Head to head: Bernini-R Edit Video vs Happy Horse 1.1 Image to Video

Happy Horse 1.1 Image to Video takes this one on the numbers and on the tape. It wins 3 of 4 tasks and posts the stronger aggregate score, 34.2 to 28.8, with an 81% confidence verdict. That is not a blowout, but it is a real edge: a lean win backed by repeated prompt adherence, not a coin flip dressed up as certainty.

Where Happy Horse separates itself is consistency on prompts that require multiple things to happen at once. In crowd motion, it delivered the clearer overhead Tokyo-style scramble crossing, with pedestrians moving independently in multiple directions instead of drifting into a more uniform flow. In marsh hide sun-swell, it was again more exact: the bird hide actually read as a hide, the 27 marking was clearer, the low tracking angle was stronger, and the warm-to-cool-to-warm lighting beat came through as an intentional temporal change rather than a vague mood shift.

The same pattern held in subject action. Happy Horse gave the cleaner overhead close-up and the more believable hand-and-pitcher progression, producing a rosetta that actually resolves as a rosetta. Bernini-R looked pleasant enough, but this benchmark was about precise action rendering, and it drifted into less specific latte-art shapes while missing the requested framing.

Bernini-R’s case rests on fox under stormglass, and it is a legitimate point in its favor. That was the strongest individual showing from either model: better black-tipped fox coloration, a more convincing follow-to-side camera move, and a sharper escalation into storm-filtered violet light near the granite outcrop. It proved Bernini-R can be the more cinematic and atmospherically exact model when the scene leans on tension, weather progression, and grounded pursuit.

Final call: Happy Horse 1.1 Image to Video wins. Bernini-R is not outclassed—it produced the best single clip of the matchup—but Happy Horse was the more reliable prompt follower across the set, and that broader consistency is why it gets the verdict.

How they were tested

We ran 4 fresh video tasks, generated on the fly for this matchup so neither model could prepare in advance, and had gpt-5.4 score each one. To cancel position bias, every task was judged twice — once in each presentation order — and every number reported here, including the headline totals, is the average of both passes. Bernini-R Edit Video scored 28.8 to Happy Horse 1.1 Image to Video's 34.2.

1. Crowd motion

A busy Tokyo scramble crossing seen from above, dozens of pedestrians crossing in different directions, each moving independently without merging or warping into one another, overcast daylight, 16:9.

Winner: Happy Horse 1.1 Image to Video — Model B better matches the prompt with a clearer overhead Tokyo-style scramble crossing and pedestrians moving independently in multiple directions while remaining well separated and stable over time. Model A is also strong and realistic, but it feels less like a dense scramble crossing and shows slightly less convincing independent crowd flow and overall polish. (Second judge pass, order swapped — scores are the average of both: Model B better matches the prompt with a clear overhead view of a Tokyo-style scramble crossing and pedestrians moving in multiple directions with mostly independent, coherent trajectories. Model A is visually solid, but the crossing pattern is less scramble-like and the crowd motion appears more uniformly directed, reducing prompt adherence and overall dynamism.)

2. Marsh hide sun-swell

A single continuous 16:9 shot from a slow knee-high dolly move along the edge of a reed-choked salt marsh at low tide, tracking a cinnamon-colored marsh harrier as it glides just above the water and then banks past a weathered bird hide marked with a faded teal number 27; midway through the clip a thick fast cloud slides over the late-afternoon sun so the whole marsh dims from warm gold to cool muted gray, the harrier’s shadow dissolves, the silver ripples flatten, then sunlight gradually returns with believable softness and direction as the bird lifts over the reeds, naturalistic and smooth with no cuts.

Winner: Happy Horse 1.1 Image to Video — Model B matches the prompt more closely with a convincing marsh harrier glide past a bird hide clearly marked 27, stronger low-angle tracking, and a believable warm-to-cool-to-warm lighting progression. Model A captures a marsh setting and bird motion, but the hide reads more like a small boat or object rather than a weathered bird hide, the number marking is less clear, and the temporal lighting change is less evident and coherent. (Second judge pass, order swapped — scores are the average of both: Model B matches the prompt more closely with a low, continuous marsh-edge tracking shot, a visible bird hide marked 27, and a convincing warm-to-cool-to-warm lighting transition as the harrier glides and lifts over reeds. Model A has decent continuity and a readable marsh setting, but the hide looks more like a small boat or marker, the camera feels too elevated and distant, and the specific sun-dimming beat is much less evident.)

3. Subject action

A barista's hands pouring latte art: the milk stream forms a clean rosetta in the crema with natural, fluid wrist motion, no cuts, overhead close-up, soft café light, 16:9.

Winner: Happy Horse 1.1 Image to Video — Model B matches the prompt more closely with a clear overhead close-up, natural hand positioning, and a clean rosetta that develops coherently across frames. Model A has pleasing café lighting and decent realism, but the pattern reads more like concentric tulip/heart forms than a clean rosetta, and the motion progression is less convincing. (Second judge pass, order swapped — scores are the average of both: Model B matches the prompt closely with an overhead close-up, soft café lighting, and a clean rosetta formed through believable hand and pitcher progression. Model A has decent visual quality, but it misses the requested overhead framing and the latte art evolves into a less clear rosetta with weaker adherence to the specified action.)

4. Fox under stormglass

A single continuous 16:9 shot beginning with a gentle handheld follow behind a lean black-tipped red fox trotting along a narrow path through heather on a coastal headland, the camera slowly arcs to its side as wind combs the grass and distant gulls wheel; the light evolves from pale predawn blue to a bruised storm-filtered violet as a rain curtain advances offshore, and the fox shifts from cautious sniffing to a sudden alert sprint toward a lichen-covered granite outcrop, building a tense, watchful mood through motion, pacing, and darkening atmosphere in one unbroken take.

Winner: Bernini-R Edit Video — Model A matches the prompt more closely with a black-tipped red fox on a coastal path, a convincing camera move from behind to the side, and a strong progression into stormy violet light near a lichen-covered granite outcrop. Model B is attractive and coherent, but it looks more painterly, misses the specified black-tipped coloration, and the storm escalation and tense sprint feel less clearly realized. (Second judge pass, order swapped — scores are the average of both: Model A matches the prompt more completely, with a convincing continuous handheld follow on a coastal headland, clear progression to storm-filtered violet light, and the fox reaching a lichen-covered granite outcrop under an advancing rain curtain. Model B is attractive and coherent, but it feels more stylized and misses some of the specified atmospheric evolution and tense sprinting behavior compared with Model A.)

See every prompt and the full side-by-side outputs in the interactive Head-to-Head.