This guide is based on Stable Diffusion web UI (A1111)
This guide generates text "hidden" in the image. To see it, you'll have to squint or stand back. It's the trend of the moment, and I'm sure you can't wait to find out how it works!
"Hidden" text examples
[[ Look at the attachments for 512x512 examples of masks ]]
1. ControlNet Installation
On A1111 WebUI, go to the extension tab and follow the steps in the image below
Then we have to restart stable diffusion to complete the installation.
2. Download and install the ControlNet Model used for generating hidden text images.
Download control_v1p_sd15_qrcode_monster.safetensors and place the safetensors file inside :
📁 stable-diffusion-webui/extensions/sd-webui-controlnet/models
3. Generating the image with text2Img
The params below are used to generate the image of this guide. But you can try whatever you like.
Once you unfold the ControlNet tab, follow the settings below. Here we'll use the "Ingrid" text image for our generation. It's a simple b&w image.
The dimension are 768x512 because I wanted a large text. You can use any size and ratio you like. ❗ Be sure to match the mask dimensions with the width/height parameters of stable diffusion.
[[ see attachments to download sample masks ]]
The Control Weight variable is very important, as it defines the subtlety of text appearance
[1...1.15] = very subtle or big lettrine
[1.15...1.35] = good result 👌
[1.35...2] = more pronounced or thin lettrine
4. Result
Now that everything is ready, you can click on generate and admire the results 🎉
I hope this tutorial will be helpful to many. I wish you a happy generation friends !! 🤗