How to make 3D Documentary like Fern (Nano Banana Tutorial)

You may already know that Google Nano Banana can generate perspectives.

But did you know it can now replace 3D animation artists?

Take a look at this 3D animated documentary video and see for yourself.

On the night of December 22nd, 1999, a man named Dante was working late at his office.

At around 9:50 p.m., his dog began to growl at the window.

Curious, Dante went to investigate, he walked up to the window to see what the dog was growling at.

Outside, a thick fog had started to roll in, shrouding the night in silence.

But what Dante didn't realize was that someone or something was already watching him.

Within the next six minutes, Dante's life would depend on every choice he made.

In this video, I'll walk you through the prompts, tools, and techniques I used so you can make your own too.

Okay, the first piece of the puzzle is generating the image.

We want to get that 3D rendered style with faceless humanoid figures, smooth surfaces, simple textures, and even different perspectives of the same scene.

We're going for something that looks like it was animated in Blender, but it's actually not.

I'll show you the prompts I used to generate these images and a tip to write a prompt for any image you want.

Just find an image in any style you want and feed that to ChatGPT, then ask, "Give me a prompt to generate image like this, the style should be 3D rendered minimalist animation."

Then it will give you a detailed prompt that can generate a close matching image.

Just for the first image, I recommend using Google Image FX, after you enter your prompt, you will get four different images, if you see one you love, you can click this download option to save it.

If you think my editing is improved, that's because I've been putting more time and effort to make them just for you, your comment will help me cherish the effort I put into my work.

Back to the video.

Okay, once you have your first image, we are taking it to the next step, which is generating the perspectives.

And no, it's not straightforward as pressing a button, but don't worry, I will show you how I did it step-by-step so you can take that to inspiration.

The character in this image is standing, but I want an image of him sitting and writing something on the paper.

This is where we use Google Nano Banana, there are many ways to use Nano Banana.

And out of them, Gemini and AI Studio are the free options, I personally like to use the AI Studio, but try both and see what works best for you.

Simply feed the image into the AI Studio and tell what kind of changes you want, since I want to change the pose, I typed, "Make this character sit down and pretending to be writing something on the paper."

Then run the prompt and see the change.

It's simple as that, you just need to give it clear instructions.

So, first I will get this image animated using an AI video generator named ChatGLM.

That can be used for free, there is a guide I made previously on free AI video tools and how you can access them, so watch that one if you're having difficulty on accessing this tool.

To animate this image, go to AI video, upload the image, below that you can enter a prompt.

And here's the tip, first describe the character's actions, in here, character is writing, camera zoom in slow.

If you don't want any camera motion, type no camera motion.

Then use keywords like 3D rendered animation, smooth motion, high quality afterwards.

Play with the settings to get maximum quality results and simply click generate.

You may see a watermark at the bottom right, you can use a video watermark remover or just trim that part out when editing.

In our video, after this scene played, camera switched to side view.

To get that perspective, simply copy the image we generated and paste it in the text box, now type, "Create me another image of this scene but from a different camera angle," and inside the brackets, type the perspective you want.

Use this prompt to generate an image that have exact same elements but with a different view.

Let's say you want a window here instead of the bookshelves, copy the new image, paste it in the text box and type, "Remove only the bookshelves in the wall in front and add a window, one tree trunk can be seen in outside and it's night time."

I was happy with the result, so I took it to video generation tool, adjusted the prompt to give it a new camera motion and generated the video.

So the next scene, I have an angled zoom out view which shows the room, and the character gets up and walks toward the window.

Here's how I did that.

First, I uploaded both images that show front and side view and typed this, "These are two perspectives of the same scene (front view and side view), generate me another scene of the same scene but with different camera angle, make sure it's taken from a corner of the room so the whole room is visible."

It generated a good image, but the desk was facing the wrong direction.

So I included that image in the prompt and asked, "Rotate the desk (keep the walls, window, bookshelves same), the window must be on his left side and bookshelves must be on behind to him."

Now the desk is facing the right direction, but the window and bookshelves are not in correct position, so I copied and pasted this new image and typed, "Swap the window and the bookshelves, keep everything else same."

Which it gave the result I wanted, I wanted to add some chairs and a carpet, so I used the newly generated image and typed, "Place two small chairs in front of the desk, add a red color carpet to floor."

So the result can be used again to generate the same scene but with the character standing in front of the window, I used following prompt to that, "Generate another image of the same scene but with the character standing near the window, looking outside."

The result was perfect, but I wanted to move the chair a little bit away from the table, so it looks like chair being pushed away when he stands up in the video.

So, I used this new image and typed, "Keep everything same and bring the black office chair little bit away from the desk and slightly turn it toward the window."

Okay, so now we have two images of same scene with character sitting in one and standing near the window in the other.

First, I brought the image where character is sitting into the AI video tool and typed, "Character stops writing and looks at the window, 3D rendered animation, smooth motion, high quality," alongside with no camera motion.

After generating the video, I saved the last frame of this video as an image and uploaded again, this time, I upload the image where character is standing as the end frame, for the prompt I typed, "Character walks to the window."

And that's how I created this scene where the character stands up and walks to the window.

Okay, so I'm not going to get you bored with more image creations, the technique is the same no matter what you create, just use a reference image and type what you want to see clearly and in simple.

If you want to see the prompts I used for each image, I've composed a PDF with all of the prompts, so if you need that, type 'prompt' in the comment section so I will send the link to you.

Anyway, let's take a look at the results one more time, thanks you for watching and I'll see you next time.

On the night of December 22nd, 1999, a man named Dante was working late at his office.

At around 9:50 p.m., his dog began to growl at the window.

Curious, Dante went to investigate, he walked up to the window to see what the dog was growling at.

Outside, a thick fog had started to roll in, shrouding the night in silence.

But what Dante didn't realize was that someone or something was already watching him.

Within the next six minutes, Dante's life would depend on every choice he made.