after - CivArchive (CivitAI Archive)

after - v1.0

NSFW

### AFTER - Ultra-Realistic Photographic Fine-Tune (SD 1.5)

AFTER is a high-fidelity, hyper-realistic photographic checkpoint developed as an advanced Machine Learning experiment. The core objective of this project was to break the native square constraints (512x512) of the Stable Diffusion 1.5 architecture, achieving superior vertical composition and extreme raw texture density straight out of the first pass.

---

### 📊 Technical Specifications & Training Pipeline

* Aesthetic Profile: Ultra-Realistic Photography & Fine Raw Skin Texture

* Base Model: Candy (SD 1.5 framework, natively trained at 640x960)

* Native Fine-Tuning Resolution: 768x960 pixels (4:5 Aspect Ratio)

* Dataset Source Resolution: 1024x1024 pixels (High-density 1:1 raw photographs)

* Dataset Volume: 680 heavily curated, ultra-high-resolution portrait and fashion images

* Training Duration: 20 Complete Epochs

* Hardware Setup: Trained locally on an NVIDIA GeForce RTX 4080 Super (16GB GDDR6X)

---

### ⚡ Multi-Resolution Native Generation (Zero Hires. Fix)

Thanks to the structural inheritance from the Candy model combined with a high-density 1024x1024 dataset, the U-Net can scale cleanly across multiple aspect ratios. You can safely generate images in 4 different resolutions on the first pass without using Hires. Fix or experiencing body/head duplication artifacts:

* 512x640 (Fast rendering / Compact portrait)

* 640x960 (Direct Candy aspect ratio compatibility)

* 768x960 (Native fine-tuning balance)

* 768x1024 (Maximum vertical extrapolation)

Recommended Settings: For optimal results, keep your CFG Scale low (between 4.0 and 5.0). The model converges perfectly at CFG 4.5 using samplers like Euler Max.

---

### 🚀 Core Strengths (ML Experiment Successes)

* True Photographic Texture: By compressing 1024px raw dataset elements into rectangular latent dimensions, the model achieves a "Texture Squish" effect, delivering realistic skin pores, fine hair strands, and authentic fabric weaves (like sheer lace, satin, and heavy denim) that rival modern heavy models (SDXL/Flux).

* Impeccable Vertical Geometry: Complete elimination of standard SD 1.5 torso/head duplication vices in tall formats up to 1024px, maintaining perfect golden-ratio body proportions.

* Advanced Optical Dynamics: Flawless rendering of complex lighting scenarios, including soft volumetric blind/shutter shadows, high-contrast rim lighting, and natural camera bokeh mimicry.

* High Inference Efficiency: Massive VRAM and latency savings, allowing huge batch generations directly in txt2img.

---

### ⚠️ Known Limitations & Future Roadmap

* Amorphous Hands: Because this initial phase of the ML experiment heavily prioritized macro body proportions, spatial balance, and global lighting behavior, the U-Net can still produce structural inconsistencies on fine extremities (fingers and hand anatomy) in certain seeds.

* Next Steps: This model is an ongoing R&D project. The current version completely validates our vertical resolution and texture scaling theories. The next iteration will feature an expanded dataset with aggressive anatomical reinforcement tags to isolate and fix hand fidelity.

⚠️ CONTENT WARNING (NSFW):

This merge is uncensored and capable of generating high-quality explicit NSFW content and nudity. It has a tendency towards revealing clothing in casual settings.

For SFW results: Strong negative prompts are highly recommended (e.g., nude, nipples, explicit, nsfw).

⚠️ LICENSE & PERMISSIONS (READ BEFORE DOWNLOADING)

1. PERSONAL USE ONLY This model is provided free of charge for Personal, Non-Profit, and Research use only. You may use it to create images for your personal portfolio.

2. STRICTLY NO REDISTRIBUTION

❌ DO NOT re-upload this file to Civitai, Hugging Face, or any other platform.
❌ DO NOT host this model on third-party generation services (e.g., Tensor.art, Mage.space, Telegram Bots).

3. COMMERCIAL RESTRICTIONS Using this model or its outputs for commercial revenue (Influencers, Ads, Stock Photos) without a license is PROHIBITED.

💼 COMMERCIAL SERVICES & COMMISSIONS

I do not sell the model file for commercial use. Instead, I offer premium AI solutions for brands and agencies:

✨ Exclusive AI Influencers: I create and manage consistent digital personas for Instagram/Social Media.
🏢 Corporate B2B LoRAs: Custom training for brand identity and mascots.
📸 High-End Image Packs: Monthly content packages for your brand.

To hire me for professional AI Modeling services: 📩 Contact: [[email protected]]

I recommend using the Adetailer extension.

Use this extension to fix hand errors:

https://github.com/licyk/advanced_euler_sampler_extension

Use these recommended settings for generation:

Sampling method: Euler_Max

Sampling steps: 30-50

CFG Scale: 2.0 - 7.0

Skip clip: 1-2

Sampling method: DPM++ 2M

Sampling steps: 18-30

CFG Scale: 2.0 - 7.0