<h1 id="krea-2-version:"><span style="color:rgb(21, 170, 191)">Krea 2 Version: </span></h1><h3 id="https:www.patreon.comtensorvizion"><a target="_blank" rel="ugc" href="https://www.patreon.com/TensorVizion/posts/electricamber-2-165792554?source=storefront">https://www.patreon.com/tensorvizion</a></h3><h2 id="what-it-does-well:"><span style="color:rgb(121, 80, 242)"><strong>What it does well:</strong></span></h2><p>. Blonde and brunette hair both get that warm amber glow without blowing out</p><p>. Glasses (round, cat-eye, aviators, thick frames, thin wire — it handles them)</p><p>. Sunglasses too, especially the reflective kind</p><p>. Skin tones stay warm and natural, not orange (I spent way too long fixing that)</p><p>. Backgrounds pick up subtle neon/electric tones without stealing focus</p><h2 id="what-to-know-going-in:"><span style="color:rgb(130, 201, 30)"><strong>What to know going in:</strong></span></h2><p>It's not a full style override. It's more like a mood layer. Your base prompt still matters a lot. If you push the weight too high the lighting can get a little much. Like, "standing inside the sun" much. You've been warned. Works best with portraits and upper-body shots. Full body is fine but the effect is most obvious up close.</p><h3 id="trigger-word:-shades">Trigger Word: shades</h3><h3 id="base-models:-z-image-turbo"><span style="color:rgb(253, 126, 20)"><strong>Base models: Z Image Turbo</strong></span></h3><p>Weight: 0.7 – 0.9 (0.8 is my sweet spot honestly)</p><p>Steps: 28–35</p><p>CFG: 5–7</p><p>Sampler: euler</p><h3 id="base-model:-flux-1-dev-fp8"><span style="color:rgb(21, 170, 191)">Base Model: Flux 1 Dev FP8</span></h3><p>Weight: 0.4-0.6</p><p>Steps: 22-24</p><p>CFG: 5</p><p>Samplers: DPM 2 ++ Karras</p><h3 id="base-model:-qwen-2512"><span style="color:rgb(190, 75, 219)">Base Model: Qwen 2512</span></h3><p>Weight: 0.7-1.0</p><p>Steps: 24</p><p>CFG: 8</p><p>Samplers: DPM 2</p><p></p><h2 id="premium-models-and-workflows"><span style="color:rgb(250, 82, 82)">Premium Models &amp; Workflows</span></h2><h3 id="kofi:-https:www.kofi.comtensorvizion">KoFI: <a target="_blank" rel="ugc" href="https://www.kofi.com/tensorvizion">https://www.kofi.com/tensorvizion</a></h3><p></p>

ElectricAmber

<p><strong><em>Trained on Z-Image-Turbo. </em></strong></p>

ElectricAmber-ZiT_c1-st3000.safetensors

Flux.2 Klein 9B

Qwen 2512

Flux 1 Dev


# 

## Model Introduction

This model is trained using the [Civision/model training](https://modelscope.ai/civision/modelTraining) environment and computing resources provided by ModelScope.

* Model Type: LoRA
* Base Model: [Tongyi-MAI/Z-Image-Turbo](https://modelscope.ai/models/Tongyi-MAI/Z-Image-Turbo)
* Training Code: [DiffSynth-Studio](https://github.com/modelscope/DiffSynth-Studio)
* Data amount：20
* Total training steps：4000
* License: Apache-2.0

## Inference Code

Install [DiffSynth-Studio](https://github.com/modelscope/DiffSynth-Studio):

```bash
pip install diffsynth
```

Inference:

```python
from diffsynth.pipelines.z_image import ZImagePipeline, ModelConfig
import torch

pipe = ZImagePipeline.from_pretrained(
    torch_dtype=torch.bfloat16,
    device="cuda",
    model_configs=[
        ModelConfig(model_id="Tongyi-MAI/Z-Image-Turbo", origin_file_pattern="transformer/*.safetensors"),
        ModelConfig(model_id="Tongyi-MAI/Z-Image-Turbo", origin_file_pattern="text_encoder/*.safetensors"),
        ModelConfig(model_id="Tongyi-MAI/Z-Image-Turbo", origin_file_pattern="vae/diffusion_pytorch_model.safetensors"),
    ],
    tokenizer_config=ModelConfig(model_id="Tongyi-MAI/Z-Image-Turbo", origin_file_pattern="tokenizer/"),
)
pipe.load_lora(pipe.dit, ModelConfig(model_id="tensorvizion/ElectricAmber-ZiT", origin_file_pattern="ElectricAmber-ZiT_c1-st4000.safetensors"))
prompt = "a cat"
image = pipe(prompt=prompt, num_inference_steps=8, cfg_scale=1)
image.save("image.jpg")
```


Files with SHA256: 7a9dc0639deda063be710eb41583d76ab5f81e2ee2932c8b0a83c0f8a50b09f7