The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface. The fastest local inference engine in the world.
Fix regression with ltxv VAE.
comfyanonymousf4cdedea62e8
diff --git a/comfy/sd.py b/comfy/sd.pyindex 89f0eb24e..16b0acc89 100644--- a/comfy/sd.py+++ b/comfy/sd.py@@ -313,18 +313,6 @@ class VAE:self.first_stage_model = AutoencodingEngine(regularizer_config={'target': "comfy.ldm.models.autoencoder.DiagonalGaussianRegularizer"},encoder_config={'target': "comfy.ldm.modules.diffusionmodules.model.Encoder", 'params': ddconfig},decoder_config={'target': "comfy.ldm.modules.diffusionmodules.model.Decoder", 'params': ddconfig})- elif "decoder.conv_in.conv.weight" in sd:- ddconfig = {'double_z': True, 'z_channels': 4, 'resolution': 256, 'in_channels': 3, 'out_ch': 3, 'ch': 128, 'ch_mult': [1, 2, 4, 4], 'num_res_blocks': 2, 'attn_resolutions': [], 'dropout': 0.0}- ddconfig["conv3d"] = True- ddconfig["time_compress"] = 4- self.upscale_ratio = (lambda a: max(0, a * 4 - 3), 8, 8)- self.latent_dim = 3- self.latent_channels = ddconfig['z_channels'] = sd["decoder.conv_in.conv.weight"].shape[1]- self.first_stage_model = AutoencoderKL(ddconfig=ddconfig, embed_dim=sd['post_quant_conv.weight'].shape[1])- self.memory_used_decode = lambda shape, dtype: (1500 * shape[2] * shape[3] * shape[4] * (4 * 8 * 8)) * model_management.dtype_size(dtype)- self.memory_used_encode = lambda shape, dtype: (900 * max(shape[2], 2) * shape[3] * shape[4]) * model_management.dtype_size(dtype)- self.working_dtypes = [torch.bfloat16, torch.float16, torch.float32]-elif "decoder.layers.1.layers.0.beta" in sd:self.first_stage_model = AudioOobleckVAE()self.memory_used_encode = lambda shape, dtype: (1000 * shape[2]) * model_management.dtype_size(dtype)@@ -357,6 +345,17 @@ class VAE:self.memory_used_encode = lambda shape, dtype: (70 * max(shape[2], 7) * shape[3] * shape[4]) * model_management.dtype_size(dtype)self.upscale_ratio = (lambda a: max(0, a * 8 - 7), 32, 32)self.working_dtypes = [torch.bfloat16, torch.float32]+ elif "decoder.conv_in.conv.weight" in sd:+ ddconfig = {'double_z': True, 'z_channels': 4, 'resolution': 256, 'in_channels': 3, 'out_ch': 3, 'ch': 128, 'ch_mult': [1, 2, 4, 4], 'num_res_blocks': 2, 'attn_resolutions': [], 'dropout': 0.0}+ ddconfig["conv3d"] = True+ ddconfig["time_compress"] = 4+ self.upscale_ratio = (lambda a: max(0, a * 4 - 3), 8, 8)+ self.latent_dim = 3+ self.latent_channels = ddconfig['z_channels'] = sd["decoder.conv_in.conv.weight"].shape[1]+ self.first_stage_model = AutoencoderKL(ddconfig=ddconfig, embed_dim=sd['post_quant_conv.weight'].shape[1])+ self.memory_used_decode = lambda shape, dtype: (1500 * shape[2] * shape[3] * shape[4] * (4 * 8 * 8)) * model_management.dtype_size(dtype)+ self.memory_used_encode = lambda shape, dtype: (900 * max(shape[2], 2) * shape[3] * shape[4]) * model_management.dtype_size(dtype)+ self.working_dtypes = [torch.bfloat16, torch.float16, torch.float32]else:logging.warning("WARNING: No VAE weights detected, VAE not initalized.")self.first_stage_model = None