72 lines
2.9 KiB
Markdown
72 lines
2.9 KiB
Markdown
# Sigma Schedule Interpolated Mean 🎭🅐🅓
|
|
## Documentation
|
|
- Class name: ADE_SigmaScheduleWeightedAverageInterp
|
|
- Category: Animate Diff 🎭🅐🅓/sample settings/sigma schedule
|
|
- Output node: False
|
|
- Repo Ref: https://github.com/Kosinkadink/ComfyUI-AnimateDiff-Evolved.git
|
|
|
|
此节点旨在通过基于指定的权重范围和插值方法在两个给定的sigma计划之间进行插值来创建一个新的sigma计划。它有效地将两个输入计划的特性融合成一个新的计划，从而在扩散过程中实现动态调整。
|
|
|
|
## Input types
|
|
### Required
|
|
- schedule_A
|
|
- 第一个要插值的sigma计划。
|
|
- Comfy dtype: SIGMA_SCHEDULE
|
|
- Python dtype: SigmaSchedule
|
|
- schedule_B
|
|
- 第二个要插值的sigma计划。
|
|
- Comfy dtype: SIGMA_SCHEDULE
|
|
- Python dtype: SigmaSchedule
|
|
- weight_A_Start
|
|
- 插值中第一个sigma计划的起始权重。
|
|
- Comfy dtype: FLOAT
|
|
- Python dtype: float
|
|
- weight_A_End
|
|
- 插值中第一个sigma计划的结束权重。
|
|
- Comfy dtype: FLOAT
|
|
- Python dtype: float
|
|
- interpolation
|
|
- 用于混合sigma计划的插值方法。
|
|
- Comfy dtype: COMBO[STRING]
|
|
- Python dtype: InterpolationMethod
|
|
|
|
## Output types
|
|
- sigma_schedule
|
|
- Comfy dtype: SIGMA_SCHEDULE
|
|
- 插值后的结果sigma计划。
|
|
- Python dtype: SigmaSchedule
|
|
|
|
## Usage tips
|
|
- Infra type: CPU
|
|
- Common nodes: unknown
|
|
|
|
## Source code
|
|
```python
|
|
class InterpolatedWeightedAverageSigmaScheduleNode:
|
|
@classmethod
|
|
def INPUT_TYPES(s):
|
|
return {
|
|
"required": {
|
|
"schedule_A": ("SIGMA_SCHEDULE",),
|
|
"schedule_B": ("SIGMA_SCHEDULE",),
|
|
"weight_A_Start": ("FLOAT", {"default": 0.5, "min": 0.0, "max": 1.0, "step": 0.001}),
|
|
"weight_A_End": ("FLOAT", {"default": 0.5, "min": 0.0, "max": 1.0, "step": 0.001}),
|
|
"interpolation": (InterpolationMethod._LIST,),
|
|
}
|
|
}
|
|
|
|
RETURN_TYPES = ("SIGMA_SCHEDULE",)
|
|
CATEGORY = "Animate Diff 🎭🅐🅓/sample settings/sigma schedule"
|
|
FUNCTION = "get_sigma_schedule"
|
|
|
|
def get_sigma_schedule(self, schedule_A: SigmaSchedule, schedule_B: SigmaSchedule,
|
|
weight_A_Start: float, weight_A_End: float, interpolation: str):
|
|
validate_sigma_schedule_compatibility(schedule_A, schedule_B)
|
|
# get reverse weights, since sigmas are currently reversed
|
|
weights = InterpolationMethod.get_weights(num_from=weight_A_Start, num_to=weight_A_End,
|
|
length=schedule_A.total_sigmas(), method=interpolation, reverse=True)
|
|
weights = weights.to(schedule_A.model_sampling.sigmas.dtype).to(schedule_A.model_sampling.sigmas.device)
|
|
new_sigmas = schedule_A.model_sampling.sigmas * weights + schedule_B.model_sampling.sigmas * (1.0-weights)
|
|
combo_schedule = schedule_A.clone()
|
|
combo_schedule.model_sampling.set_sigmas(new_sigmas)
|
|
return (combo_schedule,) |