diff options
Diffstat (limited to 'modules/sd_samplers_kdiffusion.py')
| -rw-r--r-- | modules/sd_samplers_kdiffusion.py | 53 |
1 files changed, 43 insertions, 10 deletions
diff --git a/modules/sd_samplers_kdiffusion.py b/modules/sd_samplers_kdiffusion.py index aa7f106b..f076fc55 100644 --- a/modules/sd_samplers_kdiffusion.py +++ b/modules/sd_samplers_kdiffusion.py @@ -1,6 +1,7 @@ from collections import deque
import torch
import inspect
+import einops
import k_diffusion.sampling
from modules import prompt_parser, devices, sd_samplers_common
@@ -56,6 +57,7 @@ class CFGDenoiser(torch.nn.Module): self.nmask = None
self.init_latent = None
self.step = 0
+ self.image_cfg_scale = None
def combine_denoised(self, x_out, conds_list, uncond, cond_scale):
denoised_uncond = x_out[-uncond.shape[0]:]
@@ -67,19 +69,36 @@ class CFGDenoiser(torch.nn.Module): return denoised
+ def combine_denoised_for_edit_model(self, x_out, cond_scale):
+ out_cond, out_img_cond, out_uncond = x_out.chunk(3)
+ denoised = out_uncond + cond_scale * (out_cond - out_img_cond) + self.image_cfg_scale * (out_img_cond - out_uncond)
+
+ return denoised
+
def forward(self, x, sigma, uncond, cond, cond_scale, image_cond):
if state.interrupted or state.skipped:
raise sd_samplers_common.InterruptedException
+ # at self.image_cfg_scale == 1.0 produced results for edit model are the same as with normal sampling,
+ # so is_edit_model is set to False to support AND composition.
+ is_edit_model = shared.sd_model.cond_stage_key == "edit" and self.image_cfg_scale is not None and self.image_cfg_scale != 1.0
+
conds_list, tensor = prompt_parser.reconstruct_multicond_batch(cond, self.step)
uncond = prompt_parser.reconstruct_cond_batch(uncond, self.step)
+ assert not is_edit_model or all([len(conds) == 1 for conds in conds_list]), "AND is not supported for InstructPix2Pix checkpoint (unless using Image CFG scale = 1.0)"
+
batch_size = len(conds_list)
repeats = [len(conds_list[i]) for i in range(batch_size)]
- x_in = torch.cat([torch.stack([x[i] for _ in range(n)]) for i, n in enumerate(repeats)] + [x])
- image_cond_in = torch.cat([torch.stack([image_cond[i] for _ in range(n)]) for i, n in enumerate(repeats)] + [image_cond])
- sigma_in = torch.cat([torch.stack([sigma[i] for _ in range(n)]) for i, n in enumerate(repeats)] + [sigma])
+ if not is_edit_model:
+ x_in = torch.cat([torch.stack([x[i] for _ in range(n)]) for i, n in enumerate(repeats)] + [x])
+ sigma_in = torch.cat([torch.stack([sigma[i] for _ in range(n)]) for i, n in enumerate(repeats)] + [sigma])
+ image_cond_in = torch.cat([torch.stack([image_cond[i] for _ in range(n)]) for i, n in enumerate(repeats)] + [image_cond])
+ else:
+ x_in = torch.cat([torch.stack([x[i] for _ in range(n)]) for i, n in enumerate(repeats)] + [x] + [x])
+ sigma_in = torch.cat([torch.stack([sigma[i] for _ in range(n)]) for i, n in enumerate(repeats)] + [sigma] + [sigma])
+ image_cond_in = torch.cat([torch.stack([image_cond[i] for _ in range(n)]) for i, n in enumerate(repeats)] + [image_cond] + [torch.zeros_like(self.init_latent)])
denoiser_params = CFGDenoiserParams(x_in, image_cond_in, sigma_in, state.sampling_step, state.sampling_steps)
cfg_denoiser_callback(denoiser_params)
@@ -88,7 +107,10 @@ class CFGDenoiser(torch.nn.Module): sigma_in = denoiser_params.sigma
if tensor.shape[1] == uncond.shape[1]:
- cond_in = torch.cat([tensor, uncond])
+ if not is_edit_model:
+ cond_in = torch.cat([tensor, uncond])
+ else:
+ cond_in = torch.cat([tensor, uncond, uncond])
if shared.batch_cond_uncond:
x_out = self.inner_model(x_in, sigma_in, cond={"c_crossattn": [cond_in], "c_concat": [image_cond_in]})
@@ -104,7 +126,13 @@ class CFGDenoiser(torch.nn.Module): for batch_offset in range(0, tensor.shape[0], batch_size):
a = batch_offset
b = min(a + batch_size, tensor.shape[0])
- x_out[a:b] = self.inner_model(x_in[a:b], sigma_in[a:b], cond={"c_crossattn": [tensor[a:b]], "c_concat": [image_cond_in[a:b]]})
+
+ if not is_edit_model:
+ c_crossattn = [tensor[a:b]]
+ else:
+ c_crossattn = torch.cat([tensor[a:b]], uncond)
+
+ x_out[a:b] = self.inner_model(x_in[a:b], sigma_in[a:b], cond={"c_crossattn": c_crossattn, "c_concat": [image_cond_in[a:b]]})
x_out[-uncond.shape[0]:] = self.inner_model(x_in[-uncond.shape[0]:], sigma_in[-uncond.shape[0]:], cond={"c_crossattn": [uncond], "c_concat": [image_cond_in[-uncond.shape[0]:]]})
@@ -115,7 +143,10 @@ class CFGDenoiser(torch.nn.Module): elif opts.live_preview_content == "Negative prompt":
sd_samplers_common.store_latent(x_out[-uncond.shape[0]:])
- denoised = self.combine_denoised(x_out, conds_list, uncond, cond_scale)
+ if not is_edit_model:
+ denoised = self.combine_denoised(x_out, conds_list, uncond, cond_scale)
+ else:
+ denoised = self.combine_denoised_for_edit_model(x_out, cond_scale)
if self.mask is not None:
denoised = self.init_latent * self.mask + self.nmask * denoised
@@ -198,6 +229,7 @@ class KDiffusionSampler: self.model_wrap_cfg.mask = p.mask if hasattr(p, 'mask') else None
self.model_wrap_cfg.nmask = p.nmask if hasattr(p, 'nmask') else None
self.model_wrap_cfg.step = 0
+ self.model_wrap_cfg.image_cfg_scale = getattr(p, 'image_cfg_scale', None)
self.eta = p.eta if p.eta is not None else opts.eta_ancestral
k_diffusion.sampling.torch = TorchHijack(self.sampler_noises if self.sampler_noises is not None else [])
@@ -260,13 +292,14 @@ class KDiffusionSampler: self.model_wrap_cfg.init_latent = x
self.last_latent = x
-
- samples = self.launch_sampling(t_enc + 1, lambda: self.func(self.model_wrap_cfg, xi, extra_args={
+ extra_args={
'cond': conditioning,
'image_cond': image_conditioning,
'uncond': unconditional_conditioning,
- 'cond_scale': p.cfg_scale
- }, disable=False, callback=self.callback_state, **extra_params_kwargs))
+ 'cond_scale': p.cfg_scale,
+ }
+
+ samples = self.launch_sampling(t_enc + 1, lambda: self.func(self.model_wrap_cfg, xi, extra_args=extra_args, disable=False, callback=self.callback_state, **extra_params_kwargs))
return samples
|