AI 创作者用 Kling 4.0 实测完整能量饮料广告制作流程

Through a complete energy drink ad, leedain_leader_in explores how Kling 4.0 handles product details, character consistency, and natural dialogue, shifting more of the creative process from editing to directing.

From product close-ups to action sequences and behind-the-scenes dialogue, @leedain_leader_in puts Kling 4.0 to the test with a fictional energy drink ad. She explores how consistent details, natural performances, and greater creative control can shape a more streamlined AI advertising workflow.

Notes from the Creator

Hi, I’m AI creator @leedain_leader_in.

I share AI video-making techniques with 216,000 followers on social media, using practical tutorials to walk viewers through the creative process. My videos have accumulated more than 65 million views. As the head of Leader In Company, an AI education and content business, I run AI training programs for companies and institutions, as well as cohort-based AI video masterclasses. I’m also the author of the bestselling book AI Strategy Course.

My signature format is the “ad tutorial”: I first create work that meets the standard of a real advertisement, then share how I made it so others can learn from the process. When evaluating a model, I focus on two questions: Can it produce content good enough for advertising? And can it reproduce that quality consistently? This reflects my creative philosophy: “Humans above AI.” AI is a tool; the creative vision and direction still come from people.

Alongside my tutorials, I continue to create product commercials. In these videos, the product takes center stage. I build scenes around it that feel like real brand campaigns, using AI to bring to life worlds that would be difficult to film in reality. The video below is a commercial for a jelly product. It builds a complete world around the product, with hills made of rainbow-colored jelly, a vintage convertible floating across a jelly sea, and a mint-colored sky dotted with suspended planets. The video alternates between fashion-inspired character scenes and macro close-ups that reveal the translucent texture of gummy bears, bringing the brand’s atmosphere and the product’s details together in one piece.

In this kind of work, I always focus on two things: whether the product’s texture and shape remain consistent throughout, and whether the brand’s intended mood carries through from scene to scene. These were also the two criteria I used to evaluate Kling 4.0.

For this test, I used only Kling 4.0 to create a fictional energy drink ad, “KLING 4.0 ENERGY DRINK,” as a complete Reels video from beginning to end.

The Reel opens with leaked behind-the-scenes footage from an ad shoot. After I say, “You can do it easily with Kling,” it transitions into the finished drink commercial. I appear in both sections using the same facial reference image. In the first half, subtitles are added to the behind-the-scenes footage to make it feel like a real ad set. The second half moves into the drink commercial itself, ending with the words “Created with KlingAI 4.0.”

Full video (43 seconds)

All the videos shown in the following test sections are the original clips generated for this Reel.

When creating ad tutorials, I most often run into three challenges:

The product tends to deform as soon as it moves. In product close-ups—opening a can, drinking from it, or ice falling into the drink—the can’s pull tab can warp, and the logo may become distorted, as if melting.
The character changes when the shot changes. The same person or product can look subtly different from one scene to the next, and fixing those inconsistencies in post-production takes the most time.
The dialogue sounds like it’s being read from a script. Lines without emotion or natural emphasis can instantly undermine the realism of behind-the-scenes footage.

More expressive product close-ups: natural action and smooth camera movement

In a drink commercial, the most expensive shots to produce are the product-focused “sizzle shots.”

A can emerging from ice, the instant its pull tab opens, or an ice cube dropping into a fizzy drink—these moments communicate the product’s chill and energy. Filming them for real would require a high-speed camera and specialized rigging. They are also among the hardest shots to generate with AI: the pull tab may deform, or the liquid may look as if it is melting. For this test, I selected four product-only shots from an 18-second drink commercial to see how the model handled these challenges.

Kling 4.0 · Shots 1, 2, 8 and 9 from an 18-second multi-shot video · 1080p · 9:16 · Product reference image used.

Drink commercial: 0.0–2.0 sec and 12.3–15.3 sec

The can-opening mechanism is understood correctly. The pull tab lifts around the rivet as its pivot, and the drinking opening forms along the original cut line at the correct size, faithfully recreating how a real can opens. The finger doesn’t cover the opening, and the tab doesn’t incorrectly rotate sideways or detach. This is the first time I’ve seen a model follow mechanical instructions with such precision.
The ice and carbonation behave convincingly. As the ice cube breaks through the water’s surface, the water parts around it, while the air pulled in by the cube forms bubbles that rise to the top. You can even see bubbles inside the ice and the way light refracts through it.
The condensation actually runs. The droplets on the can aren’t just a static texture: one really does travel down the label. That’s exactly the kind of detail you want in a cold-drink commercial.
Even the smallest text is clear and readable. The KLING 4.0 logo is legible, as are “ENERGY DRINK” and the small slogan at the bottom, “OPEN YOUR IMAGINATION”—with no spelling errors. Small packaging text is one of the most common weak points in AI video, but here it remains crisp even in the original 1080p footage, alongside a convincing, filmed-in-camera metallic finish. The highlights on each droplet and the fine texture of the aluminum are also rendered beautifully. The ice, bubbles, can-opening action, and condensation all look accurate enough to work in commercial advertising footage.
I was even more impressed by how this changed the workflow. Before, I had to generate separate clips of ice, carbonation, and the product multiple times, then pick usable one-second segments and stitch them together. I also had to composite the text onto the product label in post-production. This time, the physical effects and packaging text were handled in a single generation. I didn’t even need to do the compositing afterward.

Multiple references come together smoothly: consistent characters from drinking to jumping over steps

To make a complete commercial, the product shots and character shots need to feel as if they belong to the same world. In this sequence, the character drinks from a can beneath a blue sky, sets it on a table, then immediately takes off running through a city plaza and leaps over a low set of steps. The 18-second video includes 10 hard-cut shots and uses 10 reference images—covering the product, locations, scenes, liquid, and face—to put the model’s ability to integrate multiple references and maintain character consistency to a rigorous test.

Kling 4.0 · Shots 3–7 and 10 from an 18-second multi-shot video · 1080p · 9:16 · 10 reference images

Drink commercial: 2.0–12.3 sec and 15.3–18.0 sec

Even with 10 reference images integrated at once, my face, outfit, and the can remain consistent across shots. The running sequence also reaches the level of a real sports commercial.

The 10 references come together in one world. The model receives reference images for the product, location, and character’s face, yet it selects the relevant image for each shot. Rather than simply copying the poses in the reference photos, it follows the actions and camera movements specified for each shot.
The face and outfit remain consistent. From a close-up of the character drinking to a full-body running shot, the face, bangs, ponytail, white-and-lime jacket, shorts, and running shoes all stay the same. The face also looks like the same person in the indoor studio footage from Test Scene 3.
The product stays consistent. The can changes position and angle as it appears in the ice, in the character’s hand, at their lips, and on the table, but the logo, lime-green diagonal stripe, silver finish, and overall proportions remain unchanged. Even when the can is tilted to drink, the logo follows the perspective accurately while remaining readable.
The drinking action feels natural. The angle at which the lips meet the rim matches the tilt of the wrist, and the cheeks don’t look strained.
The running feels like it belongs in a real sports commercial. The arms and legs alternate naturally, while the ponytail and jacket move with believable momentum. The character clearly moves forward, and the ground and background respond accordingly. Even when stepping onto the ledge, the legs and shoes don’t deform. The action flows without a pause. Pedestrians in the background also walk naturally, bringing the scene to life.
Three versions of the same person can move independently in one shot. In SHOT 7, three identical versions of Leader In run in separate lanes on the lime-green track. They share the same face and outfit, but each has independent limb movements, spacing, and shadows. When the same character is duplicated multiple times in a single frame, it’s common to see faces blend together or movements overlap completely. Here, the model generates all three characters as directed.

More realistic behind-the-scenes footage: natural performances, dialogue, and camera transitions

My Reel doesn’t begin with the finished commercial. It opens with “leaked behind-the-scenes footage” from an ad shoot. To make viewers believe that they’re seeing a real production set—and make the reveal that follows more effective—the footage needs to feel as realistic as possible. The director and crew are worried about how difficult the shoot will be. I’m having my makeup touched up when I answer, “You can do it easily with Kling.” This was the most challenging test: a 23-second video with more than five people, dialogue from three characters, and carefully directed focus shifts and close-ups of skin.

Kling 4.0 · 23 seconds · 1080p · 9:16 · With audio · 6 reference images (5 composition references and 1 face reference)

Overall, the emotion, emphasis, and pronunciation in the dialogue feel so natural that no corrections are needed.

The lines sound like real performances—that’s what I liked most. The director’s relaxed, casual explanation, the crew member’s genuine concern, and my bright, confident response each have their own natural emotion and emphasis. The Korean pronunciation also sounds natural. There’s no need to correct the audio or record the dialogue again; it’s ready to use.
The gestures match the dialogue. When the director says, “This is the drink,” their finger rests on the can’s photo. When they say, “This is a photo of Ms. Leader In,” they point to my picture. The photos on the tablet also appear in exactly the specified order: the can, the person, and the running scene.
The shot composition follows the directions. From the wide shot, to a two-person shot with the tablet, to the explanation of the camera movement, and finally to the call and zoom-in, the shot divisions and timing across the 23-second video all match the instructions.
The shallow depth of field is handled correctly. In SHOTS 2 and 3, only the two men in the foreground are in focus. The director, makeup artist, and stylist in the background are all blurred at the same depth. The model avoids a common AI mistake: selectively bringing only my face into focus.
The focus pull and continuous push-in work together. As the director turns, focus immediately shifts from the blurred background to my face, and the camera continues moving in until it reaches a close-up. Even in real filming, this kind of camera choreography takes considerable effort.
Skin texture and the hands around the face are detailed. In the close-up, you can see skin texture, the iris, and even fine strands of hair around the bangs. The face and fingers remain clear and undistorted, even as a makeup brush touches the cheek and the stylist adjusts the collar.
The character’s identity stays consistent. With just one facial reference image, I look like the same person in this indoor studio footage and in the midday outdoor commercial from Test Scene 2.

Final thoughts

Kling 4.0 is the first model that can realize the ad I envision “without compromise.”

For this test, every asset in the Reel—from the behind-the-scenes footage to the product close-ups and character action scenes—was created with Kling 4.0. The dialogue in the behind-the-scenes sequence has natural emotion, emphasis, and pronunciation, and can be used as is. The physical effects in the drink commercial are also accurate enough for commercial production. The biggest change is that I no longer need to switch between different tools for different scenes or repeatedly fix the characters and products in post-production. The whole production can be completed in a single model.

Kling 4.0 has also significantly changed the way I work. My focus is shifting from post-production editing to directing and shot planning. I can now generate an 18-second ad with 10 shots, or a 23-second video with 4 shots, in a single go. That cuts down the time spent splitting up shots and stitching them together, leaving me more time to design the shots and plan the camera movements.

Content that used to be more complicated to produce can also be handed over to AI. For example, dialogue scenes can be generated with expressive performances, without separate voice-generation or lip-sync tools. This opens up more possibilities for behind-the-scenes videos and narrative-driven commercials. Product close-ups are also more achievable: even small slogans on packaging can be reproduced accurately, expanding the range of product shots that AI can create for brand advertising.

Tutorials are also easier to reproduce. With organized reference images and clear instructions for every shot, it’s possible to generate the desired result consistently. That makes it easier for my followers to use the same approach and recreate similar work.

I believe the next stage of AI video shouldn’t be about producing images that are merely “amazing.” It should be about creating “a production set you can confidently entrust with the job.” Kling 4.0 has taken a clear step in that direction.

Thanks to @leedain_leader_in for sharing how Kling 4.0 helps bring together product details, character consistency, and natural performances in a single advertising workflow.

With this, our Kling 4.0 creator review series comes to an end. Thank you for following along, and revisit our previous stories for more firsthand insights from creators.