Table of contents
What is this? The preview image looks terrible!
Can I see these pictures without my glasses?
Let the AI drawing move and come into reality - LoRA model-3D - Ana
What is this? The preview image looks terrible!
It's a LoRA, conceived and trained by Elldreth and TheAlly, that produces consistent, usable 3D relief images without any plugins or extensions, or expensive headsets*. Just enable LoRA, set intensity to 1, and prompt as usual.
If you don't have 3D anaglyph glasses, yes the image will look terrible!
What is an embossed image?
Anaglyph images are 3D images that use color to create the illusion of depth. This image is composed of two slightly different versions of the same image, one in red and one in blue. To see the full 3D effect, you'll need to wear a pair of relief glasses, which have a red lens and a blue lens.
The glasses allow you to see two different versions of an image, and your brain fuses them together to create a sense of depth. The rationale for using red and blue in these two images is that red and blue are the most distant colors in the spectrum and therefore provide the greatest contrast, resulting in a more pronounced 3D effect.
*Camera glasses are cheap and available on Amazon and other retailers.
Can I see these pictures without my glasses?
Here's a basic relief viewer. This will allow you to preview the image with some depth.
3Dthis.com - 3D Stereoscopic Viewer free online application
https://3dthis.com/stereo.htm
Precautions!
While this LoRA can consistently produce excellent results, it outputs images where the 3D effects don't seem to be working, sometimes producing choppy results. This is experimentation! As with all stable diffusion generations, there is an element of luck involved in seeding - please reroll.
Some modes may be better than others! We've tested some major models, our own mix of Lucid and Churned, and received positive results. It doesn't seem to work well with NAI and Anything v3, but does well with Berry's Mix.
Hiring fixes seems to either reduce the 3D effect, or produce a loss in image quality - be careful.
Also note that the sample images for this model are meant to give you an idea of what the model is capable of and to demonstrate the possibilities of depth effects, and are not necessarily 100% reproducible - due to the setup of my web UI.
You can use the depth map script plugin on Automatic1111's SD UI to create anaglyphs, it does a much better job! But I think LoRA and hints are more accessible to many people than complex extensions.