New CheckpointLoaderSimple to load checkpoints without a config.

comfy/ldm/models/diffusion/ddpm.py

@@ -81,7 +81,7 @@ class DDPM(torch.nn.Module):
         super().__init__()
         assert parameterization in ["eps", "x0", "v"], 'currently only supporting "eps" and "x0" and "v"'
         self.parameterization = parameterization
-        print(f"{self.__class__.__name__}: Running in {self.parameterization}-prediction mode")
+        # print(f"{self.__class__.__name__}: Running in {self.parameterization}-prediction mode")
         self.cond_stage_model = None
         self.clip_denoised = clip_denoised
         self.log_every_t = log_every_t
@@ -522,8 +522,8 @@ class LatentDiffusion(DDPM):
     """main class"""
 
     def __init__(self,
-                 first_stage_config,
-                 cond_stage_config,
+                 first_stage_config={},
+                 cond_stage_config={},
                  num_timesteps_cond=None,
                  cond_stage_key="image",
                  cond_stage_trainable=False,
@@ -562,8 +562,6 @@ class LatentDiffusion(DDPM):
 
         # self.instantiate_first_stage(first_stage_config)
         # self.instantiate_cond_stage(cond_stage_config)
-        self.first_stage_config = first_stage_config
-        self.cond_stage_config = cond_stage_config
 
         self.cond_stage_forward = cond_stage_forward
         self.clip_denoised = False

comfy/sd.py

@@ -317,9 +317,7 @@ class VAE:
         pixel_samples = pixel_samples.cpu().movedim(1,-1)
         return pixel_samples
 
-    def decode_tiled(self, samples):
-        tile_x = tile_y = 64
-        overlap = 8
+    def decode_tiled(self, samples, tile_x=64, tile_y=64, overlap = 8):
         model_management.unload_model()
         output = torch.empty((samples.shape[0], 3, samples.shape[2] * 8, samples.shape[3] * 8), device="cpu")
         self.first_stage_model = self.first_stage_model.to(self.device)
@@ -656,3 +654,103 @@ def load_checkpoint(config_path, ckpt_path, output_vae=True, output_clip=True, e
     sd = load_torch_file(ckpt_path)
     model = load_model_weights(model, sd, verbose=False, load_state_dict_to=load_state_dict_to)
     return (ModelPatcher(model), clip, vae)
+
+
+def load_checkpoint_guess_config(ckpt_path, fp16=False, output_vae=True, output_clip=True, embedding_directory=None):
+    sd = load_torch_file(ckpt_path)
+    sd_keys = sd.keys()
+    clip = None
+    vae = None
+
+    class WeightsLoader(torch.nn.Module):
+        pass
+
+    w = WeightsLoader()
+    load_state_dict_to = []
+    if output_vae:
+        vae = VAE()
+        w.first_stage_model = vae.first_stage_model
+        load_state_dict_to = [w]
+
+    if output_clip:
+        clip_config = {}
+        if "cond_stage_model.model.transformer.resblocks.22.attn.out_proj.weight" in sd_keys:
+            clip_config['target'] = 'ldm.modules.encoders.modules.FrozenOpenCLIPEmbedder'
+        else:
+            clip_config['target'] = 'ldm.modules.encoders.modules.FrozenCLIPEmbedder'
+        clip = CLIP(config=clip_config, embedding_directory=embedding_directory)
+        w.cond_stage_model = clip.cond_stage_model
+        load_state_dict_to = [w]
+
+    sd_config = {
+        "linear_start": 0.00085,
+        "linear_end": 0.012,
+        "num_timesteps_cond": 1,
+        "log_every_t": 200,
+        "timesteps": 1000,
+        "first_stage_key": "jpg",
+        "cond_stage_key": "txt",
+        "image_size": 64,
+        "channels": 4,
+        "cond_stage_trainable": False,
+        "monitor": "val/loss_simple_ema",
+        "scale_factor": 0.18215,
+        "use_ema": False,
+    }
+
+    unet_config = {
+        "use_checkpoint": True,
+        "image_size": 32,
+        "out_channels": 4,
+        "attention_resolutions": [
+            4,
+            2,
+            1
+        ],
+        "num_res_blocks": 2,
+        "channel_mult": [
+            1,
+            2,
+            4,
+            4
+        ],
+        "use_spatial_transformer": True,
+        "transformer_depth": 1,
+        "legacy": False
+    }
+
+    if len(sd['model.diffusion_model.input_blocks.1.1.proj_in.weight'].shape) == 2:
+        unet_config['use_linear_in_transformer'] = True
+
+    unet_config["use_fp16"] = fp16
+    unet_config["model_channels"] = sd['model.diffusion_model.input_blocks.0.0.weight'].shape[0]
+    unet_config["in_channels"] = sd['model.diffusion_model.input_blocks.0.0.weight'].shape[1]
+    unet_config["context_dim"] = sd['model.diffusion_model.input_blocks.1.1.transformer_blocks.0.attn2.to_k.weight'].shape[1]
+
+    sd_config["unet_config"] = {"target": "ldm.modules.diffusionmodules.openaimodel.UNetModel", "params": unet_config}
+    model_config = {"target": "ldm.models.diffusion.ddpm.LatentDiffusion", "params": sd_config}
+
+    if unet_config["in_channels"] > 4: #inpainting model
+        sd_config["conditioning_key"] = "hybrid"
+        sd_config["finetune_keys"] = None
+        model_config["target"] = "ldm.models.diffusion.ddpm.LatentInpaintDiffusion"
+    else:
+        sd_config["conditioning_key"] = "crossattn"
+
+    if unet_config["context_dim"] == 1024:
+        unet_config["num_head_channels"] = 64 #SD2.x
+    else:
+        unet_config["num_heads"] = 8 #SD1.x
+
+
+    model = instantiate_from_config(model_config)
+    model = load_model_weights(model, sd, verbose=False, load_state_dict_to=load_state_dict_to)
+
+    if unet_config["context_dim"] == 1024 and unet_config["in_channels"] == 4: #only SD2.x non inpainting models are v prediction
+        cond = torch.zeros((1, 2, unet_config["context_dim"]), device="cpu")
+        x_in = torch.rand((1, unet_config["in_channels"], 8, 8), device="cpu", generator=torch.manual_seed(1))
+        out = model.apply_model(x_in, torch.tensor([999], device="cpu"), cond)
+        if out.mean() < -0.6: #mean of eps should be ~0 and mean of v prediction should be ~-1
+            model.parameterization = 'v'
+
+    return (ModelPatcher(model), clip, vae)

nodes.py

@@ -202,6 +202,28 @@ class CheckpointLoader:
         ckpt_path = os.path.join(self.ckpt_dir, ckpt_name)
         return comfy.sd.load_checkpoint(config_path, ckpt_path, output_vae=True, output_clip=True, embedding_directory=self.embedding_directory)
 
+class CheckpointLoaderSimple:
+    models_dir = os.path.join(os.path.dirname(os.path.realpath(__file__)), "models")
+    ckpt_dir = os.path.join(models_dir, "checkpoints")
+
+    @classmethod
+    def INPUT_TYPES(s):
+        return {"required": { "ckpt_name": (filter_files_extensions(recursive_search(s.ckpt_dir), supported_ckpt_extensions), ),
+                              "type": (["fp16", "fp32"],),
+                              "stop_at_clip_layer": ("INT", {"default": -1, "min": -24, "max": -1, "step": 1}),
+                             }}
+    RETURN_TYPES = ("MODEL", "CLIP", "VAE")
+    FUNCTION = "load_checkpoint"
+
+    CATEGORY = "_for_testing"
+
+    def load_checkpoint(self, ckpt_name, type, stop_at_clip_layer, output_vae=True, output_clip=True):
+        ckpt_path = os.path.join(self.ckpt_dir, ckpt_name)
+        out = comfy.sd.load_checkpoint_guess_config(ckpt_path, type=="fp16", output_vae=True, output_clip=True, embedding_directory=CheckpointLoader.embedding_directory)
+        if out[1] is not None:
+            out[1].clip_layer(stop_at_clip_layer)
+        return out
+
 class LoraLoader:
     models_dir = os.path.join(os.path.dirname(os.path.realpath(__file__)), "models")
     lora_dir = os.path.join(models_dir, "loras")
@@ -837,6 +859,7 @@ NODE_CLASS_MAPPINGS = {
     "DiffControlNetLoader": DiffControlNetLoader,
     "T2IAdapterLoader": T2IAdapterLoader,
     "VAEDecodeTiled": VAEDecodeTiled,
+    "CheckpointLoaderSimple": CheckpointLoaderSimple,
 }
 
 CUSTOM_NODE_PATH = os.path.join(os.path.dirname(os.path.realpath(__file__)), "custom_nodes")