Skip to content

The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface. The fastest local inference engine in the world.

Move noise scaling to object with sampling math.

comfyanonymousc62e8361676e
diff --git a/comfy/model_sampling.py b/comfy/model_sampling.py
index e7f8bc6a3..b46202b78 100644
--- a/comfy/model_sampling.py
+++ b/comfy/model_sampling.py
@@ -11,6 +11,14 @@ class EPS:
sigma = sigma.view(sigma.shape[:1] + (1,) * (model_output.ndim - 1))
return model_input - model_output * sigma
+ def noise_scaling(self, sigma, noise, latent_image, max_denoise=False):
+ if max_denoise:
+ noise = noise * torch.sqrt(1.0 + sigma ** 2.0)
+ else:
+ noise = noise * sigma
+ if latent_image is not None:
+ noise += latent_image
+ return noise
class V_PREDICTION(EPS):
def calculate_denoised(self, sigma, model_output, model_input):
diff --git a/comfy/samplers.py b/comfy/samplers.py
index e5569322f..46d444f54 100644
--- a/comfy/samplers.py
+++ b/comfy/samplers.py
@@ -534,19 +534,13 @@ class KSAMPLER(Sampler):
else:
model_k.noise = noise
- if self.max_denoise(model_wrap, sigmas):
- noise = noise * torch.sqrt(1.0 + sigmas[0] ** 2.0)
- else:
- noise = noise * sigmas[0]
+ noise = model_wrap.inner_model.model_sampling.noise_scaling(sigmas[0], noise, latent_image, self.max_denoise(model_wrap, sigmas))
k_callback = None
total_steps = len(sigmas) - 1
if callback is not None:
k_callback = lambda x: callback(x["i"], x["denoised"], x["x"], total_steps)
- if latent_image is not None:
- noise += latent_image
-
samples = self.sampler_function(model_k, noise, sigmas, extra_args=extra_args, callback=k_callback, disable=disable_pbar, **self.extra_options)
return samples