pipeLoader v1 (Legacy)¶
Documentation¶
- Class name:
ttN pipeLoader - Category:
🌏 tinyterra/legacy - Output node:
False
The ttN pipeLoader node is designed for loading and initializing data pipelines in the context of the tinyterraNodes framework. It plays a crucial role in setting up the necessary environment and parameters for data processing and transformation tasks, facilitating seamless integration and execution of subsequent nodes in the pipeline.
Input types¶
Required¶
ckpt_name- Specifies the checkpoint name for loading model weights, crucial for initializing the model with pre-trained parameters.
- Comfy dtype:
COMBO[STRING] - Python dtype:
str
config_name- Defines the configuration name for setting up the model, affecting its behavior and parameters.
- Comfy dtype:
COMBO[STRING] - Python dtype:
str
vae_name- Identifies the VAE model name for loading, essential for specific data encoding and decoding tasks.
- Comfy dtype:
COMBO[STRING] - Python dtype:
str
clip_skip- Determines the number of CLIP layers to skip, affecting the depth of semantic analysis.
- Comfy dtype:
INT - Python dtype:
int
lora1_name- Specifies the first LoRA model to be applied, influencing model adaptation and performance.
- Comfy dtype:
COMBO[STRING] - Python dtype:
str
lora1_model_strength- Sets the strength of the first LoRA model's adaptation, affecting the model's output.
- Comfy dtype:
FLOAT - Python dtype:
float
lora1_clip_strength- Determines the CLIP strength for the first LoRA model, influencing semantic alignment.
- Comfy dtype:
FLOAT - Python dtype:
float
lora2_name- Specifies the second LoRA model to be applied, influencing model adaptation and performance.
- Comfy dtype:
COMBO[STRING] - Python dtype:
str
lora2_model_strength- Sets the strength of the second LoRA model's adaptation, affecting the model's output.
- Comfy dtype:
FLOAT - Python dtype:
float
lora2_clip_strength- Determines the CLIP strength for the second LoRA model, influencing semantic alignment.
- Comfy dtype:
FLOAT - Python dtype:
float
lora3_name- Specifies the third LoRA model to be applied, influencing model adaptation and performance.
- Comfy dtype:
COMBO[STRING] - Python dtype:
str
lora3_model_strength- Sets the strength of the third LoRA model's adaptation, affecting the model's output.
- Comfy dtype:
FLOAT - Python dtype:
float
lora3_clip_strength- Determines the CLIP strength for the third LoRA model, influencing semantic alignment.
- Comfy dtype:
FLOAT - Python dtype:
float
positive- Defines the positive conditioning for the pipeline, influencing the direction of data processing.
- Comfy dtype:
STRING - Python dtype:
str
positive_token_normalization- Adjusts the normalization of positive tokens, affecting the conditioning's impact on the model.
- Comfy dtype:
COMBO[STRING] - Python dtype:
str
positive_weight_interpretation- Defines how the weights of positive tokens are interpreted, influencing the model's focus.
- Comfy dtype:
COMBO[STRING] - Python dtype:
str
negative- Specifies the negative conditioning for the pipeline, used to guide the data processing away from certain outcomes.
- Comfy dtype:
STRING - Python dtype:
str
negative_token_normalization- Adjusts the normalization of negative tokens, affecting the conditioning's impact on the model.
- Comfy dtype:
COMBO[STRING] - Python dtype:
str
negative_weight_interpretation- Defines how the weights of negative tokens are interpreted, influencing the model's focus.
- Comfy dtype:
COMBO[STRING] - Python dtype:
str
empty_latent_width- Sets the width of the empty latent space, crucial for defining the initial state of data transformation.
- Comfy dtype:
INT - Python dtype:
int
empty_latent_height- Sets the height of the empty latent space, crucial for defining the initial state of data transformation.
- Comfy dtype:
INT - Python dtype:
int
batch_size- Specifies the batch size for processing, affecting the throughput and efficiency of the pipeline.
- Comfy dtype:
INT - Python dtype:
int
seed- Sets the seed for random number generation, ensuring reproducibility of the pipeline's operations.
- Comfy dtype:
INT - Python dtype:
int
Optional¶
model_override- Allows for overriding the default model, enabling custom model usage within the pipeline.
- Comfy dtype:
MODEL - Python dtype:
str
clip_override- Allows for overriding the default CLIP model, enabling custom semantic analysis.
- Comfy dtype:
CLIP - Python dtype:
str
optional_lora_stack- Enables the use of an optional LoRA stack, allowing for enhanced model adaptation.
- Comfy dtype:
LORA_STACK - Python dtype:
str
Output types¶
pipe- Comfy dtype:
PIPE_LINE - Represents the initialized pipeline, ready for data processing and transformation tasks.
- Python dtype:
dict
- Comfy dtype:
model- Comfy dtype:
MODEL - Outputs the model used within the pipeline after initialization.
- Python dtype:
str
- Comfy dtype:
positive- Comfy dtype:
CONDITIONING - Outputs the positive conditioning applied within the pipeline.
- Python dtype:
str
- Comfy dtype:
negative- Comfy dtype:
CONDITIONING - Outputs the negative conditioning applied within the pipeline.
- Python dtype:
str
- Comfy dtype:
latent- Comfy dtype:
LATENT - Outputs the latent space configuration utilized within the pipeline.
- Python dtype:
int
- Comfy dtype:
vae- Comfy dtype:
VAE - Outputs the VAE model integrated into the pipeline.
- Python dtype:
str
- Comfy dtype:
clip- Comfy dtype:
CLIP - Outputs the CLIP model integrated into the pipeline.
- Python dtype:
str
- Comfy dtype:
seed- Comfy dtype:
INT - Outputs the seed value used for random number generation within the pipeline.
- Python dtype:
int
- Comfy dtype:
Usage tips¶
- Infra type:
CPU - Common nodes:
Source code¶
class ttN_TSC_pipeLoader:
version = '1.1.2'
@classmethod
def INPUT_TYPES(cls):
return {"required": {
"ckpt_name": (folder_paths.get_filename_list("checkpoints"), ),
"config_name": (["Default",] + folder_paths.get_filename_list("configs"), {"default": "Default"} ),
"vae_name": (["Baked VAE"] + folder_paths.get_filename_list("vae"),),
"clip_skip": ("INT", {"default": -1, "min": -24, "max": 0, "step": 1}),
"lora1_name": (["None"] + folder_paths.get_filename_list("loras"),),
"lora1_model_strength": ("FLOAT", {"default": 1.0, "min": -10.0, "max": 10.0, "step": 0.01}),
"lora1_clip_strength": ("FLOAT", {"default": 1.0, "min": -10.0, "max": 10.0, "step": 0.01}),
"lora2_name": (["None"] + folder_paths.get_filename_list("loras"),),
"lora2_model_strength": ("FLOAT", {"default": 1.0, "min": -10.0, "max": 10.0, "step": 0.01}),
"lora2_clip_strength": ("FLOAT", {"default": 1.0, "min": -10.0, "max": 10.0, "step": 0.01}),
"lora3_name": (["None"] + folder_paths.get_filename_list("loras"),),
"lora3_model_strength": ("FLOAT", {"default": 1.0, "min": -10.0, "max": 10.0, "step": 0.01}),
"lora3_clip_strength": ("FLOAT", {"default": 1.0, "min": -10.0, "max": 10.0, "step": 0.01}),
"positive": ("STRING", {"default": "Positive","multiline": True}),
"positive_token_normalization": (["none", "mean", "length", "length+mean"],),
"positive_weight_interpretation": (["comfy", "A1111", "compel", "comfy++", "down_weight"],),
"negative": ("STRING", {"default": "Negative", "multiline": True}),
"negative_token_normalization": (["none", "mean", "length", "length+mean"],),
"negative_weight_interpretation": (["comfy", "A1111", "compel", "comfy++", "down_weight"],),
"empty_latent_width": ("INT", {"default": 512, "min": 64, "max": MAX_RESOLUTION, "step": 8}),
"empty_latent_height": ("INT", {"default": 512, "min": 64, "max": MAX_RESOLUTION, "step": 8}),
"batch_size": ("INT", {"default": 1, "min": 1, "max": 64}),
"seed": ("INT", {"default": 0, "min": 0, "max": 0xffffffffffffffff}),
},
"optional": {"model_override": ("MODEL",), "clip_override": ("CLIP",), "optional_lora_stack": ("LORA_STACK",),},
"hidden": {"prompt": "PROMPT", "ttNnodeVersion": ttN_TSC_pipeLoader.version, "my_unique_id": "UNIQUE_ID",}}
RETURN_TYPES = ("PIPE_LINE" ,"MODEL", "CONDITIONING", "CONDITIONING", "LATENT", "VAE", "CLIP", "INT",)
RETURN_NAMES = ("pipe","model", "positive", "negative", "latent", "vae", "clip", "seed",)
FUNCTION = "adv_pipeloader"
CATEGORY = "🌏 tinyterra/legacy"
def adv_pipeloader(self, ckpt_name, config_name, vae_name, clip_skip,
lora1_name, lora1_model_strength, lora1_clip_strength,
lora2_name, lora2_model_strength, lora2_clip_strength,
lora3_name, lora3_model_strength, lora3_clip_strength,
positive, positive_token_normalization, positive_weight_interpretation,
negative, negative_token_normalization, negative_weight_interpretation,
empty_latent_width, empty_latent_height, batch_size, seed, model_override=None, clip_override=None, optional_lora_stack=None, prompt=None, my_unique_id=None):
model: ModelPatcher | None = None
clip: CLIP | None = None
vae: VAE | None = None
# Create Empty Latent
latent = sampler.emptyLatent(None, batch_size, empty_latent_width, empty_latent_height)
samples = {"samples":latent}
# Load models
model, clip, vae = loader.load_checkpoint(ckpt_name, config_name)
if model_override is not None:
model = model_override
if clip_override is not None:
clip = clip_override
if optional_lora_stack is not None:
for lora in optional_lora_stack:
model, clip = loader.load_lora(lora[0], model, clip, lora[1], lora[2])
if lora1_name != "None":
model, clip = loader.load_lora(lora1_name, model, clip, lora1_model_strength, lora1_clip_strength)
if lora2_name != "None":
model, clip = loader.load_lora(lora2_name, model, clip, lora2_model_strength, lora2_clip_strength)
if lora3_name != "None":
model, clip = loader.load_lora(lora3_name, model, clip, lora3_model_strength, lora3_clip_strength)
# Check for custom VAE
if vae_name != "Baked VAE":
vae = loader.load_vae(vae_name)
# CLIP skip
if not clip:
raise Exception("No CLIP found")
clipped = clip.clone()
if clip_skip != 0:
clipped.clip_layer(clip_skip)
positive = loader.nsp_parse(positive, seed, title='pipeLoader Positive', my_unique_id=my_unique_id)
positive_embeddings_final, positive_pooled = advanced_encode(clipped, positive, positive_token_normalization, positive_weight_interpretation, w_max=1.0, apply_to_pooled='enable')
positive_embeddings_final = [[positive_embeddings_final, {"pooled_output": positive_pooled}]]
negative = loader.nsp_parse(negative, seed, title='pipeLoader Negative', my_unique_id=my_unique_id)
negative_embeddings_final, negative_pooled = advanced_encode(clipped, negative, negative_token_normalization, negative_weight_interpretation, w_max=1.0, apply_to_pooled='enable')
negative_embeddings_final = [[negative_embeddings_final, {"pooled_output": negative_pooled}]]
image = ttNsampler.pil2tensor(Image.new('RGB', (1, 1), (0, 0, 0)))
pipe = {"model": model,
"positive": positive_embeddings_final,
"negative": negative_embeddings_final,
"vae": vae,
"clip": clip,
"samples": samples,
"images": image,
"seed": seed,
"loader_settings": {"ckpt_name": ckpt_name,
"vae_name": vae_name,
"lora1_name": lora1_name,
"lora1_model_strength": lora1_model_strength,
"lora1_clip_strength": lora1_clip_strength,
"lora2_name": lora2_name,
"lora2_model_strength": lora2_model_strength,
"lora2_clip_strength": lora2_clip_strength,
"lora3_name": lora3_name,
"lora3_model_strength": lora3_model_strength,
"lora3_clip_strength": lora3_clip_strength,
"refiner_ckpt_name": None,
"refiner_vae_name": None,
"refiner_lora1_name": None,
"refiner_lora1_model_strength": None,
"refiner_lora1_clip_strength": None,
"refiner_lora2_name": None,
"refiner_lora2_model_strength": None,
"refiner_lora2_clip_strength": None,
"clip_skip": clip_skip,
"positive": positive,
"positive_l": None,
"positive_g": None,
"positive_token_normalization": positive_token_normalization,
"positive_weight_interpretation": positive_weight_interpretation,
"positive_balance": None,
"negative": negative,
"negative_l": None,
"negative_g": None,
"negative_token_normalization": negative_token_normalization,
"negative_weight_interpretation": negative_weight_interpretation,
"negative_balance": None,
"empty_latent_width": empty_latent_width,
"empty_latent_height": empty_latent_height,
"batch_size": batch_size,
"seed": seed,
"empty_samples": samples,}
}
return (pipe, model, positive_embeddings_final, negative_embeddings_final, samples, vae, clip, seed)