作者:闲散的不务正业的运维
Table of contents
1. Online address of Stable Diffusion model:
This is very important, as long as you click, you can send it.
https://inscode.csdn.net/@inscode/Stable-Diffusion
2. How to use Stable Diffusion
1. Click InsCode on the CSDN homepage
2. Find Stable Diffusion in the Template Center, click Use and Run
3. Choose a configuration you like to buy
4. Click WEB UI on the workbench to start online use
5. Operation page display
3. Model related version and parameter configuration:
Enter the AI drawing operation page, and you will find the following content:
- Model: You can choose the model you need
- Generation method: including Vincent graph, graph graph, additional functions, image information, model merging, training, Deforum, gallery browser, settings, extensions, etc.
- prompt word
- reverse cue word
- Sampling method
- Sampling iteration steps
- Width, height: image size
- Generate batches, the number of each batch: the number of batches generated each time, the number of each batch
- cue word relevance
- random seed
4. Prompt words and reverse prompt words for image generation:
-
Image generation prompts: generate corresponding images based on the words, phrases, or a sentence or a paragraph you provide
-
Reverse cue words: Related content you don't want to see
Image generation related parameter configuration:
- Model: GuoFeng3
- Generation method: Vincent diagram
- Prompt words for image generation: a sexy and beautiful woman
- Reverse cue words: colorful, festive
- Other configurations: that is, use the default parameters
5. Seeds and corresponding pictures:
- picture one
- picture two
- picture three
- picture four
- picture five
6. Summary
Generally speaking, corresponding pictures can be generated according to the relevant prompt words. It may be that the prompt words I gave are relatively few, so the generated pictures are somewhat similar. In short, the function is still very good.
It is recommended to use multiple prompt words, as detailed as possible, and the image that may be generated will be closer to the desired result.
I tried to add more prompt words, as detailed as possible, and sure enough, the picture was generated in the direction I expected:
Generate graph: