Loading
November 22, 2026

What you’re seeing on screen right now are ads created entirely using AI, not for small creators, but for some of the biggest brands in the world. Instead of hiring video editors, these companies directly went to AI specialists or prompt engineers, and this is not a future prediction, this is happening right now. AI has basically become the new cameraman, the new editor, and the new cinematographer, and honestly, the results are shocking, as some of these clips look better than real production. Did you see the jump from Nano Banana to Nano Banana Pro? The improvement is insane, and we have reached a point where you genuinely cannot tell what is real and what is AI-generated. Also, we’ll talk about the best video editing model that we have ever seen, which is the world’s first unified multi-modal model that does everything in one place. Think of it like Nano Banana, but for videos, where you can add, remove, or modify anything you want, such as changing the background or environment, or adding, swapping, or removing objects. So in this video, I’ll tell you the best techniques that you can start using in your own workflow because, like it or not, the industry has already changed and the time to ignore AI has long gone. I’ll be using Higgsfield to test out the latest techniques, capabilities, and models.

What has changed? Think about this for a second. ChatGPT came out only three years ago and today almost everyone uses it. At first, people were skeptical, unsure, and hesitant, and then suddenly it became the default tool, and right now I don’t know anyone who doesn’t use ChatGPT. The same shift is happening with AI video, where for the past years thousands of new tools launched, but most were not good enough for real creative work. They were fun to experiment with, but not something you would use in a real client project until now. Nano Banana 2 just released, and for the first time, this feels genuinely production-ready, something you can confidently use in real edits, ads, and client projects. This is what makes this upgrade a massive jump. First, physics-based simulations where you get real-world weight, gravity, reflections, materials, and motion, which means editors can now generate realistic product spins and floating shots without 3D or VFX, and you can also create natural slow-motion movement without simulations.

Second, Google-level world knowledge where the new models understand maps, architecture, lighting, and real environments, meaning as an editor you can build a studio room, cafe, or subway scene that looks exactly like the real location, and you can also instantly recreate YouTube-style backgrounds or trending creator setups. Third, geolocation rendering, which means you just enter coordinates and time and it will give you accurate sunlight, shadows, and atmosphere, allowing an editor to match outdoor lighting in multiple shots using AI. Fourth, and the most important one, is character consistency, where faces, geometry, and scenes stay consistent across generations, meaning you can create 10 to 20 shots of the same actor for an ad or story without face glitches, and build a complete sequence with wide, medium, and close-up shots with matching appearances.

Fifth, knowledge-based visuals where the models know how objects move and interact naturally, meaning hair, clothing, and props move realistically without manual animation, so with these upgrades you can start creating visuals for actual client projects. The first part that I wanted to try was visual effects inside Higgsfield because usually when we are working on client projects and need a wow factor, we rely on VFX, which takes hours and hours of work from tracking, compositing, masking, and keyframing, taking a few days. Inside Higgsfield, I found hundreds of plug-and-play VFX templates where you don’t have to build anything manually. You literally just replace the placeholder image with your own videos and the effect is done. Here are my favorite effects that I tried. First, Earth zoom in and out, which is perfect for viral Instagram reset videos or any cinematic location reveal transition that normally takes multiple layers and keyframes, but here you can generate it in one click.

The second one that I really liked is the wireframe effect, which turns your subject or background into a digital wireframe model, making it super futuristic, clean, and technical, and perfect for montage transitions, tech-style videos, or sci-fi edits. The third one is building explosion, which is a full building explosion you can drop into a montage or action-style edit, something that would normally require 3D assets or stock VFX, done in just a few minutes. Honestly, these are just a few. There are hundreds more inside the library that you can mix, remix, and stack to create insane transitions in minutes by just picking the movement and uploading your frame so your shot instantly feels like a real production. The second feature that I really like is camera movement in Higgsfield, which solves one of the biggest problems in client projects when they give you static shots with zero movement. Normally we would add keyframes and a fake zoom-in that doesn’t look truly cinematic, but inside Higgsfield you can get simple camera movements like dolly in, dolly out, handheld drift, or subtle parallax motion, or go for epic camera moves like bullet time, FPV drone, Dutch angle, and through-object shots that feel like they were shot using real rigs, making even boring still shots cinematic and dynamic.

The next feature that I really liked is making your own AI character or avatar in Higgsfield, which is basically a digital version of yourself or your client that you can use for endless content. You just upload 20 to 30 photos of yourself, the system learns your unique look in a few minutes, and after that you can generate infinite selfies, portraits, and stylized shots without ever taking a real photo, which is why people are creating AI influencers who post every day without a single photoshoot. For editors and brands, this is extremely useful because if a client wants to promote products, you can create their character using 20 to 30 images, place the character plus product in Nano Banana, animate the shot with any video model, and add camera movements for a complete photoshoot or video shoot without cameras, sets, or retakes. Next, I want to talk about Soul Inpaint inside Higgsfield, which is AI-based retouching for video where you just highlight a part of a shot and type a prompt to change clothes, hair, objects, or the background while keeping the exact same face identity so the character never breaks. This is insanely useful for video editors to fix outfits, change styles instantly, remove background distractions, add props, or create multiple ad variations without reshoots, masking, rotoscoping, or frame-by-frame work. Now we come to the craziest part, the most insane AI video editing model we have ever seen, Kling O1, which is Nano Banana for videos, letting you edit a video like a Photoshop file to change backgrounds, swap objects, replace people, ask for new angles, or modify the whole scene just by typing what you want. It gets even crazier when you upload reference images of your character from multiple angles along with logos, products, props, or style references and tell it how to integrate them, as the model understands visual context and can even track multiple characters separately with pixel-perfect consistency.

✨ IMAGE PROMPT
Prompt Prompt A Uploaded my photo cinematic, ultra-realistic close-up portrait of a handsome young man with long, thick, wavy black hair and a same beard, looking thoughtfully toward the right side of the frame. His face is captured in a striking side profile, with a small beige medical bandage across the bridge of his nose. He wears a simple white crew-neck T-shirt. Dramatic warm golden sunlight illuminates the side of his face, highlighting his sharp facial features, natural skin texture, and flowing hair. Dark, moody, softly blurred charcoal-grey background with subtle warm bokeh lights on the right. Subject positioned on the far left, leaving generous empty negative space on the right. Shallow depth of field, cinematic color grading, dramatic chiaroscuro lighting, professional editorial photography, 85mm portrait lens, photorealistic details, 8K quality, widescreen 16:9 composition.

To use it, go to Higgsfield, navigate to Kling O1 advanced video editing under video, upload a reference video along with your images or elements, describe what you want in plain English in the prompt box, and hit generate, giving editors a VFX, compositing, and cleanup team inside a single model. Everything I’ve showed you in this video is just the beginning, as Higgsfield has barely-scratched-the-surface features like click-to-add, draw-to-video, sketch-to-video, UGC factory, video upscale, product placement, and multi-reference editing. The platform is moving so fast that the best way to learn is by trial and error. Test the models, break things, and tweak prompts to figure out what works for your style, because this is the future of editing and we are just getting started.

Template

XML file

Song link

Leave a Reply

Your email address will not be published. Required fields are marked *