i

modules/core.py

@@ -2,13 +2,17 @@ import random
 import torch
 import numpy as np
 
+import comfy.model_management
+import comfy.sample
+import comfy.utils
+import latent_preview
+
 from comfy.sd import load_checkpoint_guess_config
-from nodes import VAEDecode, KSamplerAdvanced, EmptyLatentImage, CLIPTextEncode
+from nodes import VAEDecode, EmptyLatentImage, CLIPTextEncode, common_ksampler
 
 
 opCLIPTextEncode = CLIPTextEncode()
 opEmptyLatentImage = EmptyLatentImage()
-opKSamplerAdvanced = KSamplerAdvanced()
 opVAEDecode = VAEDecode()
 
 
@@ -42,24 +46,42 @@ def decode_vae(vae, latent_image):
 
 
 @torch.no_grad()
-def ksample(unet, positive_condition, negative_condition, latent_image, add_noise=True, noise_seed=None, steps=25, cfg=9,
+def ksampler(model, positive, negative, latent, seed=None, steps=30, cfg=9.0, sampler_name='euler_ancestral', scheduler='normal', denoise=1.0, disable_noise=False, start_step=None, last_step=None, force_full_denoise=False):
-            sampler_name='euler_ancestral', scheduler='normal', start_at_step=None, end_at_step=None,
+    seed = seed if isinstance(seed, int) else random.randint(1, 2 ** 64)
-            return_with_leftover_noise=False):
+
-    return opKSamplerAdvanced.sample(
+    device = comfy.model_management.get_torch_device()
-        add_noise='enable' if add_noise else 'disable',
+    latent_image = latent["samples"]
-        noise_seed=noise_seed if isinstance(noise_seed, int) else random.randint(1, 2 ** 64),
+
-        steps=steps,
+    if disable_noise:
-        cfg=cfg,
+        noise = torch.zeros(latent_image.size(), dtype=latent_image.dtype, layout=latent_image.layout, device="cpu")
-        sampler_name=sampler_name,
+    else:
-        scheduler=scheduler,
+        batch_inds = latent["batch_index"] if "batch_index" in latent else None
-        start_at_step=0 if start_at_step is None else start_at_step,
+        noise = comfy.sample.prepare_noise(latent_image, seed, batch_inds)
-        end_at_step=steps if end_at_step is None else end_at_step,
+
-        return_with_leftover_noise='enable' if return_with_leftover_noise else 'disable',
+    noise_mask = None
-        model=unet,
+    if "noise_mask" in latent:
-        positive=positive_condition,
+        noise_mask = latent["noise_mask"]
-        negative=negative_condition,
+
-        latent_image=latent_image,
+    preview_format = "JPEG"
-    )[0]
+    if preview_format not in ["JPEG", "PNG"]:
+        preview_format = "JPEG"
+
+    previewer = latent_preview.get_previewer(device, model.model.latent_format)
+
+    pbar = comfy.utils.ProgressBar(steps)
+
+    def callback(step, x0, x, total_steps):
+        preview_bytes = None
+        if previewer:
+            preview_bytes = previewer.decode_latent_to_preview_image(preview_format, x0)
+        pbar.update_absolute(step + 1, total_steps, preview_bytes)
+
+    samples = comfy.sample.sample(model, noise, steps, cfg, sampler_name, scheduler, positive, negative, latent_image,
+                                  denoise=denoise, disable_noise=disable_noise, start_step=start_step, last_step=last_step,
+                                  force_full_denoise=force_full_denoise, noise_mask=noise_mask, callback=callback, seed=seed)
+    out = latent.copy()
+    out["samples"] = samples
+    return (out, )
 
 
 @torch.no_grad()

modules/default_pipeline.py

@@ -23,20 +23,20 @@ def process(positive_prompt, negative_prompt, width=1024, height=1024, batch_siz
 
     empty_latent = core.generate_empty_latent(width=width, height=height, batch_size=batch_size)
 
-    sampled_latent = core.ksample(
+    sampled_latent = core.ksampler(
-        unet=xl_base.unet,
+        model=xl_base.unet,
-        positive_condition=positive_conditions,
+        positive=positive_conditions,
-        negative_condition=negative_conditions,
+        negative=negative_conditions,
-        latent_image=empty_latent,
+        latent=empty_latent,
-        steps=30, start_at_step=0, end_at_step=20, return_with_leftover_noise=True, add_noise=True
+        steps=30, start_step=0, last_step=20, disable_noise=False, force_full_denoise=False
     )
 
-    sampled_latent = core.ksample(
+    sampled_latent = core.ksampler(
-        unet=xl_refiner.unet,
+        model=xl_refiner.unet,
-        positive_condition=positive_conditions_refiner,
+        positive=positive_conditions_refiner,
-        negative_condition=negative_conditions_refiner,
+        negative=negative_conditions_refiner,
-        latent_image=sampled_latent,
+        latent=sampled_latent,
-        steps=30, start_at_step=20, end_at_step=30, return_with_leftover_noise=False, add_noise=False
+        steps=30, start_step=20, last_step=30, disable_noise=True, force_full_denoise=True
     )
 
     decoded_latent = core.decode_vae(vae=xl_refiner.vae, latent_image=sampled_latent)