Model Name: Cherry Fish

Version: v1.0 + MOONLIGHT (baked in stylized VAE)

Model Overview: This custom Stable Diffusion merge model excels at producing highly stylized, semi-realistic visuals with a strong emphasis on vibrant neon lighting and polished character renders. It showcases a unique balance between cartoon-like aesthetics and photorealistic details, making it particularly suited for character-centric artwork and NSFW content. The model demonstrates remarkable consistency in generating human-like characters with flawless skin textures, expressive eyes, and smooth anatomical compositions.

Model Components:

Checkpoints Used:
- "jedpointIL_V2"
- "ntrMIXIllustriousXL_xii"
LoRAs Added:
- "Style CherryMouseStreet | NoobAi V-Pred V1"
- "detaILeReij"
BAKED IN VAE (MOONLIGHT VERSION)
- LUNA XL VAE

Visual Style & Strengths: This model stands out for its remarkable versatility across various themes and environments. It can produce everything from neon-lit, cyberpunk-inspired indoor scenes to bright, well-lit outdoor settings. The model's adaptability allows it to handle both semi-realistic and anime-inspired styles effortlessly, making it suitable for a wide range of artistic projects. Whether generating character-focused compositions, casual outdoor scenarios, or futuristic urban aesthetics, the model consistently delivers polished, visually appealing results with strong attention to lighting and textures.

Prompting Tips for Best Results:

Use Tag-Heavy Prompts:
- This model performs best with detailed, specific prompts using multiple tags to control the output. Break down the elements you want in your image, such as clothing, expressions, environments, and lighting.
Control Lighting and Environments:
- Include detailed tags for lighting and ambiance, such as "soft neon lighting," "ambient glow," or "daylight with shadows" to guide the output effectively.
Adjust Style with Tags:
- To achieve semi-realistic visuals, include tags like "3d," "semi-realistic," and "detailed textures."
- For a more anime-inspired style, use tags like "anime style," "cel shading," and add "3d" to the negative prompt.
Iterate with Precision:
- Refine your prompts by iterating on successful outputs. Adjust tags incrementally to achieve your desired look.
Utilize Embeddings:
- Your favorite pony quality embeddings will most likely still work well with this model. If you'd like to use the same embeddings I do, check out Pony PDXL Negative Embeddings.

MY POS PROMPT:

masterpiece, best quality, amazing quality, high-resolution at the beginning of the prompt and then i like to use it again at the end.

MY NEG PROMPT:

lowres, worst quality, bad quality, bad anatomy, jpeg artifacts, signature, watermark, old, oldest, censored, bar censor, as well as adding in anything else you don't want to see that my be popping up during use

Technical Notes:

Recommended Settings:
- Resolution: 768x1344
- Sampler: Euler a
- Steps: 30
- CFG Scale: 5.5-7
Strengths:
- Highly detailed character generation
- Excellent lighting and color handling
- Robust NSFW and SFW content creation
Weaknesses:
- May default to certain themes (e.g., neon-lit settings)
- Occasional over-polishing can make images look too smooth or artificial

Search AI Models & Addons

very few results