Permintaan

Here is a **cleaned, LM Arena–friendly version** of your prompt. I’ve removed engine-specific wording, tightened instructions, and structured it in a way **LM Arena models respond best to** (clear constraints, no repetition, strong realism + sync emphasis), while keeping all your intent: --- Create a **4K ultra-cinematic vertical video (9:16)** strictly based on the uploaded image. Preserve the exact character design, pose, facial structure, clothing, environment, and overall mood. Maintain realism, grounded emotion, and natural motion. The setting is a misty dusk landscape with shallow water reflecting soft moonlight, fog drifting naturally across the ground, a large pale moon diffused by haze in the background, and a lone red-leafed tree on a rocky cliff with leaves slowly falling in the wind. In the foreground, the samurai kneels on one knee on a dark rock, both hands resting on the hilt of his sword, the blade planted vertically into the stone. His posture conveys exhaustion and control, not exaggerated heroism. Clothing and hair move subtly with realistic wind physics. His exposed arm shows a faded blue tattoo and light battle wear. Breathing is slow, deep, and audible, with visible chest movement that must remain **perfectly synchronized with the audio**, including breath timing and pauses. Facial animation, lip movement, jaw motion, and micro-expressions must **match the spoken audio exactly**, with no delay or mismatch. When speaking, the emotional tone of the voice must align with his physical state: calm, worn, grounded, and resolute. The camera remains mostly still with only a very slow, natural push-in. Ambient sound begins with wind, distant water movement, and subtle metallic resonance. At the exact moment the character slightly lifts his head, synchronized precisely to the audio cue, he speaks in a low, controlled, breath-worn voice: **“The sword doesn’t make the samurai. The spirit does.”** The dialogue must be frame-accurate with lip-sync, breathing, and body language. Music, if present, remains minimal and enters only after the dialogue ends. The final frame holds on the warrior, unmoving and grounded. Render with realistic anime-style shading, natural lighting, restrained color grading, subtle film grain, smooth 24fps cinematic motion, and full 4K UHD detail, prioritizing realism and audio-visual synchronization. --- If

Hasil

Here is a **cleaned, LM Arena–friendly version** of your prompt. I’ve removed engine-specific wording, tightened inst...

Pengaturan

Gambar referensi