README.md

---
tags:
- text-to-image
- lora
- diffusers
- template:diffusion-lora
- realism
- portrait
widget:
- text: >-
    b1 portrait, Captured at eye-level, a close-up shot captures a young man
    with curly black hair, wearing a red and navy blue jacket with a white
    stripe down the front. His jacket is adorned with a black collar and a white
    logo with a blue letter "F" in the center. Behind him, a young woman in a
    navy blue and red tank top, also wearing a white hat, is seated on the
    beach. The background is a sandy beach, with a clear blue sky dotted with
    white fluffy clouds.
  output:
    url: images/1.png
- text: >-
    b1 portrait, A close-up shot of a young Asian woman with brown hair and blue
    eyes. She is wearing a purple polo shirt with a collar. Her hair is pulled
    back in a ponytail. Her eyes are blue and her lips are pink. The background
    is blurred, but it appears to be a grassy field. The sky is gray
  output:
    url: images/2.png
- text: >-
    b1 portrait, a young Asian girl with short dark brown hair, wearing a red
    and green sun hat. She is wearing a blue sleeveless shirt with a yellow
    stripe down the center of the shirt. The shirt has a blue collar with a
    black button on it. The girls lips are painted pink. The backdrop is a light
    blue with white daisy prints on it, adding a pop of color to the scene.
  output:
    url: images/3.png
- text: >-
    b1 portrait, a close-up shot captures a young womans face, adorned with a
    green beanie and a beige cardigan. She is wearing a pair of gold rimmed
    glasses with a white label that reads "QDUROY" in bold black letters. Her
    eyes are a piercing blue, her eyebrows are a darker shade of brown, and her
    lips are a lighter pink. Her hair is a dark brown, adding a pop of color to
    her face. The background is a stark white, creating a stark contrast to her
    outfit.
  output:
    url: images/4.png
- text: >-
    b1 portrait, Captured at eye-level, a close-up shot of a young woman with
    long brown hair, wearing a white cap, a white shirt, and a red jacket with
    black buttons. Her hair is styled in two braids, adding a touch of texture
    to her face. Her eyes are a piercing blue, and her eyebrows are a darker
    shade of brown. The background is a light brown, and the womans hair is
    cascading over her shoulders.
  output:
    url: images/5.png
- text: >-
    b1 portrait, a close-up shot of a womans face is angled towards the left
    side of the frame. The womans eyes are painted in a vibrant shade of yellow,
    green, and brown, while her lips are painted with white teeth. She is
    wearing a pair of gold hoop earrings, adding a pop of color to her face. The
    background is a light blue, and the womans hair is pulled back in a
    ponytail.
  output:
    url: images/6.png
base_model:
- prithivMLmods/Flux.1-Merged
instance_prompt: b1 portrait
license: other
license_name: flux-1-dev-non-commercial-license
license_link: https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md
new_version: strangerzonehf/Blaze.1-Portrait-Revised
---
![asfsadfsfsedfsef.png](https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/lV-r_htnneLS7UzasDzjI.png)

<Gallery />

‎

# **Blaze.1 Portrait Image Fidelity `[ Base Flux.1 Merged ]`**

Blaze.1 Portrait is a flux-based adapter designed to generate high-quality, realism-driven super-portraits. This advanced model underwent rigorous training using a diverse set of images to ensure exceptional output in portrait generation. With 72 variations and a total of 1080 images per variation, the training process involved meticulous curation and processing of 14-bit raw Camera Uncompressed Format (CUF) images. For comprehensive insights, refer to the detailed documentation on training procedures, data curation, and parameter optimization.

# **Image Processing Parameters**  

| **Parameter**             | **Value**                               | **Parameter**             | **Value**                    |
|---------------------------|-----------------------------------------|---------------------------|------------------------------|
| LR Scheduler              | Constant                                | Noise Offset              | 0.03                         |
| Optimizer                 | AdamW                                   | Multires Noise Discount   | 0.1                          |
| Network Dim               | 64                                      | Multires Noise Iterations | 10                           |
| Network Alpha             | 32                                      | *Repeat & Steps*        | *<span style="color:orangered">30 & 38,000</span>*|
| *Epoch*                | *<span style="color:orangered">45</span>* | Save Every N Epochs       | 1                            |

The model was trained with an overall image count of 1080 images, repeated across variations to create a total dataset of 1100 repetition per image.

Training was conducted for 4 epochs per set, resulting in a cumulative total of 38,000 steps, with the image labeler *Florence2-en (natural language & English)*.

# **Dataset Details**

The dataset consists of highly realistic images, meticulously curated for optimal model performance. Smaller adapter versions were initially trained on batches, and high-resolution images were later generated from these adapters for further training.

Pre-training upscaling techniques were employed to enhance pixel clarity, ensuring superior image fidelity during model inference.

Image compression details:

- *Minimum compression:* 1.4 MB  
- *Maximum compression:* 10.1 MB  
- *Average compression:* 3.2 MB

# **Training Infrastructure**

Blaze.1 Portrait was trained using state-of-the-art hardware:

- *GPU Configuration:* NVIDIA A100 SXM (2x)
- *vCPU:* 32 cores
- *VRAM:* 160 GB
- *RAM:* 250 GB
- *Disk Space:* 400 GB
- *Network Memory:* High-speed interconnects for efficient data transfer

# **Training Runtime**

Total machine runtime: *<span style="color:orangered">8 hours 14 minutes</span>*

# **Trigger Words**

To generate images using Blaze.1 Portrait, utilize the trigger phrase:

- *<span style="color:orangered">b1 portrait</span>*
  
Incorporating this trigger in your prompts ensures optimal model performance and high-quality portrait generation.

# **Download Model**

Weights for Blaze.1 Portrait are available in Safetensors format. You can download them from the Files & Versions tab using the link below:

[Download Blaze.1 Portrait Weights](/strangerzonehf/Blaze.1-Portrait/tree/main)


```bibtex
@misc{strangerzonehf2024blaze1,
    title={Blaze.1 Portrait: High-Quality Realism-Driven Portrait Generation with Flux-Based LoRA Adapter}, 
    author={StrangerzoneHF},
    year={2024},
    note={public model repository on Hugging Face},
    url={https://huggingface.co/strangerzonehf/Blaze.1-Portrait},
    model_trainer={@prithivMLmods},
    dataset_curator={@prithivMLmods}
```