Over the past few days, I discovered an AI tool that converts images into 3D models, and I’ve been having a lot of fun playing with it. In this article, I’ll walk you through the entire process. The first step is to create an account on this website here, which currently offers the tool for free. You can generate up to 20 models per day, which is more than enough for casual use.
The site is in Chinese, so unless you're fluent, you’ll need to translate the page. The translation method depends on the browser you're using. I’m using Google Chrome, which simply requires clicking a small translation icon in the address bar.
Once the site is translated, follow the instructions to create an account. There are several methods available; I chose the email option, which sends a verification code. After creating your account, log in. You should see an interface similar to this:
Now you're ready to create your first 3D model. All you need is a picture of your character with a simple background. Keep in mind that the image should remain within an R rating—just make sure to cover any “naughty bits.”
I’ve created a LoRA that helps generate three views of your character—front, left, and back—which you can find here. You don’t necessarily need all three; a single front-facing image is enough to generate a model. However, if you plan to rig the character later, it helps if the character is in a T-pose. My LoRA doesn’t generate an exact T-pose (the arms are slightly lowered to keep the character inside the frame), and this can cause issues around the armpits where the arms may merge with the torso in the 3D model. If you can produce a perfect T-pose, your rigging results will be much cleaner.
For this example, I’ll use my OC character LoRA Not Starlight. After generating an image using this prompt: "dazturnaround, simple background, grey background, turnaround, multiple views, ((from front, straight-on, from side, from behind)), st4rl1ght, 1girl, blonde hair, amber eyes, dark eyeshadow, white hairband, superhero, white dress, dress, pantyhose, belt, knee boots, yellow footwear, full body, dark background" and running a hires fix, this is the result:
After some cleanup and using ADetailer for the face, here is the final image:
Next, you’ll need to split the three views into separate images:
Once that’s done, upload the images into the proper slots, select the resolution of the generated mesh and click Generate Immediately. After a few minutes of processing—congratulations! You’ll receive a fully textured 3D model of your character.
The model usually looks great right away. The texturing can be slightly imperfect at times, but if you have some modeling experience, you can refine it further: remesh, repaint, rig, and even animate it.
You can also try the site’s automatic bone binding option and watch your character come to life:
I hope you found this article helpful—and that you have tons of fun bringing your characters into 3D!