Update the README to align with the latest node capabilities, shifting the description from "granular nodes" to "exhaustive control" over all API parameters. Key changes include documenting new features like live tooltips, metadata recovery (actual_seed), and improved batch processing. The node reference was also updated to clearly distinguish between required and optional parameters and to reflect the latest supported models.
Replace the unified inference node with dedicated nodes for Flux.1/2, Qwen, and Z models to provide model-specific parameters and better UX. Refactor the API client to simplify image model detection using strict prefixes (black-forest-labs/, Qwen/, Tongyi-MAI/) and remove unused caching logic. Update __init__.py to register all new nodes and print a concise load summary.
Remove the separate model selector node and merge its functionality into
the unified image generation node. Update API client comments from Italian
to English and add parameter presets for new models (Z-Image, FLUX.1-Kontext).
Update api_client.py to support new SiliconFlow model features including cfg, output_format, prompt upsampling/enhancement, safety tolerance, image prompts, aspect ratio, and raw mode. Refactor code and docstrings from Italian to English, change image input from a list to a single base64 string, and introduce conditional payload building to only send set parameters.
- Implemented API client for SiliconFlow image generation.
- Created nodes for model selection and image inference.
- Added configuration management for API key.
- Updated README with installation instructions and node descriptions.
- Enhanced .gitignore to exclude sensitive files.
