Update the workflow in April 12th with ipa plus 2.0

This workflow can turn your flat illustration into a 3D image without entering any prompt word.

How to use: https://youtu.be/AuFbkvgMsSE

It contains six modules:

1.Input modules:

You should input your flat illustration here,the clearer the better.

2.Model modules:

Checkpoint: revanimated  https://civitai.com/models/7371/rev-animated

Lora：3D rendering style  https://civitai.com/models/73756/3d-rendering-style

3.Gemini modules:

I use Gemini to read your illustration and guess the prompt word. You should have your api of gemini to make it work.Please install this node from manager.(https://github.com/ZHO-ZHO-ZHO/ComfyUI-Gemini)

First: Install Comfyui-Gemini from manager

Second: apply the api: https://makersuite.google.com/app/apikey

Third: add your Gemini api into config.json(which is in the comfyui_gemini folder),and save it.

Then restart your comfyui.

If you can't use Gemin, you also can use wd1.4 to replace it.

4.Controlnet modules:

I use scribble model here,  If you want a more accurate result, please switch to lineart model. I think scribble can make its 3D structure more reasonable and lineart model make it more accurate.

There is an example for the differences:

5.Ipadapter modules:

ipadapter models:  https://huggingface.co/h94/IP-Adapter

Please use IP-Adapter for SD 1.5 here.

6.3D result modules:

Your can get the 3D image here.o