By Kathy Mittens in News — 12 Dec 2025

Smart Banana for Smart AI Generation

Hey everyone! Here’s a new update from MeshZEUS.

You can now generate 2D images using Banana, Google’s powerful neural network. This makes it even easier to create detailed previews that can later be turned into 3D models.

For this test, I prepared the prompt for generating the model in advance:

I prepared the prompt for generating the model in advance:
A photorealistic image of an anthropomorphic female cat with light blue fur, standing upright on two legs like a human. She has a sexy, confident appearance with long, slender legs and is wearing sneakers. Her outfit includes tight-fitting jeans with a brown leather belt and a white shirt. She has violet eyes, her arms are crossed over her chest, and she has a displeased, annoyed expression. Highly detailed fur texture, realistic lighting, natural shadows, and lifelike proportions. white background

Let’s create two previews — one using the previous model and another using Google Banana.

Even though we used exactly the same prompt, the results turned out very different. Each model interpreted the description in its own way.

Banana’s blue cat actually looks way more human-like and a lot less cartoonish. She’s got long legs, a clearly feminine shape, and yep — that grumpy face we asked for is definitely there.

Let’s generate a 3D model based on the image from Banana and see how it turns out. As we can see the blue cat looks good from all angles and matches our expectations. You can export the model and place it on your land in Second Life.