qnloft
/
ComfyUI
mirror of https://github.com/comfyanonymous/ComfyUI.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Some optimizations to euler a.

This commit is contained in:
comfyanonymous 2024-12-03 06:11:52 -05:00
parent cdc3b97dd5
commit c1b92b719d
1 changed files with 21 additions and 16 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

25
comfy/k_diffusion/sampling.py
View File

@ -175,12 +175,14 @@ def sample_euler_ancestral(model, x, sigmas, extra_args=None, callback=None, dis
sigma_down, sigma_up = get_ancestral_step(sigmas[i], sigmas[i + 1], eta=eta)
if callback is not None:
callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigmas[i], 'denoised': denoised})
if sigma_down == 0:
x = denoised
else:
d = to_d(x, sigmas[i], denoised)
# Euler method
dt = sigma_down - sigmas[i]
x = x + d * dt
if sigmas[i + 1] > 0:
x = x + noise_sampler(sigmas[i], sigmas[i + 1]) * s_noise * sigma_up
x = x + d * dt + noise_sampler(sigmas[i], sigmas[i + 1]) * s_noise * sigma_up
return x
@torch.no_grad()
@ -192,19 +194,22 @@ def sample_euler_ancestral_RF(model, x, sigmas, extra_args=None, callback=None,
for i in trange(len(sigmas) - 1, disable=disable):
denoised = model(x, sigmas[i] * s_in, **extra_args)
# sigma_down, sigma_up = get_ancestral_step(sigmas[i], sigmas[i + 1], eta=eta)
downstep_ratio = 1 + (sigmas[i+1]/sigmas[i] - 1) * eta
sigma_down = sigmas[i+1] * downstep_ratio
alpha_ip1 = 1 - sigmas[i+1]
alpha_down = 1 - sigma_down
renoise_coeff = (sigmas[i+1]**2 - sigma_down**2*alpha_ip1**2/alpha_down**2)**0.5
if callback is not None:
callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigmas[i], 'denoised': denoised})
if sigmas[i + 1] == 0:
x = denoised
else:
downstep_ratio = 1 + (sigmas[i + 1] / sigmas[i] - 1) * eta
sigma_down = sigmas[i + 1] * downstep_ratio
alpha_ip1 = 1 - sigmas[i + 1]
alpha_down = 1 - sigma_down
renoise_coeff = (sigmas[i + 1]**2 - sigma_down**2 * alpha_ip1**2 / alpha_down**2)**0.5
# Euler method
sigma_down_i_ratio = sigma_down / sigmas[i]
x = sigma_down_i_ratio * x + (1 - sigma_down_i_ratio) * denoised
if sigmas[i + 1] > 0 and eta > 0:
x = (alpha_ip1/alpha_down) * x + noise_sampler(sigmas[i], sigmas[i + 1]) * s_noise * renoise_coeff
if eta > 0:
x = (alpha_ip1 / alpha_down) * x + noise_sampler(sigmas[i], sigmas[i + 1]) * s_noise * renoise_coeff
return x
@torch.no_grad()