[core] ConsisID

examples/research_projects/gligen/demo.ipynb

@@ -26,8 +26,7 @@
     "%load_ext autoreload\n",
     "%autoreload 2\n",
     "\n",
-    "import torch\n",
-    "from diffusers import StableDiffusionGLIGENTextImagePipeline, StableDiffusionGLIGENPipeline"
+    "from diffusers import StableDiffusionGLIGENPipeline"
    ]
   },
   {
@@ -36,16 +35,17 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "import os\n",
+    "from transformers import CLIPTextModel, CLIPTokenizer\n",
+    "\n",
     "import diffusers\n",
     "from diffusers import (\n",
     "    AutoencoderKL,\n",
     "    DDPMScheduler,\n",
-    "    UNet2DConditionModel,\n",
-    "    UniPCMultistepScheduler,\n",
     "    EulerDiscreteScheduler,\n",
+    "    UNet2DConditionModel,\n",
     ")\n",
-    "from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer\n",
+    "\n",
+    "\n",
     "# pretrained_model_name_or_path = 'masterful/gligen-1-4-generation-text-box'\n",
     "\n",
     "pretrained_model_name_or_path = '/root/data/zhizhonghuang/checkpoints/models--masterful--gligen-1-4-generation-text-box/snapshots/d2820dc1e9ba6ca082051ce79cfd3eb468ae2c83'\n",
@@ -122,6 +122,7 @@
     "\n",
     "import numpy as np\n",
     "\n",
+    "\n",
     "boxes = np.array([x[1] for x in gen_boxes])\n",
     "boxes = boxes / 512\n",
     "boxes[:, 2] = boxes[:, 0] + boxes[:, 2]\n",

src/diffusers/__init__.py

@@ -90,6 +90,7 @@
             "AutoencoderTiny",
             "CogVideoXTransformer3DModel",
             "CogView3PlusTransformer2DModel",
+            "ConsisIDTransformer3DModel",
             "ConsistencyDecoderVAE",
             "ControlNetModel",
             "ControlNetXSAdapter",
@@ -269,6 +270,7 @@
             "CogVideoXPipeline",
             "CogVideoXVideoToVideoPipeline",
             "CogView3PlusPipeline",
+            "ConsisIDPipeline",
             "CycleDiffusionPipeline",
             "FluxControlImg2ImgPipeline",
             "FluxControlNetImg2ImgPipeline",
@@ -583,6 +585,7 @@
             AutoencoderTiny,
             CogVideoXTransformer3DModel,
             CogView3PlusTransformer2DModel,
+            ConsisIDTransformer3DModel,
             ConsistencyDecoderVAE,
             ControlNetModel,
             ControlNetXSAdapter,
@@ -741,6 +744,7 @@
             CogVideoXPipeline,
             CogVideoXVideoToVideoPipeline,
             CogView3PlusPipeline,
+            ConsisIDPipeline,
             CycleDiffusionPipeline,
             FluxControlImg2ImgPipeline,
             FluxControlNetImg2ImgPipeline,

src/diffusers/models/__init__.py

@@ -51,6 +51,7 @@
     _import_structure["modeling_utils"] = ["ModelMixin"]
     _import_structure["transformers.auraflow_transformer_2d"] = ["AuraFlowTransformer2DModel"]
     _import_structure["transformers.cogvideox_transformer_3d"] = ["CogVideoXTransformer3DModel"]
+    _import_structure["transformers.consisid_transformer_3d"] = ["ConsisIDTransformer3DModel"]
     _import_structure["transformers.dit_transformer_2d"] = ["DiTTransformer2DModel"]
     _import_structure["transformers.dual_transformer_2d"] = ["DualTransformer2DModel"]
     _import_structure["transformers.hunyuan_transformer_2d"] = ["HunyuanDiT2DModel"]
@@ -120,6 +121,7 @@
             AuraFlowTransformer2DModel,
             CogVideoXTransformer3DModel,
             CogView3PlusTransformer2DModel,
+            ConsisIDTransformer3DModel,
             DiTTransformer2DModel,
             DualTransformer2DModel,
             FluxTransformer2DModel,

src/diffusers/models/transformers/__init__.py

@@ -4,6 +4,7 @@
 if is_torch_available():
     from .auraflow_transformer_2d import AuraFlowTransformer2DModel
     from .cogvideox_transformer_3d import CogVideoXTransformer3DModel
+    from .consisid_transformer_3d import ConsisIDTransformer3DModel
     from .dit_transformer_2d import DiTTransformer2DModel
     from .dual_transformer_2d import DualTransformer2DModel
     from .hunyuan_transformer_2d import HunyuanDiT2DModel