«Image Size». This option allows you to set the size of the resulting image. I highly recommend that you use the 512x512 size. After you select the one you like from the many images you have created, you can use the «Upscale» / «Scale» command to enlarge the image to a size of 2048x2048 and download it to your computer.
«Inference Steps». The picture we create at the first stage consists of so-called random noise. The program, analyzing our request, reduces this noise step by step and gives us the final version of the image. This option determines the number of steps after which we obtain a sufficiently detailed image. You can try changing it by setting a value in the range from 25 to 50. Increasing the value affects system performance. The optimal ratio of speed and quality of Easy Diffusion is achieved at a value of 25.
«Guidance Scale». This option allows us to tell Easy Diffusion to analyze our request more closely so that the final image it produces more accurately matches its description. The optimal value is 7.5. The higher the value, the less creativity Easy Diffusion has.
«LoRA» (Low-rank adaptation). LoRA models are used to fine-tune and speed up the large models that we previously downloaded and installed and will use to create our images. There are various LoRA models that are suitable for enhancing some part of the image (for example, eyes) or style (for example, anime). By default, they are not installed on the system. You can download them from the website civitai.com by typing «LoRA» in its search bar. We do not change this option and leave it disabled (empty). However, I recommend that after training, you go to civitai.com and download the models that suit you for further experiments.
Seamless Tiling. This option allows you to make a double image seamlessly stitched together from one of your images, that is, make a panoramic image. We do not change this option and leave it disabled (empty).
«Output Format». This option specifies the final image format. You can choose three: jpeg, png and webp. I recommend leaving jpeg.
«Image Quality». The default value is 75. If you want a higher quality image but need to wait a little longer to create it, set the value to 95.
We will look at the «Render Settings» functions later in the exercise. For now we leave them unchanged.
Lets start creating our first image.
In order to create our first image from text, in the «Enter Prompt» window we need to write our first text request («Prompt»).
Lets do the following:
In the «Enter Prompt» window, write the following request: «Artificial Intelligence that will change the world, beautiful cyborg face.» Please note that Easy Diffusion only understands queries in English. If you are not very good at it, please use any online translator available to you.
In the «Image Settings» section, select the «sdxlYamersRealism_version2» algorithm.
Image Size: 512 x 512.
Leave all other parameters unchanged and click the «Make image» button.
If you open the application backend window, you will see how the Easy Diffusion system works:
Easy Diffusion may need to download additional files to create our image. After which, after a few minutes, she will begin to create a new image based on our request.
Switch to the Easy Diffusion graphical interface window.
Hooray!
You have created your first unique image using an Artificial Intelligence system. Its located to the right of the Toolbar in the workspace of the user interface:
If you dont get the same picture as me, dont be alarmed and try to create it again (I didnt succeed the first time either). Very often in Easy Diffusion you can get the desired image the second, third or fifth time. This is a feature of the program. This often happens with simple queries, as in our example. Well look at complex queries in the next chapter.
Our next step will be to explore the enhancement and manipulation options that most often need to be done after we create any image. To do this, hover your mouse over the new image and you will see the following commands:
«Upscale». If you like the new image and feel there is no need to improve it further, run the Upscale command. After which the image will be enlarged to a maximum resolution of 2040 x 2048 and you can save it on your computer.
«Fix faces». The team corrects faces in images. Most often, fingers and facial details are affected in images.
«Draw another 25 steps.» If you think the image needs to be improved, run the «Draw another 25 steps» command. It often happens that there will be more details in the picture and its quality will become better. It also happens that no significant changes are noticeable. This requires patience and perseverance.
«Make Similar Images.» With this command we tell the system to make five more similar versions of the images and the first received version.
«Download Image». The command saves your image in the location you want.
«JSON». The command saves the description and all parameters of the image you created in a marked file. You can load it the next time you want to create a similar image by clicking the «Load from a file» button.
«Use as Input». This is a very useful command that allows you to improve the image from time to time. When using it, the Artificial Intelligence system takes as a basis not only our request, but also your image, on the basis of which a new image is created. We will look at the work of this command in more detail in the next chapter.
Now lets take another 25 steps to improve our image by running the command «Draw another 25 steps»:
As a result, we will get a similar image, but with some changes:
Now lets go back to our first image of «Artificial Intelligence» below.
Hover your mouse over the image and select the «Make Similar Image» command. This command will allow us to create five more new image variations from the original one.
Five new options will be placed in one program window (for clarity, I left four):
You can use this function an unlimited number of times until you get the perfect image. This is exactly what I did when choosing pictures for the book cover.
Once you select the perfect one, you can delete all other images. Hover your mouse over the image you want to delete and click the cross in the back right corner:
As a result, only those images that you need for further processing will remain in the window.
As I noted earlier, all the models that we additionally installed in Easy Diffusion, without exception, have one significant drawback they do not draw eyes or fingers well, and very often various strange artifacts appear in the pictures. To correct the eyes, in the image we like, we need to move the mouse over the picture again and execute the «Fix Faces» command:
As an example, Ill leave one AI face and run the Fix Faces command on that image.
It is necessary to make a small remark and clarify that the «Fix Faces» command quite often does not improve the image, and sometimes spoils it. However, its always worth trying out what might come out of your experiments. Be bolder and experiment more often.
In my case, the faces turned out a little blurry or blurry:
The final result (which we like best) needs to be converted to a higher resolution and saved on our hard drive in order to later use it for further work. To do this, execute the «Upscale» command by hovering the mouse cursor over the picture.
After this, execute the «Download Image» command. The downloaded image will have a resolution of 2040 x 2048 pixels per inch. This is quite enough to design a book cover or make a beautiful postcard.
Congratulations!