Update app.py

app.py

@@ -1,35 +1,60 @@
-import gradio as gr 
+import gradio as gr
 import spaces
+from gradio_litmodel3d import LitModel3D
+
 import os
 import shutil
-import random
 os.environ['SPCONV_ALGO'] = 'native'
 from typing import *
 import torch
 import numpy as np
 import imageio
-import cv2
-import torchvision.transforms.functional as TF
-from gradio_litmodel3d import LitModel3D
 from easydict import EasyDict as edict
 from PIL import Image, ImageOps
 from trellis.pipelines import TrellisImageTo3DPipeline
 from trellis.representations import Gaussian, MeshExtractResult
 from trellis.utils import render_utils, postprocessing_utils
+
+import os
+import random
+import torch
+import torchvision.transforms.functional as TF
+
 from diffusers import ControlNetModel, StableDiffusionXLControlNetPipeline, AutoencoderKL
 from diffusers import DDIMScheduler, EulerAncestralDiscreteScheduler
 from controlnet_aux import PidiNetDetector, HEDdetector
 from diffusers.utils import load_image
 from huggingface_hub import HfApi
 from pathlib import Path
+from PIL import Image, ImageOps
+import torch
+import numpy as np
+import cv2
+import os
+import random
 from gradio_imageslider import ImageSlider
 
 style_list = [
+    {
+        "name": "(No style)",
+        "prompt": "{prompt}",
+        "negative_prompt": "longbody, lowres, bad anatomy, bad hands, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality",
+    },
+    {
+        "name": "Cinematic",
+        "prompt": "cinematic still {prompt} . emotional, harmonious, vignette, highly detailed, high budget, bokeh, cinemascope, moody, epic, gorgeous, film grain, grainy",
+        "negative_prompt": "anime, cartoon, graphic, text, painting, crayon, graphite, abstract, glitch, deformed, mutated, ugly, disfigured",
+    },
     {
         "name": "3D Model",
         "prompt": "professional 3d model {prompt} . octane render, highly detailed, volumetric, dramatic lighting",
         "negative_prompt": "ugly, deformed, noisy, low poly, blurry, painting",
     },
+    {
+        "name": "Anime",
+        "prompt": "anime artwork {prompt} . anime style, key visual, vibrant, studio anime,  highly detailed",
+        "negative_prompt": "photo, deformed, black and white, realism, disfigured, low contrast",
+    },
 ]
 styles = {k["name"]: (k["prompt"], k["negative_prompt"]) for k in style_list}
 STYLE_NAMES = list(styles.keys())
@@ -50,36 +75,31 @@ def start_session(req: gr.Request):
     user_dir = os.path.join(TMP_DIR, str(req.session_hash))
     os.makedirs(user_dir, exist_ok=True)
     
+    
 def end_session(req: gr.Request):
     user_dir = os.path.join(TMP_DIR, str(req.session_hash))
     shutil.rmtree(user_dir)
 
 @spaces.GPU
-def preprocess_image(
-    image: Image.Image,
-    prompt: str = "",
-    negative_prompt: str = "",
-    style_name: str = "",
-    num_steps: int = 25,
-    guidance_scale: float = 5,
-    controlnet_conditioning_scale: float = 1.0,
-    req: gr.Request = None  
-) -> Tuple[Image.Image, Image.Image]:
-    user_dir = os.path.join(TMP_DIR, str(req.session_hash))
-    os.makedirs(user_dir, exist_ok=True)  
-
-    width, height = image['composite'].size
+def preprocess_image(image: Image.Image, 
+                    prompt: str = "",
+                    negative_prompt: str = "",
+                    style_name: str = "",
+                    num_steps: int = 25,
+                    guidance_scale: float = 5,
+                    controlnet_conditioning_scale: float = 1.0,
+                    ) -> Image.Image:
+    width, height  = image['composite'].size
     ratio = np.sqrt(1024. * 1024. / (width * height))
     new_width, new_height = int(width * ratio), int(height * ratio)
     image = image['composite'].resize((new_width, new_height))
     image = ImageOps.invert(image)
 
-    print("image:", type(image))
+    print("image:",type(image))
 
     prompt, negative_prompt = apply_style(style_name, prompt, negative_prompt)
-
+    
     print("params:", prompt, negative_prompt, style_name, num_steps, guidance_scale, controlnet_conditioning_scale)
-
     output = pipe_control(
         prompt=prompt,
         negative_prompt=negative_prompt,
@@ -88,15 +108,15 @@ def preprocess_image(
         controlnet_conditioning_scale=controlnet_conditioning_scale,
         guidance_scale=guidance_scale,
         width=new_width,
-        height=new_height
-    ).images[0]
-
-    processed_image_path = os.path.join(user_dir, 'processed_image.png')
-    output.save(processed_image_path)
-
+        height=new_height).images[0]
+    
     processed_image = pipeline.preprocess_image(output)
+    return (image, processed_image)
 
-    return image, processed_image
+def preprocess_images(images: List[Tuple[Image.Image, str]]) -> List[Image.Image]:
+    images = [image[0] for image in images]
+    processed_images = [pipeline.preprocess_image(image) for image in images]
+    return processed_images
 
 def pack_state(gs: Gaussian, mesh: MeshExtractResult) -> dict:
     return {
@@ -142,28 +162,48 @@ def get_seed(randomize_seed: bool, seed: int) -> int:
 @spaces.GPU
 def image_to_3d(
     image: Image.Image,
+    multiimages: List[Tuple[Image.Image, str]],
+    is_multiimage: bool,
     seed: int,
     ss_guidance_strength: float,
     ss_sampling_steps: int,
     slat_guidance_strength: float,
     slat_sampling_steps: int,
+    multiimage_algo: Literal["multidiffusion", "stochastic"],
     req: gr.Request,
 ) -> Tuple[dict, str]:
     user_dir = os.path.join(TMP_DIR, str(req.session_hash))
-    outputs = pipeline.run(
-        image[1],
-        seed=seed,
-        formats=["gaussian", "mesh"],
-        preprocess_image=False,
-        sparse_structure_sampler_params={
-            "steps": ss_sampling_steps,
-            "cfg_strength": ss_guidance_strength,
-        },
-        slat_sampler_params={
-            "steps": slat_sampling_steps,
-            "cfg_strength": slat_guidance_strength,
-        },
-    )
+    if not is_multiimage:
+        outputs = pipeline.run(
+            image[1],
+            seed=seed,
+            formats=["gaussian", "mesh"],
+            preprocess_image=False,
+            sparse_structure_sampler_params={
+                "steps": ss_sampling_steps,
+                "cfg_strength": ss_guidance_strength,
+            },
+            slat_sampler_params={
+                "steps": slat_sampling_steps,
+                "cfg_strength": slat_guidance_strength,
+            },
+        )
+    else:
+        outputs = pipeline.run_multi_image(
+            [image[0] for image in multiimages],
+            seed=seed,
+            formats=["gaussian", "mesh"],
+            preprocess_image=False,
+            sparse_structure_sampler_params={
+                "steps": ss_sampling_steps,
+                "cfg_strength": ss_guidance_strength,
+            },
+            slat_sampler_params={
+                "steps": slat_sampling_steps,
+                "cfg_strength": slat_guidance_strength,
+            },
+            mode=multiimage_algo,
+        )
     video = render_utils.render_video(outputs['gaussian'][0], num_frames=120)['color']
     video_geo = render_utils.render_video(outputs['mesh'][0], num_frames=120)['normal']
     video = [np.concatenate([video[i], video_geo[i]], axis=1) for i in range(len(video))]
@@ -200,7 +240,38 @@ def extract_gaussian(state: dict, req: gr.Request) -> Tuple[str, str]:
     torch.cuda.empty_cache()
     return gaussian_path, gaussian_path
 
+def prepare_multi_example() -> List[Image.Image]:
+    multi_case = list(set([i.split('_')[0] for i in os.listdir("assets/example_multi_image")]))
+    images = []
+    for case in multi_case:
+        _images = []
+        for i in range(1, 4):
+            img = Image.open(f'assets/example_multi_image/{case}_{i}.png')
+            W, H = img.size
+            img = img.resize((int(W / H * 512), 512))
+            _images.append(np.array(img))
+        images.append(Image.fromarray(np.concatenate(_images, axis=1)))
+    return images
+
+def split_image(image: Image.Image) -> List[Image.Image]:
+    image = np.array(image)
+    alpha = image[..., 3]
+    alpha = np.any(alpha>0, axis=0)
+    start_pos = np.where(~alpha[:-1] & alpha[1:])[0].tolist()
+    end_pos = np.where(alpha[:-1] & ~alpha[1:])[0].tolist()
+    images = []
+    for s, e in zip(start_pos, end_pos):
+        images.append(Image.fromarray(image[:, s:e+1]))
+    return [preprocess_image(image) for image in images]
+
 with gr.Blocks(delete_cache=(600, 600)) as demo:
+    gr.Markdown("""
+    ## Sketch to 3D with TRELLIS
+    1. Fast sketch to image with SDXL Flash, using [@xinsir](https://huggingface.co/xinsir) [scribble sdxl controlnet](https://huggingface.co/xinsir/controlnet-scribble-sdxl-1.0) and [sdxl flash](https://huggingface.co/sd-community/sdxl-flash)
+    2. Scalable and versatile image to 3D generation using [TRELLIS](https://trellis3d.github.io/)
+    ### 🎨🖌️ draw or upload a sketch and click "Generate" to create a 3D asset ✨
+    
+    """)
     with gr.Row():
         with gr.Column():
             with gr.Column():    
@@ -245,10 +316,14 @@ with gr.Blocks(delete_cache=(600, 600)) as demo:
                             ss_guidance_strength = gr.Slider(0.0, 10.0, label="Guidance Strength", value=7.5, step=0.1)
                             ss_sampling_steps = gr.Slider(1, 50, label="Sampling Steps", value=12, step=1)
                         gr.Markdown("Stage 2: Structured Latent Generation")
-                    with gr.Row():
+                        with gr.Row():
                             slat_guidance_strength = gr.Slider(0.0, 10.0, label="Guidance Strength", value=3.0, step=0.1)
                             slat_sampling_steps = gr.Slider(1, 50, label="Sampling Steps", value=12, step=1)
-                        
+                        multiimage_algo = gr.Radio(["stochastic", "multidiffusion"], label="Multi-image Algorithm", value="stochastic")
+                   
+            with gr.Tab(label="Multiple Images", id=1, visible=False) as multiimage_input_tab:
+                    multiimage_prompt = gr.Gallery(label="Image Prompt", format="png", type="pil", height=300, columns=3)
+            
             with gr.Accordion(label="GLB Extraction Settings", open=False):
                 mesh_simplify = gr.Slider(0.9, 0.98, label="Simplify", value=0.95, step=0.01)
                 texture_size = gr.Slider(512, 2048, label="Texture Size", value=1024, step=512)
@@ -269,6 +344,7 @@ with gr.Blocks(delete_cache=(600, 600)) as demo:
                 download_glb = gr.DownloadButton(label="Download GLB", interactive=False)
                 download_gs = gr.DownloadButton(label="Download Gaussian", interactive=False)  
     
+    is_multiimage = gr.State(False)
     do_preprocess = gr.State(True)
     output_buf = gr.State()
 
@@ -284,10 +360,23 @@ with gr.Blocks(delete_cache=(600, 600)) as demo:
             run_on_click=True,
             examples_per_page=64,
         )
+    with gr.Row(visible=False) as multiimage_example:
+        examples_multi = gr.Examples(
+            examples=prepare_multi_example(),
+            inputs=[image_prompt],
+            fn=split_image,
+            outputs=[multiimage_prompt],
+            run_on_click=True,
+            examples_per_page=8,
+        )
 
     demo.load(start_session)
     demo.unload(end_session)
     
+    multiimage_input_tab.select(
+        lambda: tuple([True, gr.Row.update(visible=False), gr.Row.update(visible=True)]),
+        outputs=[is_multiimage, single_image_example, multiimage_example]
+    )
     image_prompt.clear(
         fn=reset_canvas,
         outputs = [image_prompt]
@@ -302,6 +391,11 @@ with gr.Blocks(delete_cache=(600, 600)) as demo:
         inputs=[image_prompt, prompt, negative_prompt, style, num_steps, guidance_scale, controlnet_conditioning_scale],
         outputs=[image_prompt_processed],
     )
+    multiimage_prompt.upload(
+        preprocess_images,
+        inputs=[multiimage_prompt],
+        outputs=[multiimage_prompt],
+    )
 
     generate_btn.click(
         get_seed,
@@ -309,7 +403,7 @@ with gr.Blocks(delete_cache=(600, 600)) as demo:
         outputs=[seed],
     ).then(
         image_to_3d,
-        inputs=[image_prompt_processed, seed, ss_guidance_strength, ss_sampling_steps, slat_guidance_strength, slat_sampling_steps],
+        inputs=[image_prompt_processed, multiimage_prompt, is_multiimage, seed, ss_guidance_strength, ss_sampling_steps, slat_guidance_strength, slat_sampling_steps, multiimage_algo],
         outputs=[output_buf, video_output],
     ).then(
         lambda: tuple([gr.Button(interactive=True), gr.Button(interactive=True)]),
@@ -343,15 +437,13 @@ with gr.Blocks(delete_cache=(600, 600)) as demo:
         lambda: gr.Button(interactive=False),
         outputs=[download_glb],
     )
-    
-# Launch the Gradio app
+
 if __name__ == "__main__":
-    pipeline = TrellisImageTo3DPipeline.from_pretrained("cavargas10/TRELLIS")
+    pipeline = TrellisImageTo3DPipeline.from_pretrained("JeffreyXiang/TRELLIS-image-large")
     pipeline.cuda()
 
     device = "cuda" if torch.cuda.is_available() else "cpu"
     
-    #scribble controlnet
     controlnet = ControlNetModel.from_pretrained(
     "xinsir/controlnet-scribble-sdxl-1.0",
     torch_dtype=torch.float16
@@ -371,4 +463,4 @@ if __name__ == "__main__":
         pipeline.preprocess_image(Image.fromarray(np.zeros((512, 512, 3), dtype=np.uint8)))    # Preload rembg
     except:
         pass
-    demo.launch(show_error=True)
+    demo.launch()