B strongly captures the fox courier skidding through a rain-slick city with wildflow | Anthropic: Claude Fable 5.1: Model B nails the core action and mood — a dete…
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 more cleanly fulfills the centered one-point-perspective brief, with shelves and consistently shrinking books converging to a clear vanishing point under warm overhead lights. Model B is atmospheric and convincing, but its extra architectural depth and carpet make the aisle feel less simple and direct. (Second judge pass, order swapped — scores are the average of both: Model A presents a cleaner, more consistent one-point perspective, with the shelves and books receding evenly toward a c | Anthropic: Claude Fable 5.1: Model A delivers a very clean, symmetric one-po…
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 has especially clear, natural-looking hands gripping the threads, while Model B more convincingly captures the candid act of tying a friendship bracelet onto someone’s wrist, with appealing depth and context. Model B’s fingers are slightly less crisp anatomically, but the overall scene is more engaging and still plausible. (Second judge pass, order swapped — scores are the average of both: Model B more clearly conveys a candid moment of tying a friendship bracelet onto someone’s wrist, w | Anthropic: Claude Fable 5.1: Model B delivers a convincing candid lifestyle…
Clay regatta pit crew
A stop-motion claymation scene of a quirky river regatta pit stop: three teenage rowers hauling a narrow racing shell onto a dock while a mechanic in yellow waders tightens an oarlock, a small terrier steals a striped sock, and a grandmother timekeeper bangs a brass bell, all arranged in a coherent triangular composition with wet wooden planks, bobbing lane buoys, and overcast morning light, 16:9
OpenAI: GPT-6 Luna: Model B more clearly shows the rowers hauling the shell while the mechanic works at an oarlock, and its dog, bell-ringing timekeeper, wet dock, and claymation style fit the prompt cohesively. Model A includes more visible regatta details and lane buoys, but the mechanic appears to be working on a bell rather than an oarlock, and the scene feels more cluttered. (Second judge pass, order swapped — scores are the average of both: Model B more clearly shows the mechanic tightening an oarlock while t | Anthropic: Claude Fable 5.1: Model B delivers a convincing Aardman-style cla…
Eleven shuttlecocks drill
An indoor badminton training hall viewed from slightly above, with exactly eleven white shuttlecocks scattered across a varnished court, each fully visible and individually countable, plus one junior player lunging to retrieve the nearest shuttlecock, realistic sports photography, even gym lighting, uncluttered background, 16:9
OpenAI: GPT-6 Luna: Model A has a clear junior-player lunge, a clean indoor court, and shuttlecocks that are mostly easy to distinguish, though it appears to show twelve rather than exactly eleven. Model B has a dynamic sports-photo composition but shows substantially more than eleven shuttlecocks, with several clustered together and difficult to count individually. (Second judge pass, order swapped — scores are the average of both: Model A is closer to the requested eleven individually countable shuttlecocks, thou | Anthropic: Claude Fable 5.1: Model A delivers an uncluttered hall viewed fro…
Matchup powered by OpenRouter.