The thrill of seeing illustrations come to life as animation...! [Trying to make an anime OP-style MV with AI Vol. 3 | Animation Edition]

Hello, this is Kurono!
For this third installment, I've roughly summarized the process of creating illustrations and animating them for when I tried making an anime opening-style animation. (It took a month due to various things... haha)

It was a production process where I felt a lot of the joy of creation, so I would be happy if this article could serve as a reference or a starting point for others who want to try making something similar.

Topic 1: AI tools and adjustment tools used

I mainly used the following tools for illustration creation, animation, and video production. Details will be introduced in later topics.

Illustration: Nano Banana (Google AI Studio, Leonardo.Ai), Midjourney
Illustration adjustment: CLIP STUDIO PAINT
Animation: Vidu AI
Video editing: Filmora

Topic 2: [Illustration] You guys are from slightly different world lines... (NanoBanana/Midjourney)

Before animating, I first create the illustrations, but the hardest part wasconsistency in character style.
I already had image sketches of the characters from before, but since they were made in Midjourney and each character had a different style, I had to start by fixing that...

(1) Create a three-view drawing of the character
NanoBananaIf you first create athree-view drawing (in this case, a four-view drawing)in , the subsequent work becomes much easier. It helps unify the image, and it's more efficient than creating them each time when a scene comes up where you think, 'I want this scene from behind.'
If you want to unify illustrations to NanoBanana's anime style, it is recommended to use NanoBanana from the stage of creating character concepts.
I realized later that when creating other characters, attaching this boy's illustration as a reference image and adding instructions like 'anime style, match the style, make it a girl, make the hair blue...' allows for smoother creation of consistent illustrations.

This is because adjusting this girl was particularly difficult. Since I was correcting based on what I made in Midjourney, I really struggled to match the style with the boy. Once the image is solidified in your mind, you don't want to change the character design, and it was difficult to bring the art style closer from there. Even if I attached the image from (1) and gave instructions in NanoBanana, it would still slightly drag the previous style along.

In the end, I was able to manage the adjustments by creating various scenes in NanoBanana and repeatedly having her appear on the same screen as the boy to make the art style similar.

So, the art style is slightly different between the beginning and the end...

(2) Create illustrations by specifying characters and scenes

This illustration was created with Leonardo.Ai (Model: NanoBanana).
The prompt was adjusted while consulting with ChatGPT.

I wanted the boy's expression to be a bit more flustered, and the girl's bangs were a little different from what I imagined, so I corrected them using the method described later.

Topic 3: [Adjustment] Fixing that facial discomfort (CLIP STUDIO)

Even if the character turns out well, the bear might be strange, or the pose might be lackluster... I think there are probably better ways, but since I couldn't get it right no matter how many times I tried, I decided to 'redraw just that part' inCLIP STUDIOredrawing just that part. I was also bothered by the slightly yellowish color, so I corrected that as well.

At the time, perhaps due to the size of the original illustration, it sometimes wouldn't become 16:9, so I added the missing left and right sides inMidjourney.
Note that Leonardo.Ai seems to allow aspect ratio specification now.
However, I want it in 1920x1080, but it's 1344x768, so it's a bit sad that the size is small...

Topic 4: [Animation] Amazing, it actually animates properly (Vidu)

(1) Animating with Vidu

Viduis truly amazing...!! It feels like it understands my intentions and properly turns them into animation. I've only tried DomoAI and Leonardo.Ai (Model: Hailuo 2.3 Fast) besides this, but personally, Vidu is the easiest for me to get my preferred results and is the easiest to use. I'm also grateful that I can give instructions in Japanese...! I feel like instructions such as 'do XX at 3 seconds' sometimes work well too. (Though it might be a coincidence, it's a bit subtle...)

(2) Mass-producing illustrations with Vidu captures

If there is a good scene during the animation, you can turn it into an image using Frame Extraction. You can do this from the icon at the bottom right of the created animation.

(3) Creating connected animations

I created animations like Illustration A → Illustration B, Illustration B → Illustration C, and connected them to make a long video.
Vidu has an Extend feature, but considering it uses 16 credits for 4 seconds and 20 credits for 5 seconds, I personally feel it is more efficient to specify the final capture of the previous animation as the starting frame 1. (2-5 seconds is 10 credits)
By the way, if you want to loop it, you can achieve this by creating multiple animations of A → B and B → A.

※Fixing Vidu face to Banana face
This illustration is part of the Vidu animation, taken via frame extraction. The face looked slightly off in the middle of the animation, so I corrected it.

I am correcting the area around the eyes. This looks more like her...!

I feel the transition from the previous illustration (frame 1) has also become more natural.

This was a case where the Vidu animation felt a bit off, but sometimes it turns into a good illustration, so in those cases, I extract the frame and use it for a different scene.

Topic. 5: [Editing] Connecting animations into one video (Filmora)

I use Filmora for video editing. The reason is that Adobe's monthly fee was expensive...! However, for this level of usage, I think other software would be sufficient.
I used a simple approach for this work, only connecting videos and using transitions (and a few effects).

Finally

I was just so happy and excited to be able to make this anime OP. It was moving to see that even an amateur could create something like this. There is still room for improvement and many points to refine, but the experience of being able to "give shape to what is in my head" was something I never imagined a few months ago, and I think it was a truly great experience.

To think that AI can move characters so freely... I feel like I can express so many more things, and I am full of excitement for what's to come.

I want to create not just an anime OP, but a proper story as well, but I haven't been able to make much progress... These days, I'm thinking it's about time I started working on it in earnest.

💎Articles about the production are also being updated sequentially💎 ※Added (2026/7/31)

Published 2026/7/11💎

◆Related Articles◆