v2_dreamink
| Property | Value |
|---|---|
| License | OpenRAIL |
| Model Type | Stable Diffusion Embedding |
| Training Dataset Size | 32 images |
| Vector Size | 4 |
What is v2_dreamink?
v2_dreamink is a specialized embedding model for Stable Diffusion v2 (768) that creates distinctive silhouetted landscapes against vibrant, colorful backgrounds. This model represents a unique fusion of styles, trained on outputs from a merger of Inkpunk Diffusion and Dreamlike Diffusion.
Implementation Details
The model was trained using Automatic1111's Webui with specific parameters including a 0.003 embedding learning rate, batch size of 4, and gradient accumulation rate of 8. Training utilized a dataset of 32 images processed through a model merge of Inkpunk and Dreamlike Diffusion at 0.3 ratio.
- Compatible exclusively with SD v2 models (768)
- Trained for 500 steps with checkpoints every 250 steps
- Uses deterministic latent sampling method
Core Capabilities
- Generates striking silhouetted landscapes
- Excels with DPM++ samplers for sharp lines
- Particularly effective with landscape compositions
- Creates prismatic and colorful background effects
Frequently Asked Questions
Q: What makes this model unique?
This model specializes in creating dramatic silhouetted landscapes with vibrant backgrounds, combining the sci-fi influences of Inkpunk with the artistic qualities of Dreamlike Diffusion.
Q: What are the recommended use cases?
The model performs best with landscape scenes and architectural compositions. It's particularly effective when used with DPM Adaptive samplers and may not be optimal for character portraits or detailed figure work.
