update to v1.5

apps/third_party/CRM/LICENSE

@@ -1,21 +0,0 @@
-MIT License
-
-Copyright (c) 2024 TSAIL group
-
-Permission is hereby granted, free of charge, to any person obtaining a copy
-of this software and associated documentation files (the "Software"), to deal
-in the Software without restriction, including without limitation the rights
-to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
-copies of the Software, and to permit persons to whom the Software is
-furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice shall be included in all
-copies or substantial portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
-OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
-SOFTWARE.

apps/third_party/CRM/README.md

@@ -1,85 +0,0 @@
-# Convolutional Reconstruction Model
-
-Official implementation for *CRM: Single Image to 3D Textured Mesh with Convolutional Reconstruction Model*.
-
-**CRM is a feed-forward model which can generate 3D textured mesh in 10 seconds.**
-
-## [Project Page](https://ml.cs.tsinghua.edu.cn/~zhengyi/CRM/) | [Arxiv](https://arxiv.org/abs/2403.05034) | [HF-Demo](https://huggingface.co/spaces/Zhengyi/CRM) | [Weights](https://huggingface.co/Zhengyi/CRM)
-
-https://github.com/thu-ml/CRM/assets/40787266/8b325bc0-aa74-4c26-92e8-a8f0c1079382
-
-## Try CRM 🍻
-* Try CRM at [Huggingface Demo](https://huggingface.co/spaces/Zhengyi/CRM).
-* Try CRM at [Replicate Demo](https://replicate.com/camenduru/crm). Thanks [@camenduru](https://github.com/camenduru)! 
-
-## Install
-
-### Step 1 - Base
-
-Install package one by one, we use **python 3.9**
-
-```bash
-pip install torch==1.13.0+cu117 torchvision==0.14.0+cu117 torchaudio==0.13.0 --extra-index-url https://download.pytorch.org/whl/cu117
-pip install torch-scatter==2.1.1 -f https://data.pyg.org/whl/torch-1.13.1+cu117.html
-pip install kaolin==0.14.0 -f https://nvidia-kaolin.s3.us-east-2.amazonaws.com/torch-1.13.1_cu117.html
-pip install -r requirements.txt
-```
-
-besides, one by one need to install xformers manually according to the official [doc](https://github.com/facebookresearch/xformers?tab=readme-ov-file#installing-xformers) (**conda no need**), e.g.
-
-```bash
-pip install ninja
-pip install -v -U git+https://github.com/facebookresearch/xformers.git@main#egg=xformers
-```
-
-### Step 2 - Nvdiffrast
-
-Install nvdiffrast according to the official [doc](https://nvlabs.github.io/nvdiffrast/#installation), e.g.
-
-```bash
-pip install git+https://github.com/NVlabs/nvdiffrast
-```
-
-
-
-## Inference
-
-We suggest gradio for a visualized inference.
-
-```
-gradio app.py
-```
-
-![image](https://github.com/thu-ml/CRM/assets/40787266/4354d22a-a641-4531-8408-c761ead8b1a2)
-
-For inference in command lines, simply run
-```bash
-CUDA_VISIBLE_DEVICES="0" python run.py --inputdir "examples/kunkun.webp"
-```
-It will output the preprocessed image, generated 6-view images and CCMs and a 3D model in obj format.
-
-**Tips:** (1) If the result is unsatisfatory, please check whether the input image is correctly pre-processed into a grey background. Otherwise the results will be unpredictable.
-(2) Different from the [Huggingface Demo](https://huggingface.co/spaces/Zhengyi/CRM), this official implementation uses UV texture instead of vertex color. It has better texture than the online demo but longer generating time owing to the UV texturing.
-
-## Todo List
-- [x] Release inference code.
-- [x] Release pretrained models.
-- [ ] Optimize inference code to fit in low memery GPU.
-- [ ] Upload training code.
-
-## Acknowledgement
-- [ImageDream](https://github.com/bytedance/ImageDream)
-- [nvdiffrast](https://github.com/NVlabs/nvdiffrast)
-- [kiuikit](https://github.com/ashawkey/kiuikit)
-- [GET3D](https://github.com/nv-tlabs/GET3D)
-
-## Citation
-
-```
-@article{wang2024crm,
-  title={CRM: Single Image to 3D Textured Mesh with Convolutional Reconstruction Model},
-  author={Zhengyi Wang and Yikai Wang and Yifei Chen and Chendong Xiang and Shuo Chen and Dajiang Yu and Chongxuan Li and Hang Su and Jun Zhu},
-  journal={arXiv preprint arXiv:2403.05034},
-  year={2024}
-}
-```

apps/third_party/CRM/app.py

@@ -1,228 +0,0 @@
-# Not ready to use yet
-import argparse
-import numpy as np
-import gradio as gr
-from omegaconf import OmegaConf
-import torch
-from PIL import Image
-import PIL
-from pipelines import TwoStagePipeline
-from huggingface_hub import hf_hub_download
-import os
-import rembg
-from typing import Any
-import json
-import os
-import json
-import argparse
-
-from model import CRM
-from inference import generate3d
-
-pipeline = None
-rembg_session = rembg.new_session()
-
-
-def expand_to_square(image, bg_color=(0, 0, 0, 0)):
-    # expand image to 1:1
-    width, height = image.size
-    if width == height:
-        return image
-    new_size = (max(width, height), max(width, height))
-    new_image = Image.new("RGBA", new_size, bg_color)
-    paste_position = ((new_size[0] - width) // 2, (new_size[1] - height) // 2)
-    new_image.paste(image, paste_position)
-    return new_image
-
-def check_input_image(input_image):
-    if input_image is None:
-        raise gr.Error("No image uploaded!")
-
-
-def remove_background(
-    image: PIL.Image.Image,
-    rembg_session = None,
-    force: bool = False,
-    **rembg_kwargs,
-) -> PIL.Image.Image:
-    do_remove = True
-    if image.mode == "RGBA" and image.getextrema()[3][0] < 255:
-        # explain why current do not rm bg
-        print("alhpa channl not enpty, skip remove background, using alpha channel as mask")
-        background = Image.new("RGBA", image.size, (0, 0, 0, 0))
-        image = Image.alpha_composite(background, image)
-        do_remove = False
-    do_remove = do_remove or force
-    if do_remove:
-        image = rembg.remove(image, session=rembg_session, **rembg_kwargs)
-    return image
-
-def do_resize_content(original_image: Image, scale_rate):
-    # resize image content wile retain the original image size
-    if scale_rate != 1:
-        # Calculate the new size after rescaling
-        new_size = tuple(int(dim * scale_rate) for dim in original_image.size)
-        # Resize the image while maintaining the aspect ratio
-        resized_image = original_image.resize(new_size)
-        # Create a new image with the original size and black background
-        padded_image = Image.new("RGBA", original_image.size, (0, 0, 0, 0))
-        paste_position = ((original_image.width - resized_image.width) // 2, (original_image.height - resized_image.height) // 2)
-        padded_image.paste(resized_image, paste_position)
-        return padded_image
-    else:
-        return original_image
-
-def add_background(image, bg_color=(255, 255, 255)):
-    # given an RGBA image, alpha channel is used as mask to add background color
-    background = Image.new("RGBA", image.size, bg_color)
-    return Image.alpha_composite(background, image)
-
-
-def preprocess_image(image, background_choice, foreground_ratio, backgroud_color):
-    """
-    input image is a pil image in RGBA, return RGB image
-    """
-    print(background_choice)
-    if background_choice == "Alpha as mask":
-        background = Image.new("RGBA", image.size, (0, 0, 0, 0))
-        image = Image.alpha_composite(background, image)
-    else:
-        image = remove_background(image, rembg_session, force_remove=True)
-    image = do_resize_content(image, foreground_ratio)
-    image = expand_to_square(image)
-    image = add_background(image, backgroud_color)
-    return image.convert("RGB")
-
-
-def gen_image(input_image, seed, scale, step):
-    global pipeline, model, args
-    pipeline.set_seed(seed)
-    rt_dict = pipeline(input_image, scale=scale, step=step)
-    stage1_images = rt_dict["stage1_images"]
-    stage2_images = rt_dict["stage2_images"]
-    np_imgs = np.concatenate(stage1_images, 1)
-    np_xyzs = np.concatenate(stage2_images, 1)
-
-    glb_path, obj_path = generate3d(model, np_imgs, np_xyzs, args.device)
-    return Image.fromarray(np_imgs), Image.fromarray(np_xyzs), glb_path, obj_path
-
-
-parser = argparse.ArgumentParser()
-parser.add_argument(
-    "--stage1_config",
-    type=str,
-    default="configs/nf7_v3_SNR_rd_size_stroke.yaml",
-    help="config for stage1",
-)
-parser.add_argument(
-    "--stage2_config",
-    type=str,
-    default="configs/stage2-v2-snr.yaml",
-    help="config for stage2",
-)
-
-parser.add_argument("--device", type=str, default="cuda")
-args = parser.parse_args()
-
-crm_path = hf_hub_download(repo_id="Zhengyi/CRM", filename="CRM.pth")
-specs = json.load(open("configs/specs_objaverse_total.json"))
-model = CRM(specs).to(args.device)
-model.load_state_dict(torch.load(crm_path, map_location = args.device), strict=False)
-
-stage1_config = OmegaConf.load(args.stage1_config).config
-stage2_config = OmegaConf.load(args.stage2_config).config
-stage2_sampler_config = stage2_config.sampler
-stage1_sampler_config = stage1_config.sampler
-
-stage1_model_config = stage1_config.models
-stage2_model_config = stage2_config.models
-
-xyz_path = hf_hub_download(repo_id="Zhengyi/CRM", filename="ccm-diffusion.pth")
-pixel_path = hf_hub_download(repo_id="Zhengyi/CRM", filename="pixel-diffusion.pth")
-stage1_model_config.resume = pixel_path
-stage2_model_config.resume = xyz_path
-
-pipeline = TwoStagePipeline(
-    stage1_model_config,
-    stage2_model_config,
-    stage1_sampler_config,
-    stage2_sampler_config,
-    device=args.device,
-    dtype=torch.float16
-)
-
-with gr.Blocks() as demo:
-    gr.Markdown("# CRM: Single Image to 3D Textured Mesh with Convolutional Reconstruction Model")
-    with gr.Row():
-        with gr.Column():
-            with gr.Row():
-                image_input = gr.Image(
-                    label="Image input",
-                    image_mode="RGBA",
-                    sources="upload",
-                    type="pil",
-                )
-                processed_image = gr.Image(label="Processed Image", interactive=False, type="pil", image_mode="RGB")
-            with gr.Row():
-                with gr.Column():
-                    with gr.Row():
-                        background_choice = gr.Radio([
-                                "Alpha as mask",
-                                "Auto Remove background"
-                            ], value="Auto Remove background",
-                            label="backgroud choice")
-                        # do_remove_background = gr.Checkbox(label=, value=True)
-                        # force_remove = gr.Checkbox(label=, value=False)
-                    back_groud_color = gr.ColorPicker(label="Background Color", value="#7F7F7F", interactive=False)
-                    foreground_ratio = gr.Slider(
-                        label="Foreground Ratio",
-                        minimum=0.5,
-                        maximum=1.0,
-                        value=1.0,
-                        step=0.05,
-                    )
-
-                with gr.Column():
-                    seed = gr.Number(value=1234, label="seed", precision=0)
-                    guidance_scale = gr.Number(value=5.5, minimum=3, maximum=10, label="guidance_scale")
-                    step = gr.Number(value=50, minimum=30, maximum=100, label="sample steps", precision=0)
-            text_button = gr.Button("Generate 3D shape")
-            gr.Examples(
-                examples=[os.path.join("examples", i) for i in os.listdir("examples")],
-                inputs=[image_input],
-            )
-        with gr.Column():
-            image_output = gr.Image(interactive=False, label="Output RGB image")
-            xyz_ouput = gr.Image(interactive=False, label="Output CCM image")
-
-            output_model = gr.Model3D(
-                label="Output GLB",
-                interactive=False,
-            )
-            gr.Markdown("Note: The GLB model shown here has a darker lighting and enlarged UV seams. Download for correct results.")
-            output_obj = gr.File(interactive=False, label="Output OBJ")
-
-    inputs = [
-        processed_image,
-        seed,
-        guidance_scale,
-        step,
-    ]
-    outputs = [
-        image_output,
-        xyz_ouput,
-        output_model,
-        output_obj,
-    ]
-
-
-    text_button.click(fn=check_input_image, inputs=[image_input]).success(
-        fn=preprocess_image,
-        inputs=[image_input, background_choice, foreground_ratio, back_groud_color],
-        outputs=[processed_image],
-    ).success(
-        fn=gen_image,
-        inputs=inputs,
-        outputs=outputs,
-    )
-    demo.queue().launch()

apps/third_party/CRM/configs/nf7_v3_SNR_rd_size_stroke.yaml

@@ -1,21 +0,0 @@
-config:
-# others
-    seed: 1234
-    num_frames: 7
-    mode: pixel
-    offset_noise: true
-# model related
-    models:
-        config: imagedream/configs/sd_v2_base_ipmv_zero_SNR.yaml
-        resume: models/pixel.pth
-# sampler related
-    sampler:
-        target: libs.sample.ImageDreamDiffusion
-        params:
-            mode: pixel
-            num_frames: 7
-            camera_views: [1, 2, 3, 4, 5, 0, 0]
-            ref_position: 6
-            random_background: false
-            offset_noise: true
-            resize_rate: 1.0

apps/third_party/CRM/configs/specs_objaverse_total.json

@@ -1,57 +0,0 @@
-{
-  "Input": {
-    "img_num": 16,
-    "class": "all",
-    "camera_angle_num": 8,
-    "tet_grid_size": 80,
-    "validate_num": 16,
-    "scale": 0.95,
-    "radius": 3,
-    "resolution": [256, 256]
-  },
-
-  "Pretrain": {
-    "mode": null,
-    "sdf_threshold": 0.1,
-    "sdf_scale": 10,
-    "batch_infer": false,
-    "lr": 1e-4,
-    "radius": 0.5
-  },
-
-  "Train": {
-    "mode": "rnd",
-    "num_epochs": 500,
-    "grad_acc": 1,
-    "warm_up": 0,
-    "decay": 0.000,
-    "learning_rate": {
-      "init": 1e-4,
-      "sdf_decay": 1,
-      "rgb_decay": 1
-    },
-    "batch_size": 4,
-    "eva_iter": 80,
-    "eva_all_epoch": 10,
-    "tex_sup_mode": "blender",
-    "exp_uv_mesh": false,
-    "doub": false,
-    "random_bg": false,
-    "shift": 0,
-    "aug_shift": 0,
-    "geo_type": "flex"
-  },
-
-  "ArchSpecs": {
-    "unet_type": "diffusers",
-    "use_3D_aware": false,
-    "fea_concat": false,
-    "mlp_bias": true
-  },
-
-  "DecoderSpecs": {
-    "c_dim": 32,
-    "plane_resolution": 256
-  }
-}
-

apps/third_party/CRM/configs/stage2-v2-snr.yaml

@@ -1,25 +0,0 @@
-config:
-# others
-    seed: 1234
-    num_frames: 6
-    mode: pixel
-    offset_noise: true
-    gd_type: xyz
-# model related
-    models:
-        config: imagedream/configs/sd_v2_base_ipmv_chin8_zero_snr.yaml
-        resume: models/xyz.pth
-
-# eval related
-    sampler:
-        target: libs.sample.ImageDreamDiffusionStage2
-        params:
-            mode: pixel
-            num_frames: 6
-            camera_views: [1, 2, 3, 4, 5, 0]
-            ref_position: null
-            random_background: false
-            offset_noise: true
-            resize_rate: 1.0
-
-

apps/third_party/CRM/imagedream/__init__.py

@@ -1 +0,0 @@
-from .model_zoo import build_model

apps/third_party/CRM/imagedream/camera_utils.py

@@ -1,99 +0,0 @@
-import numpy as np
-import torch
-
-
-def create_camera_to_world_matrix(elevation, azimuth):
-    elevation = np.radians(elevation)
-    azimuth = np.radians(azimuth)
-    # Convert elevation and azimuth angles to Cartesian coordinates on a unit sphere
-    x = np.cos(elevation) * np.sin(azimuth)
-    y = np.sin(elevation)
-    z = np.cos(elevation) * np.cos(azimuth)
-
-    # Calculate camera position, target, and up vectors
-    camera_pos = np.array([x, y, z])
-    target = np.array([0, 0, 0])
-    up = np.array([0, 1, 0])
-
-    # Construct view matrix
-    forward = target - camera_pos
-    forward /= np.linalg.norm(forward)
-    right = np.cross(forward, up)
-    right /= np.linalg.norm(right)
-    new_up = np.cross(right, forward)
-    new_up /= np.linalg.norm(new_up)
-    cam2world = np.eye(4)
-    cam2world[:3, :3] = np.array([right, new_up, -forward]).T
-    cam2world[:3, 3] = camera_pos
-    return cam2world
-
-
-def convert_opengl_to_blender(camera_matrix):
-    if isinstance(camera_matrix, np.ndarray):
-        # Construct transformation matrix to convert from OpenGL space to Blender space
-        flip_yz = np.array([[1, 0, 0, 0], [0, 0, -1, 0], [0, 1, 0, 0], [0, 0, 0, 1]])
-        camera_matrix_blender = np.dot(flip_yz, camera_matrix)
-    else:
-        # Construct transformation matrix to convert from OpenGL space to Blender space
-        flip_yz = torch.tensor(
-            [[1, 0, 0, 0], [0, 0, -1, 0], [0, 1, 0, 0], [0, 0, 0, 1]]
-        )
-        if camera_matrix.ndim == 3:
-            flip_yz = flip_yz.unsqueeze(0)
-        camera_matrix_blender = torch.matmul(flip_yz.to(camera_matrix), camera_matrix)
-    return camera_matrix_blender
-
-
-def normalize_camera(camera_matrix):
-    """normalize the camera location onto a unit-sphere"""
-    if isinstance(camera_matrix, np.ndarray):
-        camera_matrix = camera_matrix.reshape(-1, 4, 4)
-        translation = camera_matrix[:, :3, 3]
-        translation = translation / (
-            np.linalg.norm(translation, axis=1, keepdims=True) + 1e-8
-        )
-        camera_matrix[:, :3, 3] = translation
-    else:
-        camera_matrix = camera_matrix.reshape(-1, 4, 4)
-        translation = camera_matrix[:, :3, 3]
-        translation = translation / (
-            torch.norm(translation, dim=1, keepdim=True) + 1e-8
-        )
-        camera_matrix[:, :3, 3] = translation
-    return camera_matrix.reshape(-1, 16)
-
-
-def get_camera(
-    num_frames, 
-    elevation=15, 
-    azimuth_start=0, 
-    azimuth_span=360, 
-    blender_coord=True,
-    extra_view=False,
-):
-    angle_gap = azimuth_span / num_frames
-    cameras = []
-    for azimuth in np.arange(azimuth_start, azimuth_span + azimuth_start, angle_gap):
-        camera_matrix = create_camera_to_world_matrix(elevation, azimuth)
-        if blender_coord:
-            camera_matrix = convert_opengl_to_blender(camera_matrix)
-        cameras.append(camera_matrix.flatten())
-        
-    if extra_view:
-        dim = len(cameras[0])
-        cameras.append(np.zeros(dim))  
-    return torch.tensor(np.stack(cameras, 0)).float()
-
-
-def get_camera_for_index(data_index):
-    """
-    按照当前我们的数据格式, 以000为正对我们的情况:
-    000是正面, ev: 0, azimuth: 0
-    001是左边, ev: 0, azimuth: -90
-    002是下面, ev: -90, azimuth: 0
-    003是背面, ev: 0, azimuth: 180
-    004是右边, ev: 0, azimuth: 90
-    005是上面, ev: 90, azimuth: 0
-    """
-    params = [(0, 0), (0, -90), (-90, 0), (0, 180), (0, 90), (90, 0)]
-    return get_camera(1, *params[data_index])

apps/third_party/CRM/imagedream/configs/sd_v2_base_ipmv.yaml

@@ -1,61 +0,0 @@
-model:
-  target: imagedream.ldm.interface.LatentDiffusionInterface
-  params:
-    linear_start: 0.00085
-    linear_end: 0.0120
-    timesteps: 1000
-    scale_factor: 0.18215
-    parameterization: "eps"
-
-    unet_config:
-      target: imagedream.ldm.modules.diffusionmodules.openaimodel.MultiViewUNetModel
-      params:
-        image_size: 32 # unused
-        in_channels: 4
-        out_channels: 4
-        model_channels: 320
-        attention_resolutions: [ 4, 2, 1 ]
-        num_res_blocks: 2
-        channel_mult: [ 1, 2, 4, 4 ]
-        num_head_channels: 64 # need to fix for flash-attn
-        use_spatial_transformer: True
-        use_linear_in_transformer: True
-        transformer_depth: 1
-        context_dim: 1024
-        use_checkpoint: False
-        legacy: False
-        camera_dim: 16
-        with_ip: True
-        ip_dim: 16 # ip token length
-        ip_mode: "local_resample"
-
-    vae_config:
-      target: imagedream.ldm.models.autoencoder.AutoencoderKL
-      params:
-        embed_dim: 4
-        monitor: val/rec_loss
-        ddconfig:
-          #attn_type: "vanilla-xformers"
-          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
-        lossconfig:
-          target: torch.nn.Identity
-
-    clip_config:
-      target: imagedream.ldm.modules.encoders.modules.FrozenOpenCLIPEmbedder
-      params:
-        freeze: True
-        layer: "penultimate"
-        ip_mode: "local_resample"

apps/third_party/CRM/imagedream/configs/sd_v2_base_ipmv_ch8.yaml

@@ -1,61 +0,0 @@
-model:
-  target: imagedream.ldm.interface.LatentDiffusionInterface
-  params:
-    linear_start: 0.00085
-    linear_end: 0.0120
-    timesteps: 1000
-    scale_factor: 0.18215
-    parameterization: "eps"
-
-    unet_config:
-      target: imagedream.ldm.modules.diffusionmodules.openaimodel.MultiViewUNetModel
-      params:
-        image_size: 32 # unused
-        in_channels: 8
-        out_channels: 8
-        model_channels: 320
-        attention_resolutions: [ 4, 2, 1 ]
-        num_res_blocks: 2
-        channel_mult: [ 1, 2, 4, 4 ]
-        num_head_channels: 64 # need to fix for flash-attn
-        use_spatial_transformer: True
-        use_linear_in_transformer: True
-        transformer_depth: 1
-        context_dim: 1024
-        use_checkpoint: False
-        legacy: False
-        camera_dim: 16
-        with_ip: True
-        ip_dim: 16 # ip token length
-        ip_mode: "local_resample"
-
-    vae_config:
-      target: imagedream.ldm.models.autoencoder.AutoencoderKL
-      params:
-        embed_dim: 4
-        monitor: val/rec_loss
-        ddconfig:
-          #attn_type: "vanilla-xformers"
-          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
-        lossconfig:
-          target: torch.nn.Identity
-
-    clip_config:
-      target: imagedream.ldm.modules.encoders.modules.FrozenOpenCLIPEmbedder
-      params:
-        freeze: True
-        layer: "penultimate"
-        ip_mode: "local_resample"