sd-dynamic-thresholding/dynthres_core.py

167 lines
7.8 KiB
Python

import torch, math
######################### DynThresh Core #########################
class DynThresh:
Modes = ["Constant", "Linear Down", "Cosine Down", "Half Cosine Down", "Linear Up", "Cosine Up", "Half Cosine Up", "Power Up", "Power Down", "Linear Repeating", "Cosine Repeating", "Sawtooth"]
Startpoints = ["MEAN", "ZERO"]
Variabilities = ["AD", "STD"]
def __init__(self, mimic_scale, threshold_percentile, mimic_mode, mimic_scale_min, cfg_mode, cfg_scale_min, sched_val, experiment_mode, maxSteps, separate_feature_channels, scaling_startpoint, variability_measure, interpolate_phi):
self.mimic_scale = mimic_scale
self.threshold_percentile = threshold_percentile
self.mimic_mode = mimic_mode
self.cfg_mode = cfg_mode
self.maxSteps = maxSteps
self.cfg_scale_min = cfg_scale_min
self.mimic_scale_min = mimic_scale_min
self.experiment_mode = experiment_mode
self.sched_val = sched_val
self.sep_feat_channels = separate_feature_channels
self.scaling_startpoint = scaling_startpoint
self.variability_measure = variability_measure
self.interpolate_phi = interpolate_phi
def interpretScale(self, scale, mode, min):
scale -= min
max = self.maxSteps - 1
if mode == "Constant":
pass
elif mode == "Linear Down":
scale *= 1.0 - (self.step / max)
elif mode == "Half Cosine Down":
scale *= math.cos((self.step / max))
elif mode == "Cosine Down":
scale *= math.cos((self.step / max) * 1.5707)
elif mode == "Linear Up":
scale *= self.step / max
elif mode == "Half Cosine Up":
scale *= 1.0 - math.cos((self.step / max))
elif mode == "Cosine Up":
scale *= 1.0 - math.cos((self.step / max) * 1.5707)
elif mode == "Power Up":
scale *= math.pow(self.step / max, self.sched_val)
elif mode == "Power Down":
scale *= 1.0 - math.pow(self.step / max, self.sched_val)
elif mode == "Linear Repeating":
portion = ((self.step / max) * self.sched_val) % 1.0
scale *= (0.5 - portion) * 2 if portion < 0.5 else (portion - 0.5) * 2
elif mode == "Cosine Repeating":
scale *= math.cos((self.step / max) * 6.28318 * self.sched_val) * 0.5 + 0.5
elif mode == "Sawtooth":
scale *= ((self.step / max) * self.sched_val) % 1.0
scale += min
return scale
def dynthresh(self, cond, uncond, cfgScale, weights):
mimicScale = self.interpretScale(self.mimic_scale, self.mimic_mode, self.mimic_scale_min)
cfgScale = self.interpretScale(cfgScale, self.cfg_mode, self.cfg_scale_min)
# uncond shape is (batch, 4, height, width)
conds_per_batch = cond.shape[0] / uncond.shape[0]
assert conds_per_batch == int(conds_per_batch), "Expected # of conds per batch to be constant across batches"
cond_stacked = cond.reshape((-1, int(conds_per_batch)) + uncond.shape[1:])
### Normal first part of the CFG Scale logic, basically
diff = cond_stacked - uncond.unsqueeze(1)
if weights is not None:
diff = diff * weights
relative = diff.sum(1)
### Get the normal result for both mimic and normal scale
mim_target = uncond + relative * mimicScale
cfg_target = uncond + relative * cfgScale
### If we weren't doing mimic scale, we'd just return cfg_target here
### Now recenter the values relative to their average rather than absolute, to allow scaling from average
mim_flattened = mim_target.flatten(2)
cfg_flattened = cfg_target.flatten(2)
mim_means = mim_flattened.mean(dim=2).unsqueeze(2)
cfg_means = cfg_flattened.mean(dim=2).unsqueeze(2)
mim_centered = mim_flattened - mim_means
cfg_centered = cfg_flattened - cfg_means
if self.sep_feat_channels:
if self.variability_measure == 'STD':
mim_scaleref = mim_centered.std(dim=2).unsqueeze(2)
cfg_scaleref = cfg_centered.std(dim=2).unsqueeze(2)
else: # 'AD'
mim_scaleref = mim_centered.abs().max(dim=2).values.unsqueeze(2)
cfg_scaleref = torch.quantile(cfg_centered.abs(), self.threshold_percentile, dim=2).unsqueeze(2)
else:
if self.variability_measure == 'STD':
mim_scaleref = mim_centered.std()
cfg_scaleref = cfg_centered.std()
else: # 'AD'
mim_scaleref = mim_centered.abs().max()
cfg_scaleref = torch.quantile(cfg_centered.abs(), self.threshold_percentile)
if self.scaling_startpoint == 'ZERO':
scaling_factor = mim_scaleref / cfg_scaleref
result = cfg_flattened * scaling_factor
else: # 'MEAN'
if self.variability_measure == 'STD':
cfg_renormalized = (cfg_centered / cfg_scaleref) * mim_scaleref
else: # 'AD'
### Get the maximum value of all datapoints (with an optional threshold percentile on the uncond)
max_scaleref = torch.maximum(mim_scaleref, cfg_scaleref)
### Clamp to the max
cfg_clamped = cfg_centered.clamp(-max_scaleref, max_scaleref)
### Now shrink from the max to normalize and grow to the mimic scale (instead of the CFG scale)
cfg_renormalized = (cfg_clamped / max_scaleref) * mim_scaleref
### Now add it back onto the averages to get into real scale again and return
result = cfg_renormalized + cfg_means
actualRes = result.unflatten(2, mim_target.shape[2:])
if self.interpolate_phi != 1.0:
actualRes = actualRes * self.interpolate_phi + cfg_target * (1.0 - self.interpolate_phi)
if self.experiment_mode == 1:
num = actualRes.cpu().numpy()
for y in range(0, 64):
for x in range (0, 64):
if num[0][0][y][x] > 1.0:
num[0][1][y][x] *= 0.5
if num[0][1][y][x] > 1.0:
num[0][1][y][x] *= 0.5
if num[0][2][y][x] > 1.5:
num[0][2][y][x] *= 0.5
actualRes = torch.from_numpy(num).to(device=uncond.device)
elif self.experiment_mode == 2:
num = actualRes.cpu().numpy()
for y in range(0, 64):
for x in range (0, 64):
overScale = False
for z in range(0, 4):
if abs(num[0][z][y][x]) > 1.5:
overScale = True
if overScale:
for z in range(0, 4):
num[0][z][y][x] *= 0.7
actualRes = torch.from_numpy(num).to(device=uncond.device)
elif self.experiment_mode == 3:
coefs = torch.tensor([
# R G B W
[0.298, 0.207, 0.208, 0.0], # L1
[0.187, 0.286, 0.173, 0.0], # L2
[-0.158, 0.189, 0.264, 0.0], # L3
[-0.184, -0.271, -0.473, 1.0], # L4
], device=uncond.device)
resRGB = torch.einsum("laxy,ab -> lbxy", actualRes, coefs)
maxR, maxG, maxB, maxW = resRGB[0][0].max(), resRGB[0][1].max(), resRGB[0][2].max(), resRGB[0][3].max()
maxRGB = max(maxR, maxG, maxB)
print(f"test max = r={maxR}, g={maxG}, b={maxB}, w={maxW}, rgb={maxRGB}")
if self.step / (self.maxSteps - 1) > 0.2:
if maxRGB < 2.0 and maxW < 3.0:
resRGB /= maxRGB / 2.4
else:
if maxRGB > 2.4 and maxW > 3.0:
resRGB /= maxRGB / 2.4
actualRes = torch.einsum("laxy,ab -> lbxy", resRGB, coefs.inverse())
return actualRes