Sign In

Cinematic Macro & Identity Core (SDXL LoRA Rank 128)

Download

0 variants available

No files available

Type
LoRA
Stats

33

Reviews
Published

Sep 9, 2026

Base Model

SDXL 1.0

Training
Steps: 3,000
Epochs: 100
Hash
AutoV2
2AC150F66B
Trigger Words
alisaXXXr128
default creator card background decoration
Uploads - 34

34

Followers - 84

84

Likes - 216

216

Orakul_Studio_0029.png

Key Features:

No Quantization & High-Rank Depth: Rank 128 embeds skin texture, micro-pores, water droplets, and natural subsurface scattering (SSS) without the "plastic" blurriness often caused by quantization.

Style Versatility: Thanks to the high flexibility of the weights, the model performs consistently well in both authentic macro-photorealism and 2D/Anime stylizations.

Recommended Weight Guide:

0.07 – 0.10 [2D Anime / Cel-Shading]: A subtle injection. It imparts specific eye and facial feature characteristics while preserving the base model's clean, flat 2D line art, free from unwanted noise.

0.15 – 0.25 [2.5D / Semi-Realism / Concept Art]: A volumetric concept art style reminiscent of AAA games. Deep chiaroscuro and fine details emerge while maintaining a stylized graphic look.

0.70 – 1.00 [Full Photorealism / Macro]: Maximum detail. Ideal for macro portraits, studio lighting, water droplets, pores, and realistic wet-skin physics.

Generation Settings:

Base Model: SDXL 1.0 (or fine-tuned photorealistic bases)

Sampler: dpmpp_2m / euler_ancestral

Steps: 25–35

CFG Scale: 5.0 – 7.0

Framing Note: For full-body shots, using FaceDetailer / HandDetailer is recommended to compensate for SDXL's latent space physics.