This workflow demonstrates the democratization of high-end VFX, proving that solo creators can now achieve cinematic fidelity once reserved for major studios. It is a compelling, if slightly uncanny, glimpse into the future of automated filmmaking.
Deep Dive
Prerequisite Knowledge
- No data available.
Where to go next
- No data available.
Deep Dive
I Recreated Attack On Titan Using Only AI (Higgsfield)Indexed:
Recreate ANY Anime using Higgsfield š https://higgsfield.ai?fpr=ai&fp_sid=conor10 In this video, I recreate five iconic scenes from Attack on Titan as live-action cinematic clips using AI tools inside Higgsfield, from the Colossal Titan reveal to Levi and Hangeās final goodbye. I walk through the exact image and video workflow step by step using GPT Image 2 and HappyHorse so you can turn your own favourite anime moments into realistic AI-generated scenes. Prompt Mentioned: Use @image1 as the reference. Live-action cinematic realistic style. Real-world physical look with natural skin texture, weight, and material detail. Match the framing, composition, characters, wardrobe, and environment exactly as in @image1, only translated into a live- action photorealistic look. Keep the camera angle and subject placement unchanged from @image1. contact: inquiries (at) creatingwithconor com
Have you ever wondered what Attack on Titan would look like if it was recreated into a liveaction film? I have.
That's why I took five of the most iconic moments from the entire show and rebuilt them as liveaction cinematic clips using nothing but AI. And by the end of this video, you'll know how to do the exact same thing with any of your favorite scenes. The platform that I'll be using to recreate all of our scenes is called Higsfield. And this workflow isn't just limited to Attack on Titan, but you can run it with any of your favorite anime. Since AOT is one of mine, I picked five iconic scenes to bring to life if you want to follow along with me and recreate your own favorite scenes. I've left a link to Higsfield in the description below. And just a heads up, there will be spoilers for the entire anime. Let's get into it.
The first scene is from season 2. Bold has just revealed himself as the Colossal Titan, and he tries to kidnap Aaron. So, here's how we start. When you're inside Higsfield, you'll see the navigation bar with lots of different tools at the top. First, click on image to open up the image creation workspace.
Once you're in there, you'll see a model selection. Click on that and choose GPT image 2. This is the model we're going to use for every single image in this video because it's one of the best at taking a reference frame and translating it into a photorealistic cinematic look while keeping the original composition completely intact. Now, in the reference image field, upload your screenshot from the show. I've got my clean 16 by9 frame of Birdold carrying Aaron. I'll write in this prompt. This is the prompt that I'll use for every screenshot we're going to recreate. So, save it. You can find it in the description below. Let's generate and see what comes back. The first thing you notice is how physical it feels. In the anime, this scene is intense, but it's still drawn. There's a stylization to it here. The weight is real. The posture carries actual strain, and the whole thing has this gritty war film quality. The framing is preserved exactly with the same angle and placement. But this is just the beginning. Now, we need to animate it.
Head back to the Higsfield homepage and this time click on video in the navigation bar. When the video workspace opens, you'll see the model selection on the left. Click it and choose Happy Horse. The reason I'm using Happy Horse specifically is that most AI video models have content restrictions that stop them from generating scenes from popular media like anime. Happy Horse doesn't have those same limitations, which makes it the only real option for a project like this. Keep that in mind if you ever try this yourself. Once you've selected the model, set the duration to 6 seconds, the resolution to 1080p, and the aspect ratio to 16x9.
Then upload the image we just generated as your reference frame. This becomes the starting point for the animation.
Now paste in the video prompt and generate the handheld camera movement alone sells this clip. Berthol's expression, the heavy breathing, and the tension from both characters are all elements that show physical pain. This is the kind of thing you'd see in a survival thriller.
For a first scene, this sets the bar high. The next scene, though, is a different kind of intense. This one is darker, and I'll be honest, I wasn't sure the AI was going to be able to handle it. I'm using this screenshot of a Titan about to eat Aaron's mom, and what happens next is one of the most disturbing things in the entire series.
I wanted to see what that looks like when it's real. And if the AI gets this right, it's going to be uncomfortable to watch, which means it worked. It's the same process as before. Back in Higsfield, go to image, select GPT image 2, and upload your screenshot. I'll paste in the same prompt that I showed you before and hit generate. This one is actually intimidating. This specific Titan in the anime already has this uncanny quality to her. And in the AI version, that quality is amplified. The skin texture, the scale, and her expression with her mouth open make this look like it came out of a horror film.
It didn't perfectly recreate the human's body, but that's only because in the original screenshot, Carla was drawn in a way where it was hard for the AI to notice the details. Now, let's animate it. Back to the homepage, click video, select happy horse, and apply the same settings as before. Upload this new image as the starting frame, and paste in the video prompt.
>> This is honestly unsettling, but in a good way. The fixed camera angle was the right call here. It forces you to watch it without any escape, and the sounds the Titan does make this feel all the more terrifying. Now, the next scene is one of my personal favorites from the whole show, so I hope the AI does it justice. I'm using this screenshot from season 4 where Hanji decided to stay behind alone and hold off the titans.
And before she goes, Levi places his hand on her chest, demonstrating the dedicate your heart salute, which was his way of saying goodbye. It's the most emotionally loaded moment in the entire show for me. and I wanted to see what it looks like in live action and if the weight of that goodbye is carried by actual human faces. Back in the image generation, upload your screenshot, paste in the prompt, and hit generate.
There's something about seeing this in live action that makes it hit differently. In the anime, the art style carries a lot of the emotion. But here, since it's just two people, it feels more raw. It even got their height difference right, which is quite funny.
The only thing it did wrong was that the AI mistook Hang's spears for iron poles, and it added a third one on the right.
But honestly, the main focus is our two characters. So, let's animate them. Back to video generation with the same settings. Upload the image and paste in the prompt.
The determined expression is everything in this clip. Ha's whole arc in the show is about someone who loves life and choosing to die so that others can live makes it come full circle. And that expression carries all of that. She looks happy with her decision, which means that the AI recreated it perfectly. The next scene is a completely different kind of moment.
It's very calm with little action and movement, but it affected the whole tone of the series going forward. In season 4, we see Aaron for the first time in years, and he's unrecognizable with long hair, facial hair, one eye covered with bandages, and an expression so dull you would never think this was the boy full of life we're so used to seeing. It's the moment you realize the vibe of the series has shifted completely. I have this screenshot right here, which is a close-up frame of Aaron's face. I'll upload it, paste in our prompt, and generate. This is the one that surprised me the most. The season 4 Aaron design is so specific that I wasn't sure the AI would be able to translate it faithfully, but somehow this looks exactly like him. It's like a casting director found the exact right person for this role. The bandage looks realistic, the hair falls the right way, and that cold expression is completely intact. If you showed this to someone who'd never seen the anime, they'd assume it was a still from a real film.
Now, to bring it to life, upload the image in the video generation workspace and set your settings. I'll paste in this prompt and generate.
There's almost no movement in this clip, and that's exactly what makes it work.
He just looks at you. And because the AI has rendered him with that level of realism, it carries the same weight as the original scene. That's Aaron Joerger in season 4. And now we're ending on the one that started everything. The scene that made millions of people realize this show was something completely different. The colossal Titan reveal in episode 1 is one of the most iconic moments in all of anime. Something so enormous it shouldn't exist. Staring down at a city full of people who have never seen anything like it. This is one of, if not the most important scenes in the whole show. So, let's see if the AI can replicate it. I'll upload this screenshot and generate with GPT image 2 using our prompt. I genuinely didn't expect this to work as well as it did.
The Colossal Titan is one of the most visually extreme designs in the show, and translating that into photo realism felt like it might be a step too far, but the result is honestly very accurate. The scale is there. He has his iconic steam, and the gritty texture carries the whole image. This is the most insane result we've made so far.
Now, back to video. Upload the image and paste in the final prompt.
The smoke is what sells it for me. In the anime, the steam coming off the colossal titan is one of its defining visual features, and the AI was able to recreate it, and the grin he gives to the camera is very intimidating. Now that all of our clips are done, let's assemble them into a final edit. For this, I'm using Cap Cut, but use whatever editor you're comfortable with.
Drag all five video clips into your timeline and arrange them in whatever order you like for your final edit. I want to make something that looks like a trailer, so I'm arranging them in the show's chronological order. Add some music underneath, something orchestral and cinematic that matches the tone of the show. Now, let's look at our final result.
This feels so cinematic, it could easily be a trailer for a live-action adaptation of the original anime. Let's compare it side by side with the anime version so that the differences are clearer. Honestly, both of them are incredible in their own way. With this workflow, you can recreate your favorite moments from any anime. And with Higsfield, you get access to everything you need in just one subscription. So if you want to recreate your own favorite scenes from Attack on Titan, use the link in the description to sign up to Higsfield. Thanks for watching and I'll see you in the next one.
Related Videos
Elon Muskās XAI, Fiber-Optic Drones & the New Era of US Defense & Winning the AI Arms Race
DefenseNow
250 viewsā¢2026-05-15
Decart Raises $300M to Build the Future of Realtime AI
DecartAI
252 viewsā¢2026-05-18
I Read Every Google Antigravity 2.0 Doc So You Don't Have To (13-Min Operator Playbook)
hyperautomationlabs1045
120 viewsā¢2026-05-19
Could AI change the future of cancer survival?
MotherConservative
999 viewsā¢2026-05-16
[RQ] All Preview 2 Midnight Horror School Deepfakes in Macbg Major
macbghuggylego
102 viewsā¢2026-05-15
Firefox on Android Just Added 'Shake to Summarize'
BrenTech
349 viewsā¢2026-05-19
Googleās NEW AI Just SHOCKED The Worldā¦
JulianGoldiePodcast
188 viewsā¢2026-05-21
WWDC 2026 Promises Apple Intelligence and Siri Upgrades | Episode 195
TheMacRumorsShow
104 viewsā¢2026-05-22











