In this section, we'll dive into theStable DiffusionYou'll learn the definitions, how to download, where to install, how to use, and scenarios of use for the two important miniatures in Embedding (Text Inversion) and Hypernetworks.
I. Embedding (Text Inversion) Details
1. Definition and importance
Embedding, also known as Text Inversion, is an important tool for adjusting the inputs to a large model.
Beginners often confuse it with Text Inversion, they are actually the same concept.
The role of Embedding is the packing of cue words to achieve a specific output effect by adjusting the inputs of the Unet part of the larger model.
2、Download and Installation
Embedding can be downloaded in a similar way to Lora, either through the C site or the launcher. To install, place the downloaded file directly into the embeddings folder in the Stable Diffusion root directory.
3. Usage
To use Embedding, simply select the installed Embedding by clicking on it in the additional web interface and it will be automatically invoked.Embedding is simple to invoke with a single click and requires no additional weight settings.
4. Weighting adjustments
Embedding's weight adjustment is similar to cue word adjustment, which can be directly bracketed by the cue word method and then add a colon to modify the weight. For example, by setting the weight to 0.2, the image will be slightly characterized by Embedding.
5. Use of negative cues
Embedding can be used not only in positive cue words but also plays an important role in negative cue words. By adding Embedding in negative cue words, the quality of images can be improved effectively.
Application of Embedding to Negative Cue Words
sketches.
(worst quality:2), (low quality:2), (normal quality:2), lowers, normal quality,
((monochrome)), ((grayscale)), (grayscale).
Facing away, looking away.
text, error, extra digits, fewer digits, cropped, jpeg artifacts, signature, watermark, username, blurry,
skin spots, acnes, skin blemishes, bad anatomy, fat, bad feet, cropped, poorly drawn hands, poorly drawn face, mutation, deformed,
tilted head, bad anatomy, bad hands, extra fingers, fewer digits, extra limbs, extra arms, extra legs, malformed limbs, fused fingers, and
too many fingers, long neck, cross-eyed, mutated hands, bad body, bad proportions, gross proportions, text, error, missing fingers,
missing arms, missing legs, extra digit, extra arms, extra leg, extra foot, missing fingers, mole:1.3. EasyNegative
II. Introduction to Hypernetworks
1. Definitions and functions
Hypernetworks is a network that generates weights for other networks and can be seen as a weakened Lora. it is similar to Lora in file size, functionality, and usage.
2、Download and Installation
The download of Hypernetworks can be done via the C-Site or the launcher. To install, place the downloaded file in the hypernetworks folder inside the models folder in the Stable Diffusion root directory.
3. Usage
Hypernetworks is invoked in Stable Diffusion's interface by attaching a network extension model.The invocation is similar to Lora, requiring the use of pointed brackets to enclose it, and the specification of a filename and weights.
III. Practical cases and application scenarios
1、Embedding's practical cases
For example, using Embedding's dva game character model, we can replicate the characteristics of the character dva.
By adjusting the weights of Embedding, we can control the strength of character features to generate images with consistent style.
Weight = 0.7
2、Hypernetworks application scenarios
Hypernetworks is less used in real projects, but it still has its value in specific scenarios. For example, it can be used to generate cartoon-style images, or when a weak Lora effect is required.
IV. Conclusion
In this section, we not only learn how to use Embedding and Hypernetworks, but also understand their application scenarios and development history.
I hope this article will help you better understand and use the miniatures in Stable Diffusion.
