Mattia Moretti
c58a16a401
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.
27 lines
1.5 KiB
Python
27 lines
1.5 KiB
Python
from .base import SiliconFlowBaseNode
|
|
from .api_client import fetch_image_models
|
|
|
|
class SiliconFlowFlux1Dev(SiliconFlowBaseNode):
|
|
@classmethod
|
|
def INPUT_TYPES(cls):
|
|
models = [m for m in fetch_image_models() if "flux.1-dev" in m.lower() and "kontext" not in m.lower()]
|
|
if not models: models = ["black-forest-labs/FLUX.1-dev"]
|
|
return {
|
|
"required": {
|
|
"model": (models, {"default": models[0], "tooltip": "FLUX.1 Dev model ID."}),
|
|
"prompt": ("STRING", {"multiline": True, "tooltip": "Description."}),
|
|
"image_size": (["1024x1024", "960x1280", "768x1024", "720x1440", "720x1280", "others"], {"default": "1024x1024", "tooltip": "Size presets (max 2.3M pixels)."}),
|
|
"num_inference_steps": ("INT", {"default": 20, "min": 1, "max": 30, "tooltip": "Inference steps (1-30)."}),
|
|
"seed": ("INT", {"default": 0, "min": 0, "max": 9999999999, "tooltip": "Random seed."}),
|
|
},
|
|
"optional": {
|
|
"prompt_enhancement": ("BOOLEAN", {"default": False, "tooltip": "Detail-optimized prompt."}),
|
|
}
|
|
}
|
|
|
|
def generate(self, model, prompt, image_size, num_inference_steps, seed, prompt_enhancement=False):
|
|
return self._generate_common(model, prompt, image_size=image_size, num_inference_steps=num_inference_steps, seed=seed, prompt_enhancement=prompt_enhancement)
|
|
|
|
NODE_CLASS_MAPPINGS = {"SiliconFlowFlux1Dev": SiliconFlowFlux1Dev}
|
|
NODE_DISPLAY_NAME_MAPPINGS = {"SiliconFlowFlux1Dev": "🎨 SiliconFlow — FLUX.1 Dev"}
|