The Art of Generating Furry

This article is based on my own experience as an AI director.

To generate a furry character, different aspects of the character you want to generate must be taken into account, such as: the species. There is an infinity of furry species that can be generated within the AI that it is ridiculously impossible not to leave out any character that you have in mind.

The model

The first thing that should be taken into account is the model that we should use, since there are a great variety that can generate furry characters but they are not so effective to perform certain tasks, for this reason the model that is going to be used should be chosen very well taking into account the following:

Stability: That you set the model to generate a character in a constant way.
Mixed: Having a wide way to generate an endless number of varied characters, including humans.
Intuitive: make it understandable and easy to use.

I do not recommend the use of Pony models very much, since it does not obey the third condition, since this type of model is quite complex to use it, the same with the SD1.5 and SDXL version, since most are not so stable, however, the models based on Illustrious if they are more recommended, by showing better stability and control in the generation of images.

The Prompt

To start generating a furry character, the prompt is the fundamental part for it. Depending on the type of training that the model has through it we can be guided when it comes to starting to generate our characters

Order

Ordering a prompt at the time of writing it can help us to divide and identify certain sections of the music, so that when modifying, improving or changing we can identify the section to change:

Species: It should be described what kind of furry character is going to be created, as well as its sex and anatomical shape and body colos, eyes and id wears hair or not, for example: wolf, furry female, athletic, breasts, white body, green eyes, blue eyeshadow, red hair, long hair, wavy hair
Clothes: This is where we can describe the type of clothes she would wear, example: blue dress, short sleeves, cleavage, bare shoulders, bare thighs
Background: We describe the place where our character will be, the time, and the place, example: forest, night, full moon, trees, river, rocks
Pose and Situation: Here we will describe the position, the angle and the gesture that our character will have, example: sitting, on rock, yokozuwari, from above, looking at viewer, smile, :3, half-closed eyes, seductive
Aesthetics: it is where depending on the model the quality and style of the image will be described, it is to use for example masterpiece, high quality, high resolution etc., example: detailed, dynamic light, cinematic, EasyToon, toon style, cartoon logic

Leaving the prompt armed in this way:

wolf, furry female, athletic, breasts, white body, green eyes, blue eyeshadow, red hair, long hair, wavy hair, blue dress, short sleeves, cleavage, bare shoulders, bare thighs, forest, night, full moon, trees, river, rocks, sitting, on rock, yokozuwari, from above, looking at viewer, smile, :3, half-closed eyes, seductive, detailed, dynamic light, cinematic, EasyToon, toon style, cartoon logic

Guides

To write our prompt correctly to generate a good furry image, we must be based on the training of the model, when doing so we will have an idea of how to write it, for example, the SD1.5 and SDXL models are based on natural language therefore we must write as if we were talking to a person, but the Pony and Illustrious models are based on Danbooru tags, therefore in this case we would have to visit the Danbooru tag group to select the correct words of our prompt.

Lora

In certain special cases a lora can be used.

The style: Which style will be used for our character, cartoon, anime, or realistic.
Characters: Most models can parody furry characters from wigs and series, but in special cases it can be used

The negative prompt

It is essential to use the negative prompt to generate the image correctly, especially when the model adds elements that are not desired, for example if a character was given hair when it should not have been, in the negative prompt, hair is simply placed, and other elements that would be put in the positive prompt, we would put it in the negative.

Conclusion

In order to generate a good image of furry, it should be written in sections to identify the parts of it wearing, the species, body shape, clothing, background and condition, and aesthetics. A stable and intuitive midrange should be used. Use a lora only in special cases.

I hope this guide has helped you so that you can generate your furry characters.