Kling 3.0 Video Tests 🎬 No es taaaan bueno como pensé 🥲 Tú qué opinas?? 👀 Estos días he estado poniendo a prueba Kling 3.0 y aquí van todos los resultados (los buenos y no tan buenos 😅) tal cual salen: sin cortes, sin edición, y con un solo prompt por clip Lo que más me ha sorprendido es la consistencia de la cara 😍 ha mejorado muchísimo frente a 2.6 (y, sinceramente, frente a casi cualquier generador de vídeo que haya probado) Además, con la opción Multi-shot puedes pasar de una imagen de referencia a una mini secuencia de hasta 5 escenas en un solo vídeo. No es perfecto: a veces te cuela alguna toma rara, pero aun así es un salto enorme comparado con tener que generar cada escena a mano 👀 Ah! y para generar los vídeos lo he hecho a través de la plataforma de @higgsfield.ai ✨ Este finde os grabo un mini tutorial para sacarle el máximo partido 💕 Siento mucho pero esta vez no habrá prompts... no se quedaron guardados 😓 Qué te parecen los vídeos que genera Kling?

Aria Cruz | Influencer AI

@soy_aria_cruz · Digital creator

INSTAGRAM · 2026-02-13Source

1.0Klikes

88comments

Remix This

Prompt

GLOBAL LOCK

Vertical 4:5 cinematic action clip set inside a dusty abandoned concrete building, warm tan-and-orange war-zone palette, broken daylight shafts through industrial windows, airborne dust, rubble-strewn floor. Main subject is a young adult woman with light-to-medium skin, dark hair tied back, tactical clothing in muted olive and brown, fingerless gloves, combat boots, and a modern rifle. Keep the identity locked: intense eyes, sweaty face, grim focus, practical combat styling, no glam fashion treatment. Camera grammar is aggressive and cinematic with alternating medium action framing, over-the-shoulder weapon detail, tight face close-up, top-down final angle, shallow depth of field, motion blur on recoil and falling brass. No dialogue. Audio intent is gunfire and action sound design only.

[00:00-00:02.20] Open on the tactical woman sliding or bracing low on the dusty floor inside the ruined room, leaning back on one hip while aiming her rifle up and across frame. Warm window light cuts through the background haze and dust hangs in the air. The camera sits at a medium-low angle, capturing the full body posture and the urgency of the firefight.

[00:02.20-00:03.60] Cut to an over-the-shoulder or near-weapon close-up aimed down the rifle toward an unseen target. The gun dominates the frame, background flames or hot light bloom softly at the edge, and the subject's gloved hands keep the weapon steady. The recoil feel should be sharp but readable.

[00:03.60-00:05.80] Push into an intense close-up of her face partially framed by the rifle. Her eyes are narrowed, jaw tense, and sweat-and-dust texture remains visible. Keep the muzzle and optic or barrel edge crossing the foreground while her expression sells concentration under pressure.

[00:05.80-00:07.40] Cut to a high overhead angle as she drops or rolls lower to the floor with the rifle across her body, legs bent and body turning in the dust. The choreography should feel like a continuation of the fight rather than a clean pose, with dust trails and friction visible on the ground.

[00:07.40-00:10.19] Finish on macro inserts of brass shell casings bouncing and spinning across the dusty floor near combat boots. The background stays blurred and warm, emphasizing the violence aftermath through tactile detail. End with the shell motion slowing naturally as the dust settles.

NEGATIVE PROMPT

no sci-fi armor, no polished superhero styling, no fantasy monsters, no extra soldiers entering frame, no pristine clean room, no glossy magazine makeup, no toy gun proportions, no broken rifle geometry, no warped hands, no duplicated fingers, no frozen shell casings, no bullet-time effect, no text overlays, no subtitles, no shaky phone footage, no nighttime blue grading, no blood gore splatter, no robotic face motion, no inconsistent hair length.

SHOT PROMPTS

Shot 1
Female tactical fighter sliding low on a dusty abandoned interior floor, rifle raised, warm daylight through ruined windows, cinematic combat framing.

Shot 2
Over-the-shoulder rifle close-up with gloved hands and shallow depth of field, hot light bloom in background, recoil-ready action detail.

Shot 3
Tight face close-up behind the rifle, intense focused eyes, sweat and dust texture, warm war-zone palette.

Shot 4
Top-down shot of the woman dropping or rolling with the rifle on a dusty floor, dynamic combat choreography.

Shot 5
Macro brass shell casings bouncing on dusty concrete near combat boots, shallow focus, tactile action aftermath.

SPEECH PACK

[00:00-00:10.19]
No spoken dialogue, no lip-sync, and no vocal turn-taking. Audio should be action-only: weapon fire, shell impacts, dust movement, and cinematic battlefield ambience without narration.

Why soy_aria_cruz's Tactical Woman Firefight Kling Video Went Viral — and the Formula Behind It

This short action clip is not a full narrative scene. It is a compact combat-style demo that tests whether an AI video model can preserve one character identity across multiple aggressive framings: wide body action, weapon POV detail, face close-up, overhead movement, and shell-casing macro inserts. That makes it a useful growth case for creators evaluating AI action consistency rather than just visual prettiness.

Case Snapshot

Format: vertical cinematic action test clip.

Main subject: one tactical female fighter in a ruined interior.

Scene logic: low sliding firefight pose, rifle detail, intense face close-up, overhead collapse, shell-casing aftermath.

Stylistic goal: prove combat realism and subject consistency inside a short AI-generated sequence.

Context clue: the original Instagram caption explicitly frames the post as a Kling 3.0 test reel with no edits.

What You're Seeing

The clip opens in a dusty abandoned building with warm sunlight pushing through windows and suspended particles catching the light. A dark-haired woman in tactical gear slides or braces across the floor while firing a rifle. The edit then tightens rapidly. Instead of staying in the wide shot, it jumps into weapon proximity, then an intense face close-up, then a high-angle collapse, and finally brass shell casings hitting the floor. That progression tells you the clip is built as a motion-and-consistency stress test, not as story cinema.

The most useful part is the identity lock. Even when the angle changes hard, the woman still reads as the same person: same hair, same eyes, same tactical palette, same room logic. That is exactly the kind of thing creators care about when testing whether a video model can maintain continuity through action cuts.

Shot Breakdown

Shot Type	Visible Action	Why It Matters
Wide action opener	The fighter braces low on the floor while aiming her rifle.	Establishes body mechanics, outfit readability, and environment fast.
Weapon detail	The camera cuts close to the rifle and hands.	Tests prop integrity and recoil credibility.
Face lock	Her focused eyes fill the frame behind the weapon.	Checks whether the model can preserve character identity in close-up.
Overhead movement	A top-down angle shows her lower to the ground with the rifle across the body.	Adds spatial variety and body continuity stress.
Aftermath insert	Shell casings bounce across the dusty floor.	Provides a tactile finish and tests object physics detail.

Why This Format Works

This format works because it compresses several hard AI-video problems into one short clip. Identity consistency, weapon geometry, combat body motion, environmental continuity, and small-object physics all appear inside ten seconds. That makes the video inherently interesting to creators who do not just want pretty frames; they want to know whether the model survives stress.

The clip also performs well as social content because every beat feels screenshot-worthy. The opening slide pose looks like a poster frame. The face-behind-rifle shot is emotionally intense. The shell-casing close-up feels premium and cinematic. When a short test video contains multiple usable stills, it becomes more shareable in creator communities.

Five Creative Hypotheses

The clip is intentionally short because action-model weaknesses show up quickly and do not need long runtime to be visible.
The subject was chosen to test whether a female fighter can remain visually consistent under high-motion framing.
The shell-casing ending is not random flair; it is a deliberate physics test for small fast metallic objects.
The warm dusty grade helps hide some generative softness while still feeling cinematic.
The lack of dialogue is strategic because it isolates visual-performance testing from speech-sync complexity.

How To Recreate It

Start with one locked fighter identity and one consistent ruined interior environment. Do not overbuild the scene with extra soldiers or heavy effects. The power of this format comes from focus. Use one subject, one rifle, one room, and a small run of deliberate angles. The best sequence order is: establish the body in action, move close to the weapon, move closer to the face, then widen the spatial logic again from above, and finally land on a tactile detail shot that implies combat aftermath.

For AI generation, the critical risk areas are hands, weapon integrity, and pose continuity. If the rifle warps between shots or the face drifts under pressure, the whole clip stops feeling like a reliable model test. Keep the lighting and wardrobe simple enough that continuity errors become obvious rather than hidden.

Growth Playbook

This is the kind of format that performs well with creator audiences because it invites evaluation. Instead of saying "look at this cool scene," it implicitly asks, "did the model hold up?" That creates comment energy around specific technical points: face consistency, gun realism, shell physics, and camera transitions. Posts that trigger analysis often outperform posts that only ask for admiration.

If you were building a series from this, each episode could test one difficulty domain: firefight interiors, chase scenes, water simulation, crowd continuity, creature reveals, or one-take dialogue. That would turn isolated outputs into a creator-education franchise instead of random demo drops.

FAQ

Is this a full story scene?

No. It reads more like a targeted AI action-model test than a complete narrative sequence.

What is the hardest part of recreating this well?

Keeping the same face, hands, and rifle believable across multiple fast action angles.

Why end on shell casings?

Because small metallic objects bouncing on the floor are a good stress test for timing and physics realism.

Does this require dialogue?

No. The clip works as pure action design and does not depend on spoken lines.