Trained on a versatile dataset featuring both classic off-road lifestyle and futuristic cyberpunk aesthetics. Designed to deliver crisp details, realistic vehicle proportions, and rich environment integration.Recommended Weight: <code>0.7 – 1.0</code>IMPORTANT: MOSS Image can sometimes behave unpredictably even with LoRA enabled, so slight deviations from the intended behavior are quite possible.Prompt Tips:<ul><li>Use <code>niva legend</code> as the main trigger tag.</li><li>Combine with descriptive environmental tags (e.g., rough gravel, forest river, neon-lit cyberpunk megacity) to get the desired style.</li><li>niva legend, glossy dark green paint, front three-quarter view, parked on rough gravel near a calm forest river at golden hour sunset, warm orange sunlight flare, long dramatic shadows, highly detailed glossy car body reflections, natural textures, 35mm film photography, cinematic lighting, photorealistic, 8k</li><li>A cinematic, hyper-detailed shot of a heavily modified armored Niva Legend in a neon-lit cyberpunk megacity at night, aggressive off-road brush guards, custom lifted suspension, vibrant electric cyan and hot pink neon accents, glowing headlights piercing through thick atmospheric fog, slick wet asphalt reflecting bright cyberpunk signs, rich saturated color grading, photorealistic textures, dramatic action movie framing, 8k resolution</li></ul>

Niva Legend (VAZ-2121)

NIVAV1M.safetensors

v1.0 MOSS

v1.0 Krea 2


# 

## 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: [openmoss/MOSS-Image-SFT](https://modelscope.ai/models/openmoss/MOSS-Image-SFT)
* Training Code: [DiffSynth-Studio](https://github.com/modelscope/DiffSynth-Studio)
* Data amount：42
* Total training steps：2000
* License: Apache-2.0

## Inference Code

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

```bash
pip install diffsynth
```

Inference:

```python
from diffsynth.pipelines.moss_image import MOSSImagePipeline, ModelConfig
import torch

pipe = MOSSImagePipeline.from_pretrained(
    torch_dtype=torch.bfloat16,
    device="cuda",
    model_configs=[
        ModelConfig(model_id="openmoss/MOSS-Image-SFT", origin_file_pattern="transformer/diffusion_pytorch_model*.safetensors"),
        ModelConfig(model_id="openmoss/MOSS-Image-SFT", origin_file_pattern="text_encoder/model*.safetensors"),
        ModelConfig(model_id="openmoss/MOSS-Image-SFT", origin_file_pattern="vae/diffusion_pytorch_model.safetensors", computation_dtype=torch.float32),
    ],
    tokenizer_config=ModelConfig(model_id="openmoss/MOSS-Image-SFT", origin_file_pattern="tokenizer/"),
)
pipe.load_lora(pipe.dit, ModelConfig(model_id="bazya75/NIVAV1M", origin_file_pattern="NIVAV1M_c1-st2000.safetensors"))

prompt = "a cat"
image = pipe(prompt, height=1024, width=1024, seed=0, num_inference_steps=50, cfg_scale=3.0)
image.save("image.jpg")
```


Files with SHA256: 4da6865fcdcc60417bf2269734554cb055e5851226999c68bcda9313554751d6