Head to head: H3 Max VHS vs PixVerse C1 Image To Video
H3 Max VHS vs PixVerse C1 Image To Video
By Ryan Merket · Published · Updated
A four-scene comparison puts H3 Max VHS and PixVerse C1 Image To Video through moody kitchen cinematography, liquid physics, a downhill delivery chase, and a water-balloon burst. The judging turns on whether stronger action and detail outweigh steadier framing and closer prompt fit.
The results are close and uneven, not a clean sweep: PixVerse took 7 task wins to H3 Max VHS’s 4, with 1 tie. The aggregate scores were **26.3** for H3 Max VHS and **28.6** for PixVerse, but the saved verdict for this sample is a tie. The score gap alone doesn’t settle a matchup whose task-level results change direction across scenes. PixVerse’s clearest advantage was visual energy. It often made the soda splash and buoyant ice easier to read, and rendered the balloon’s water sheet and droplets with sharper light. Its downhill runs could feel more urgent, too. But those gains sometimes came with trade-offs: a less convincing liquid, a looser match to the requested continuous follow, or a brighter look than the brief called for. H3 Max VHS had its own repeatable strengths: steadier camera movement, better adherence to some requested moods and shot structures, and—in one soda trial—a more cola-like result. The copper-pan dolly-in and wasabi delivery each split across runs, while the fluid tests leaned more often toward PixVerse. That mix makes a single headline score a poor substitute for looking at what each scene asked the models to do. Across 3 successful repeated trials, the mean paired score difference (H3 Max VHS minus PixVerse) was -2.2666666666666657; the Student-t 95% confidence interval for that mean difference was -8.256135106665418 to 3.722801773332087. This interval describes the estimated mean difference, not a probability of winning or a claim about broader capability. The assessment applies only to the saved task sample. **Final call: Too close to call (not statistical). Keep the saved tie; PixVerse’s higher aggregate score and larger task-win count don’t erase H3’s scene-specific strengths.**
Copper Pan Dolly-In
A short continuous 16:9 video clip in a moody midnight ramen bar kitchen: chef Nari Okabe rapidly tosses blue crab shells and charred scallion oil in a hammered copper pan over a roaring gas flame, sparks flicking upward as steam curls through amber practical light and faint cyan neon spill; the camera executes a smooth slow dolly-in from waist-high medium distance to an intimate close view of the pan, perfectly straight and steady with no tilt or orbit, while her hands keep cooking in motion and the atmosphere feels precise, hushed, and intensely focused.
OpenAI: GPT-6 Luna: Model B better captures the active crab-shell tossing, visible flame and sparks, steam, and moody amber-cyan lighting, with a compelling move into the pan. Model A has a steadier, more direct dolly-in, but the cooking action and atmospheric details feel less pronounced. (Second judge pass, order swapped — scores are the average of both: Model B delivers a clearer, steady progression from the chef to an intimate view of crab shells in the copper pan, with strong amber-and-cyan kitchen lighting an | Anthropic: Claude Fable 5.1: Model A delivers a clean, straight dolly-in on…
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 B clearly shows the drop and splash, followed by the cube rising toward the surface amid fizzing bubbles; Model A has an unusual suspended wave and a less convincing progression. Model B’s splash is somewhat exaggerated, but its motion and settling are more physically coherent overall. (Second judge pass, order swapped — scores are the average of both: Model B more clearly shows the drop, splash, and buoyant rise into a settled position, with convincing soda bubbles and a strong close-up c | Anthropic: Claude Fable 5.1: Model B shows a fuller sequence—impact splash,…
Wasabi Downhill Delivery
A short continuous 16:9 video clip at dawn on a steep mountain service road above terraced wasabi fields: a compact electric delivery cart branded Kiso Ferments rockets downhill carrying sloshing crates of fresh shiso, yuzu, and ice-packed tuna for a pop-up breakfast kitchen, tires spitting mist as the driver leans through tight bends; the camera performs a handheld follow, sprinting just off the cart’s rear quarter in one unbroken shot with aggressive forward momentum, lively shake, and natural motion blur on the rushing guardrails, under cold silver morning light with a breathless, high-voltage mood.
OpenAI: GPT-6 Luna: Model B better conveys the requested high-voltage downhill run, with strong cornering, tire mist, motion blur, and clearly visible ice-packed fish and greens in the foreground crates. Model A has a clearer Kiso Ferments logo and a convincing mountain setting, but its follow feels calmer and less urgent. (Second judge pass, order swapped — scores are the average of both: Model B better captures the breathless downhill chase, with pronounced motion blur, tire mist, and a tight bend while keeping t | Anthropic: Claude Fable 5.1: Model A delivers a crisp, consistent compact ca…
Fluid & particle dynamics
Cinematic slow-motion of a water balloon bursting, the sheet of water expanding and droplets scattering through the air in convincing detail against a dark background, hard side light, 16:9.
OpenAI: GPT-6 Luna: Model A has a cinematic dark setup and a convincing water membrane, but the sampled sequence mostly holds on a rounded shell with limited scattering. Model B more clearly shows the water sheet breaking apart into many sharply lit droplets, with stronger burst motion and a more satisfying progression. (Second judge pass, order swapped — scores are the average of both: Model B shows a striking, detailed water sheet that breaks into scattering droplets against a dark, hard-lit background, with a cl | Anthropic: Claude Fable 5.1: Model A shows a coherent intact-balloon-to-burs…
Matchup powered by OpenRouter.