Welcome, this is my very first homebrewed Lora!
A high-quality LoRA of Maria from the Silent Hill 2 Remake, trained on 66 carefully selected in-game screenshots most self taken. This version focuses on accurate face, hair (with visible pink tips), modest black dress under the red leather jacket, and the signature red butterfly tattoo.
Recommended Usage
Base Model: Illustrious XL (best compatibility)
Trigger Word: maria_sh2v2
Recommended Strength: 1.18 – 1.25 (1.20–1.22 is usually ideal)
Resolution: 1024×1536 or 1216×1536 (then upscale)
Example Prompts
SFW Default:
maria_sh2v2, short straight blonde bob haircut with vibrant pink tips, pale realistic skin, detailed lifelike face, blue eyes, red leather jacket slightly open over modest high-neck tight black dress, black choker, cinematic lighting, silent hill 2 remake styleSeductive / Flirty:
maria_sh2v2, looking at viewer, seductive gaze, slight seductive smile, short straight blonde bob haircut with vibrant pink tips, red leather jacket slightly open over modest high-neck black dress, dark moody atmosphereDark Dance Club Scene:
maria_sh2v2, standing in dark empty dance club, dim neon lighting, glowing neon signs, red leather jacket slightly open over modest high-neck black dress, seductive expressionNSFW:
maria_sh2v2, nude, red leather jacket open, single red butterfly tattoo on chest, seductive pose, moody lightingTips
Use negative prompt: cleavage, deep cleavage, flying butterfly, multiple butterflies, deformed tattoo, miniskirt, low cut dress
Lower strength (1.15–1.18) helps with modesty and reduces uncanny valley
Works very well with 2× upscaling (4x_NMKD-Superscale or 4x-UltraSharp) at 0.32–0.35 denoise
Trained locally on RTX 5090 with OneTrainer. Feedback is welcome! Thank you for checking it out!
### Disclaimer
This LoRA is a fan-made recreation trained for personal, non-commercial use only.
Silent Hill 2 and all related characters are owned by Konami.
This model is not affiliated with, endorsed by, or sponsored by Konami.
Please respect the original creators and use this LoRA responsibly.
Description
V2.0 is what I consider my base version, the first version was so bad I scrapped it. I learned a lot in the process.