| Text-to-image generation up to 1792x1024 via V6 and V6.1 models | Yes | No |
| --sref flag for style reference images — pin a visual style across multiple generations | Yes | No |
| Niji mode (niji 6) for anime and illustration aesthetics with dedicated weighting | Yes | No |
| Remix mode lets you edit a generated image with a new prompt without starting over | Yes | No |
| --cref character reference flag to keep a specific person or character consistent across images | Yes | No |
| Vary (Region) for inpainting — repaint specific areas of an image without touching the rest | Yes | No |
| SDXL, SD 3.5, and community checkpoints via ComfyUI or Automatic1111 interfaces | No | Yes |
| LoRA fine-tuning — load character or style LoRAs on top of any base model with a few grams of VRAM | No | Yes |
| ControlNet for pose, depth, and edge-guided generation — output follows a skeleton or sketch exactly | No | Yes |
| img2img and inpainting built into every major UI — redraw any region with a mask | No | Yes |
| No content policy enforcement when running locally — the model does what the prompt says | No | Yes |
| ComfyUI node-based workflow editor for chaining models, ControlNets, upscalers, and custom scripts | No | Yes |