Head to head: Happy Horse 1.1 Image to Video vs Wan 2.5 Image to Video
Happy Horse 1.1 Image to Video vs Wan 2.5 Image to Video
By Ryan Merket · Published · Updated
A close comparison across kayak follow shots, velodrome tracking, dusk lighting changes, and ice-in-soda physics. The models trade strengths depending on the shot’s specific demands.
This is a genuine dead heat: Happy Horse 1.1 scores 27.9 to Wan 2.5’s 27.3, and the limited confidence estimate gives no basis for calling either model better overall. The small score gap matters less than how often each model succeeds on a different kind of shot. For the kayak sequences, Wan more often holds the requested rear-quarter follow through a narrow gated channel; Happy Horse can bring more spray and urgency, but its camera sometimes swings toward a frontal view. In the velodrome, Happy Horse more reliably keeps the helmeted rider in a steady, handlebar-height sideline track. Wan’s banked turns and shifting light can be more dramatic, but its viewpoint is less consistent. The lighting transitions are similarly split: Wan often makes the lamp-switching-on moment legible, while Happy Horse gets stronger marks when the full warm-to-cool dusk progression and locked-off composition matter most. Physics results also divide. Wan sometimes keeps the ice plausibly afloat or shows a coherent impact-and-bob sequence; in other clips, Happy Horse better conveys the splash and buoyant return while Wan’s cube sinks. Neither model owns the repeatable advantage that would settle this matchup. **Final call: tie. Happy Horse 1.1 and Wan 2.5 are effectively even here; choose by the shot you need, not by this aggregate score.**
Kayak pace follow
A short continuous 16:9 river-level shot of a slalom kayaker in a bright saffron helmet sprinting downstream between green practice gates on a narrow training channel, the camera in a fluid gliding follow just off the athlete’s rear-left quarter, subtly accelerating and arcing to keep the paddler centered and sharp as the kayak yaws, water sprays, and concrete embankments, reeds, and numbered gate poles stream past in the background; cool overcast morning light with silver reflections on the ripples, focused, disciplined, and urgent mood.
OpenAI: GPT-6 Luna: Model A captures the fast slalom action with splashing water and a fluid camera arc from the athlete’s rear-left quarter, while keeping the saffron helmet and green gates prominent. Model B is clear and consistent but stays in a frontal view, so it misses the requested trailing camera movement and urgent follow-shot feel. (Second judge pass, order swapped — scores are the average of both: Model B clearly shows the narrow gated channel and maintains a sharp, consistent subject, but its view is mo | Anthropic: Claude Fable 5.1: Model A delivers the requested gliding rear/sid…
Stadium shadow roll
A short continuous 16:9 sideline shot at a neighborhood velodrome: a teenage track cyclist in a matte teal skinsuit powers through one banked turn and onto the straight while the camera glides parallel at handlebar height, matching speed; halfway through the clip a long ragged cloud slides over the late-afternoon sun, and the scene smoothly shifts from warm golden light with crisp wheel-spoke shadows to cooler muted shade, then a sliver of sun returns and briefly relights the rider’s face and the painted lane lines, all staying natural and believable, focused and energetic with a tense pre-race mood.
OpenAI: GPT-6 Luna: Model A better matches the requested parallel sideline tracking shot, with a consistent rider, teal kit, helmet, and believable velodrome setting. Model B has more dramatic turn imagery, but its changing frontal angles, less natural look, and lack of a clear shadow-to-sunlight transition weaken its fit. (Second judge pass, order swapped — scores are the average of both: Model A better sustains the parallel, handlebar-height tracking shot with coherent rider and bicycle motion, while Model B more | Anthropic: Claude Fable 5.1: Model A delivers a convincing sideline tracking…
Lighting transition
A living room at dusk as the sun sets over a few seconds: the warm light visibly dims and shifts to cool blue while a lamp flickers on, the transition smooth and believable, locked-off frame, 16:9.
OpenAI: GPT-6 Luna: Model A shows the warm-to-cool dusk shift clearly, but its lamp is already on in the first frame. Model B more directly depicts the lamp switching on as the exterior darkens, with a convincing warm lamp glow against the cool evening light. (Second judge pass, order swapped — scores are the average of both: Model B clearly progresses from dusk into cooler blue darkness while the lamp visibly comes on and warms the room. Model A shows a sunset-to-blue shift, but its lamp appears lit from the start | Anthropic: Claude Fable 5.1: Model B shows a smooth, believable progression…
Physics realism
A close-up of a single ice cube dropped into a tall glass of soda: it falls, splashes, bobs, and settles with believable buoyancy and fizzing bubbles rising naturally, soft kitchen light, 16:9.
OpenAI: GPT-6 Luna: Model A shows a more obvious splash, but the ice ends submerged and the liquid disturbance looks exaggerated. Model B keeps the ice plausibly floating at the surface with natural-looking rising bubbles and steadier motion, though the drop itself is less clearly shown. (Second judge pass, order swapped — scores are the average of both: Model B keeps the cube plausibly buoyant and shows a coherent change in its tilt with fine rising bubbles, though the drop and splash are mostly absent. Model A de | Anthropic: Claude Fable 5.1: Model A has an attractive soft-lit kitchen and…
Matchup powered by OpenRouter.