The Appeal of the Kawaii Realistic Series: From Anime and Manga to Realistic Styles
A Collection of Diverse Personalities and Cultures – The “Kawaii Realistic” Series
Hello, this is kojirom.
This time, I will introduce all 4 works of the Kawaii Realistic series.
They are all equally wonderful, regardless of whether they are anime, manga, European, or Asian style. The high quality is truly remarkable, with hardly any flaws. The level is now comparable to that of large companies such as DALL-E and Bing Image Creator.
The settings and negative prompts were based on the image data published on the author’s page in CIVITAI. The details of the settings are attached at the bottom of the page.
The positive prompt for this time is unusually long. I couldn’t think of it myself, so I asked ChatGPT and Claude for ideas, and it was decided as follows.
One girl (1girl)
Wearing warm winter clothes in a snowy town (wearing warm winter clothes in a snowy town)
In a swimsuit (wearing swimwear)
I entered the above three prompts into the X/Y/Z Plot of Stable Diffusion to generate images.
By using X/Y/Z Plot, I can visually compare the results of inputting the same prompt to different models, which is very useful.
Then, enjoy the world of cute and realistic AI images!
Kawaii Realistic Anime Mix – A0.4
Kawaii Realistic Anime Mix – A0.4 | Stable Diffusion Checkpoint | Civitai
Kawaii Realistic Asian Mix – v0.6
Kawaii Realistic Asian Mix – v0.6 | Stable Diffusion Checkpoint | Civitai
Kawaii Realistic European Mix – v0.5
Kawaii Realistic European Mix – v0.5 | Stable Diffusion Checkpoint | Civitai
Kawaii Realistic Manga Mix – v0.2
Kawaii Realistic Manga Mix – v0.2 | Stable Diffusion Checkpoint | Civitai
Kawaii Realistic Anime Mix
Note: Site Summary
VerA is more animated and VerR is more realistic.
Since it varies depending on the prompt, please use Negative Prompts or other tools to adjust the settings.
The sample VAEs all use “anything-v4.0.vae”.
Kawaii Realistic Asian Mix
Note: Site Summary
I’m combining multiple models for efficient results with fewer prompts, including NSFW content. Different versions cater to varied preferences; you can choose your favorite or blend and adjust them. Recommended negative prompts include 3D, 3D render, painting, digital painting, watermark, sepia, black & white, NG_DeepNegative_V1_75T, EasyNegative, with tags to avoid low-quality and monochrome features.
Kawaii Realistic European Mix
Note: Site Summary
I’m creating this by combining various models for effective results with minimal prompts. Although adjusted by a European model, it’s capable of producing diverse racial representations. Users can choose from different versions or merge and adjust to their preference. This includes NSFW content. The negative prompts include 3D, 3D render, painting, digital painting, watermark, sepia, black & white, NG_DeepNegative_V1_75T, EasyNegative, with filters to avoid low-quality and monochromatic outputs.
Kawaii Realistic Manga Mix
Note: Site Summary
The model is adjusted by merging the animation and illustration models with a few realistic models.
The sample VAEs all use “anything-v4.0.vae”.
- Please note that some illustrations may have unfinished details, such as hands or fingers.
- Feel free to crop or edit these images as needed.
- Due to uncertainties in the licenses of the source models used for the illustrations featured in this article, commercial use is not recommended.
Settings and Parameters for This Time’s Stable Diffusion Web UI
Negative prompt: 3d, 3d render, painting, digital painting, watermark, sepia, black & white, EasyNegative,(worst quality:2), (low quality:2), (normal quality:2), lowres, normal quality, ((monochrome)), ((grayscale)),(ugly:1.33), nsfw, underwear, bra, pantie, camel toe, nipples,
Steps: 27, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 2412681505, Size: 512×512, Model hash: 5af0d0b777, Model: kawaiiRealisticAnime_a04, VAE hash: f921fb3f29, VAE: anythingModelVAEV40_v10.pt, Denoising strength: 0.65, Clip skip: 2, Hires upscale: 1.5, Hires steps: 27, Hires upscaler: R-ESRGAN 4x+, TI hashes: “easynegative: c74b4e810b03”, Script: X/Y/Z plot, X Type: Prompt S/R, X Values: “1girl, walk around town, wearing a swimwear”, Version: v1.7.0