Image to Image
Generate images based on image input
Input image obtained from the Unity camera or external images
Image to image allows you to add new elements for blending for original image
Currently (April 2023), Image to Image also recommends the use of WebUI, as it is possible to use the Inpaint function and other functions that allow partial modifications by using WebUI.
Since stable-diffusion-for-unity runs StableDiffusionWebUI, it is possible to use the WebUI functionality from the local server URL listed in the command prompt started by StartServer from Unity.
For Image2Image on Unity
For generate image with image to Image, attach the Img2Img.cs component to the object
It is assumed that the local server is up and running.
Please refer to the following for StartServer before use.
🖥️About local serverA parameter summary is shown below. Parameters that are adjusted frequently are bolded.
ImageSource
Whether the input element is a prepared image or a camera input in Unity
Image
If Texture is selected in ImageSource, select the image to be used as input Details are listed at the bottom of the table
Prompt
Enter the words you want to generate in natural language
NegativePrompt
Input elements in natural language that you do not want to include in the generated image
Width
Width of the generated image in pixels
Height
Height of the generated image in pixels
Steps
Increase this parameter if the generated image looks blurry
Cfg Scale
Strength to follow prompts Input larger value to strongly follow the prompt.
Denoising Strength
How strongly follow the input image If you want to change the input image only slightly, decrease the value. If you want to change the image significantly, increase the value.
Seed
Fixing the seed value to generate a similar image
In case of -1, the image will be random
Batch Count
The number of images to be generated.
Select 1 to 100 images.
Generated images will be stored in StreamingAssets/StableDiffusion. Please check images by date sort in the explorer.
ExportType
Selectable output types: PNG, JPG, TGA
Apply to Image, RawImage, Material, etc
Sampler
Changing the nuances of the generated image
Model
Base trained model See the "Switch Models/LoRA" to switch new model
Lora
Additional trained Lora to generate consitent specific pattern image
See the "Switch Models/LoRA" to switch new LoRA
If you select Lora, <lora:filename:1>will be added to the Prompt and LoRA will be available
Settings when Texture is selected as input
Check >Read/Write in ImportSettings of the image.
Last updated