upgrade diffusers to use core marigold pipleines

change video tab to use vae-lite and relax time and fps
change 3d tab to store in obj format and zip up everything

README.md

@@ -4,19 +4,19 @@ emoji: 🏵️
 colorFrom: blue
 colorTo: red
 sdk: gradio
-sdk_version: 4.21.0
+sdk_version: 4.32.2
 app_file: app.py
 pinned: true
 license: cc-by-sa-4.0
 models:
-- prs-eth/marigold-lcm-v1-0
+- prs-eth/marigold-depth-lcm-v1-0
 hf_oauth: true
 hf_oauth_expiration_minutes: 43200
 ---
 
 This is a demo of Marigold-LCM, the state-of-the-art depth estimator for images in the wild.
 It combines the power of the original Marigold 10-step estimator and the Latent Consistency Models, delivering high-quality results in as little as one step.
-Find out more in our CVPR 2024 paper titled ["Repurposing Diffusion-Based Image Generators for Monocular Depth Estimation"](https://arxiv.org/abs/2312.02145)
+Find out more in our CVPR 2024 Oral paper titled ["Repurposing Diffusion-Based Image Generators for Monocular Depth Estimation"](https://arxiv.org/abs/2312.02145)
 
 ```
 @InProceedings{ke2023repurposing,

app.py

@@ -25,12 +25,14 @@ import warnings
 import zipfile
 from io import BytesIO
 
+import diffusers
 import gradio as gr
 import imageio as imageio
 import numpy as np
 import spaces
 import torch as torch
 from PIL import Image
+from diffusers import MarigoldDepthPipeline
 from gradio_imageslider import ImageSlider
 from huggingface_hub import login
 from tqdm import tqdm
@@ -38,32 +40,31 @@ from tqdm import tqdm
 from extrude import extrude_depth_3d
 from gradio_patches.examples import Examples
 from gradio_patches.flagging import FlagMethod, HuggingFaceDatasetSaver
-from marigold_depth_estimation_lcm import MarigoldDepthConsistencyPipeline
 
 warnings.filterwarnings(
     "ignore", message=".*LoginButton created outside of a Blocks context.*"
 )
 
 default_seed = 2024
+default_batch_size = 4
 
-default_image_denoise_steps = 4
+default_image_num_inference_steps = 4
 default_image_ensemble_size = 1
-default_image_processing_res = 768
+default_image_processing_resolution = 768
 default_image_reproducuble = True
 
 default_video_depth_latent_init_strength = 0.1
-default_video_denoise_steps = 1
+default_video_num_inference_steps = 1
 default_video_ensemble_size = 1
-default_video_processing_res = 768
-default_video_out_fps = 15
-default_video_out_max_frames = 100
+default_video_processing_resolution = 768
+default_video_out_max_frames = 450
 
 default_bas_plane_near = 0.0
 default_bas_plane_far = 1.0
 default_bas_embossing = 20
-default_bas_denoise_steps = 4
+default_bas_num_inference_steps = 4
 default_bas_ensemble_size = 1
-default_bas_processing_res = 768
+default_bas_processing_resolution = 768
 default_bas_size_longest_px = 512
 default_bas_size_longest_cm = 10
 default_bas_filter_size = 3
@@ -85,9 +86,9 @@ def process_image_check(path_input):
 def process_image(
     pipe,
     path_input,
-    denoise_steps=default_image_denoise_steps,
+    num_inference_steps=default_image_num_inference_steps,
     ensemble_size=default_image_ensemble_size,
-    processing_res=default_image_processing_res,
+    processing_resolution=default_image_processing_resolution,
 ):
     name_base, name_ext = os.path.splitext(os.path.basename(path_input))
     print(f"Processing image {name_base}{name_ext}")
@@ -99,22 +100,23 @@ def process_image(
 
     input_image = Image.open(path_input)
 
+    generator = torch.Generator(device=pipe.device).manual_seed(default_seed)
+
     pipe_out = pipe(
         input_image,
-        denoising_steps=denoise_steps,
+        num_inference_steps=num_inference_steps,
         ensemble_size=ensemble_size,
-        processing_res=processing_res,
-        batch_size=1 if processing_res == 0 else 0,
-        seed=default_seed,
-        show_progress_bar=False,
+        processing_resolution=processing_resolution,
+        batch_size=1 if processing_resolution == 0 else default_batch_size,
+        generator=generator,
     )
 
-    depth_pred = pipe_out.depth_np
-    depth_colored = pipe_out.depth_colored
-    depth_16bit = (depth_pred * 65535.0).astype(np.uint16)
+    depth_pred = pipe_out.prediction[0, :, :, 0]
+    depth_colored = pipe.image_processor.visualize_depth(pipe_out.prediction)[0]
+    depth_16bit = pipe.image_processor.export_depth_to_16bit_png(pipe_out.prediction)[0]
 
     np.save(path_out_fp32, depth_pred)
-    Image.fromarray(depth_16bit).save(path_out_16bit, mode="I;16")
+    depth_16bit.save(path_out_16bit)
     depth_colored.save(path_out_vis)
 
     return (
@@ -127,10 +129,9 @@ def process_video(
     pipe,
     path_input,
     depth_latent_init_strength=default_video_depth_latent_init_strength,
-    denoise_steps=default_video_denoise_steps,
+    num_inference_steps=default_video_num_inference_steps,
     ensemble_size=default_video_ensemble_size,
-    processing_res=default_video_processing_res,
-    out_fps=default_video_out_fps,
+    processing_resolution=default_video_processing_resolution,
     out_max_frames=default_video_out_max_frames,
     progress=gr.Progress(),
 ):
@@ -146,79 +147,105 @@ def process_video(
     path_out_vis = os.path.join(path_output_dir, f"{name_base}_depth_colored.mp4")
     path_out_16bit = os.path.join(path_output_dir, f"{name_base}_depth_16bit.zip")
 
-    reader = imageio.get_reader(path_input)
+    generator = torch.Generator(device=pipe.device).manual_seed(default_seed)
 
-    meta_data = reader.get_meta_data()
-    fps = meta_data["fps"]
-    size = meta_data["size"]
-    duration_sec = meta_data["duration"]
+    reader, writer, zipf = None, None, None
+    try:
+        pipe.vae, pipe.vae_tiny = pipe.vae_tiny, pipe.vae
+
+        reader = imageio.get_reader(path_input)
+
+        meta_data = reader.get_meta_data()
+        fps = meta_data["fps"]
+        size = meta_data["size"]
+        max_orig = max(size)
+        duration_sec = meta_data["duration"]
+        total_frames = int(fps * duration_sec)
+
+        out_duration_sec = out_max_frames / fps
+        if duration_sec > out_duration_sec:
+            gr.Warning(
+                f"Only the first ~{int(out_duration_sec)} seconds will be processed; "
+                f"use alternative setups such as ComfyUI Marigold node for full processing"
+            )
 
-    if fps <= out_fps:
-        frame_interval, out_fps = 1, fps
-    else:
-        frame_interval = round(fps / out_fps)
-        out_fps = fps / frame_interval
+        writer = imageio.get_writer(path_out_vis, fps=fps)
 
-    out_duration_sec = out_max_frames / out_fps
-    if duration_sec > out_duration_sec:
-        gr.Warning(
-            f"Only the first ~{int(out_duration_sec)} seconds will be processed; "
-            f"use alternative setups for full processing"
-        )
+        zipf = zipfile.ZipFile(path_out_16bit, "w", zipfile.ZIP_DEFLATED)
 
-    writer = imageio.get_writer(path_out_vis, fps=out_fps)
-    zipf = zipfile.ZipFile(path_out_16bit, "w", zipfile.ZIP_DEFLATED)
-    prev_depth_latent = None
-
-    pbar = tqdm(desc="Processing Video", total=out_max_frames)
-
-    out_frame_id = 0
-    for frame_id, frame in enumerate(reader):
-        if not (frame_id % frame_interval == 0):
-            continue
-        out_frame_id += 1
-        pbar.update(1)
-        if out_frame_id > out_max_frames:
-            break
-
-        frame_pil = Image.fromarray(frame)
-
-        pipe_out = pipe(
-            frame_pil,
-            denoising_steps=denoise_steps,
-            ensemble_size=ensemble_size,
-            processing_res=processing_res,
-            match_input_res=False,
-            batch_size=0,
-            depth_latent_init=prev_depth_latent,
-            depth_latent_init_strength=depth_latent_init_strength,
-            return_depth_latent=True,
-            seed=default_seed,
-            show_progress_bar=False,
+        last_frame_latent = None
+        latent_common = torch.randn(
+            (
+                1,
+                4,
+                (768 * size[1] + 7 * max_orig) // (8 * max_orig),
+                (768 * size[0] + 7 * max_orig) // (8 * max_orig),
+            ),
+            generator=generator,
+            device=pipe.device,
+            dtype=torch.float16,
         )
 
-        prev_depth_latent = pipe_out.depth_latent
+        out_frame_id = 0
+        pbar = tqdm(desc="Processing Video", total=min(out_max_frames, total_frames))
+
+        for frame_id, frame in enumerate(reader):
+            out_frame_id += 1
+            pbar.update(1)
+            if out_frame_id > out_max_frames:
+                break
+
+            frame_pil = Image.fromarray(frame)
+
+            latents = latent_common
+            if last_frame_latent is not None:
+                assert (
+                    last_frame_latent.shape == latent_common.shape
+                ), f"{last_frame_latent.shape}, {latent_common.shape}"
+                latents = (
+                    1 - depth_latent_init_strength
+                ) * latents + depth_latent_init_strength * last_frame_latent
+
+            pipe_out = pipe(
+                frame_pil,
+                num_inference_steps=num_inference_steps,
+                ensemble_size=ensemble_size,
+                processing_resolution=processing_resolution,
+                match_input_resolution=False,
+                batch_size=1,
+                latents=latents,
+                output_latent=True,
+            )
 
-        processed_frame = pipe_out.depth_colored
-        processed_frame = imageio.core.util.Array(np.array(processed_frame))
-        writer.append_data(processed_frame)
+            last_frame_latent = pipe_out.latent
 
-        processed_frame = (65535 * np.clip(pipe_out.depth_np, 0.0, 1.0)).astype(
-            np.uint16
-        )
-        processed_frame = Image.fromarray(processed_frame, mode="I;16")
+            processed_frame = pipe.image_processor.visualize_depth(  # noqa
+                pipe_out.prediction
+            )[0]
+            processed_frame = imageio.core.util.Array(np.array(processed_frame))
+            writer.append_data(processed_frame)
 
-        archive_path = os.path.join(
-            f"{name_base}_depth_16bit", f"{out_frame_id:05d}.png"
-        )
-        img_byte_arr = BytesIO()
-        processed_frame.save(img_byte_arr, format="png")
-        img_byte_arr.seek(0)
-        zipf.writestr(archive_path, img_byte_arr.read())
+            archive_path = os.path.join(
+                f"{name_base}_depth_16bit", f"{out_frame_id:05d}.png"
+            )
+            img_byte_arr = BytesIO()
+            processed_frame = pipe.image_processor.export_depth_to_16bit_png(
+                pipe_out.prediction
+            )[0]
+            processed_frame.save(img_byte_arr, format="png")
+            img_byte_arr.seek(0)
+            zipf.writestr(archive_path, img_byte_arr.read())
+    finally:
+        if zipf is not None:
+            zipf.close()
+
+        if writer is not None:
+            writer.close()
 
-    reader.close()
-    writer.close()
-    zipf.close()
+        if reader is not None:
+            reader.close()
+
+        pipe.vae, pipe.vae_tiny = pipe.vae_tiny, pipe.vae
 
     return (
         path_out_vis,
@@ -232,9 +259,9 @@ def process_bas(
     plane_near=default_bas_plane_near,
     plane_far=default_bas_plane_far,
     embossing=default_bas_embossing,
-    denoise_steps=default_bas_denoise_steps,
+    num_inference_steps=default_bas_num_inference_steps,
     ensemble_size=default_bas_ensemble_size,
-    processing_res=default_bas_processing_res,
+    processing_resolution=default_bas_processing_resolution,
     size_longest_px=default_bas_size_longest_px,
     size_longest_cm=default_bas_size_longest_cm,
     filter_size=default_bas_filter_size,
@@ -257,16 +284,17 @@ def process_bas(
 
     input_image = Image.open(path_input)
 
+    generator = torch.Generator(device=pipe.device).manual_seed(default_seed)
+
     pipe_out = pipe(
         input_image,
-        denoising_steps=denoise_steps,
+        num_inference_steps=num_inference_steps,
         ensemble_size=ensemble_size,
-        processing_res=processing_res,
-        seed=default_seed,
-        show_progress_bar=False,
+        processing_resolution=processing_resolution,
+        generator=generator,
     )
 
-    depth_pred = pipe_out.depth_np * 65535
+    depth_pred = pipe_out.prediction[0, :, :, 0] * 65535
 
     def _process_3d(
         size_longest_px,
@@ -275,6 +303,7 @@ def process_bas(
         scene_lights,
         output_model_scale=None,
         prepare_for_3d_printing=False,
+        zip_outputs=False,
     ):
         image_rgb_w, image_rgb_h = input_image.width, input_image.height
         image_rgb_d = max(image_rgb_w, image_rgb_h)
@@ -294,7 +323,7 @@ def process_bas(
             (image_new_w, image_new_h), Image.BILINEAR
         ).convert("I").save(image_depth_new)
 
-        path_glb, path_stl = extrude_depth_3d(
+        path_glb, path_stl, path_obj = extrude_depth_3d(
             image_rgb_new,
             image_depth_new,
             output_model_scale=(
@@ -312,22 +341,24 @@ def process_bas(
             vertex_colors=vertex_colors,
             scene_lights=scene_lights,
             prepare_for_3d_printing=prepare_for_3d_printing,
+            zip_outputs=zip_outputs,
         )
 
-        return path_glb, path_stl
+        return path_glb, path_stl, path_obj
 
-    path_viewer_glb, _ = _process_3d(
+    path_viewer_glb, _, _ = _process_3d(
         256, filter_size, vertex_colors=False, scene_lights=True, output_model_scale=1
     )
-    path_files_glb, path_files_stl = _process_3d(
+    path_files_glb, path_files_stl, path_files_obj = _process_3d(
         size_longest_px,
         filter_size,
         vertex_colors=True,
         scene_lights=False,
         prepare_for_3d_printing=True,
+        zip_outputs=True,
     )
 
-    return path_viewer_glb, [path_files_glb, path_files_stl]
+    return path_viewer_glb, [path_files_glb, path_files_stl, path_files_obj]
 
 
 def run_demo_server(pipe, hf_writer=None):
@@ -461,12 +492,12 @@ def run_demo_server(pipe, hf_writer=None):
                             )
                             image_reset_btn = gr.Button(value="Reset")
                         with gr.Accordion("Advanced options", open=False):
-                            image_denoise_steps = gr.Slider(
+                            image_num_inference_steps = gr.Slider(
                                 label="Number of denoising steps",
                                 minimum=1,
                                 maximum=4,
                                 step=1,
-                                value=default_image_denoise_steps,
+                                value=default_image_num_inference_steps,
                             )
                             image_ensemble_size = gr.Slider(
                                 label="Ensemble size",
@@ -475,13 +506,13 @@ def run_demo_server(pipe, hf_writer=None):
                                 step=1,
                                 value=default_image_ensemble_size,
                             )
-                            image_processing_res = gr.Radio(
+                            image_processing_resolution = gr.Radio(
                                 [
                                     ("Native", 0),
                                     ("Recommended", 768),
                                 ],
                                 label="Processing resolution",
-                                value=default_image_processing_res,
+                                value=default_image_processing_resolution,
                             )
                     with gr.Column():
                         image_output_slider = ImageSlider(
@@ -656,12 +687,12 @@ def run_demo_server(pipe, hf_writer=None):
                         with gr.Accordion(
                             "3D printing demo: Advanced options", open=False
                         ):
-                            bas_denoise_steps = gr.Slider(
+                            bas_num_inference_steps = gr.Slider(
                                 label="Number of denoising steps",
                                 minimum=1,
                                 maximum=4,
                                 step=1,
-                                value=default_bas_denoise_steps,
+                                value=default_bas_num_inference_steps,
                             )
                             bas_ensemble_size = gr.Slider(
                                 label="Ensemble size",
@@ -670,13 +701,13 @@ def run_demo_server(pipe, hf_writer=None):
                                 step=1,
                                 value=default_bas_ensemble_size,
                             )
-                            bas_processing_res = gr.Radio(
+                            bas_processing_resolution = gr.Radio(
                                 [
                                     ("Native", 0),
                                     ("Recommended", 768),
                                 ],
                                 label="Processing resolution",
-                                value=default_bas_processing_res,
+                                value=default_bas_processing_resolution,
                             )
                             bas_size_longest_px = gr.Slider(
                                 label="Size (px) of the longest side",
@@ -740,9 +771,9 @@ def run_demo_server(pipe, hf_writer=None):
                             0.0,  # plane_near
                             0.66,  # plane_far
                             15,  # embossing
-                            4,  # denoise_steps
+                            4,  # num_inference_steps
                             4,  # ensemble_size
-                            768,  # processing_res
+                            768,  # processing_resolution
                             512,  # size_longest_px
                             10,  # size_longest_cm
                             3,  # filter_size
@@ -755,14 +786,14 @@ def run_demo_server(pipe, hf_writer=None):
                             0.0,  # plane_near
                             0.5,  # plane_far
                             50,  # embossing
-                            2,  # denoise_steps
+                            2,  # num_inference_steps
                             1,  # ensemble_size
-                            768,  # processing_res
+                            768,  # processing_resolution
                             512,  # size_longest_px
                             10,  # size_longest_cm
                             3,  # filter_size
                             5,  # frame_thickness
-                            -15,  # frame_near
+                            -25,  # frame_near
                             1,  # frame_far
                         ],
                         [
@@ -770,9 +801,9 @@ def run_demo_server(pipe, hf_writer=None):
                             0.0,  # plane_near
                             1.0,  # plane_far
                             20,  # embossing
-                            2,  # denoise_steps
+                            2,  # num_inference_steps
                             4,  # ensemble_size
-                            768,  # processing_res
+                            768,  # processing_resolution
                             512,  # size_longest_px
                             10,  # size_longest_cm
                             3,  # filter_size
@@ -786,9 +817,9 @@ def run_demo_server(pipe, hf_writer=None):
                         bas_plane_near,
                         bas_plane_far,
                         bas_embossing,
-                        bas_denoise_steps,
+                        bas_num_inference_steps,
                         bas_ensemble_size,
-                        bas_processing_res,
+                        bas_processing_resolution,
                         bas_size_longest_px,
                         bas_size_longest_cm,
                         bas_filter_size,
@@ -836,9 +867,9 @@ def run_demo_server(pipe, hf_writer=None):
                 fn=process_pipe_image,
                 inputs=[
                     image_input,
-                    image_denoise_steps,
+                    image_num_inference_steps,
                     image_ensemble_size,
-                    image_processing_res,
+                    image_processing_resolution,
                 ],
                 outputs=[image_output_slider, image_output_files],
                 concurrency_limit=1,
@@ -854,9 +885,9 @@ def run_demo_server(pipe, hf_writer=None):
                 fn=process_pipe_image,
                 inputs=[
                     image_input,
-                    image_denoise_steps,
+                    image_num_inference_steps,
                     image_ensemble_size,
-                    image_processing_res,
+                    image_processing_resolution,
                 ],
                 outputs=[image_output_slider, image_output_files],
                 concurrency_limit=1,
@@ -868,8 +899,8 @@ def run_demo_server(pipe, hf_writer=None):
                 None,
                 None,
                 default_image_ensemble_size,
-                default_image_denoise_steps,
-                default_image_processing_res,
+                default_image_num_inference_steps,
+                default_image_processing_resolution,
             ),
             inputs=[],
             outputs=[
@@ -877,8 +908,8 @@ def run_demo_server(pipe, hf_writer=None):
                 image_output_slider,
                 image_output_files,
                 image_ensemble_size,
-                image_denoise_steps,
-                image_processing_res,
+                image_num_inference_steps,
+                image_processing_resolution,
             ],
             queue=False,
         )
@@ -902,9 +933,9 @@ def run_demo_server(pipe, hf_writer=None):
         if hf_writer is not None:
             share_components = [
                 image_input,
-                image_denoise_steps,
+                image_num_inference_steps,
                 image_ensemble_size,
-                image_processing_res,
+                image_processing_resolution,
                 image_output_slider,
                 share_content_is_legal,
                 share_transfer_of_rights,
@@ -970,9 +1001,9 @@ def run_demo_server(pipe, hf_writer=None):
                 bas_plane_near,
                 bas_plane_far,
                 bas_embossing,
-                bas_denoise_steps,
+                bas_num_inference_steps,
                 bas_ensemble_size,
-                bas_processing_res,
+                bas_processing_resolution,
                 bas_size_longest_px,
                 bas_size_longest_cm,
                 bas_filter_size,
@@ -993,9 +1024,9 @@ def run_demo_server(pipe, hf_writer=None):
                 default_bas_plane_near,
                 default_bas_plane_far,
                 default_bas_embossing,
-                default_bas_denoise_steps,
+                default_bas_num_inference_steps,
                 default_bas_ensemble_size,
-                default_bas_processing_res,
+                default_bas_processing_resolution,
                 default_bas_size_longest_px,
                 default_bas_size_longest_cm,
                 default_bas_filter_size,
@@ -1012,9 +1043,9 @@ def run_demo_server(pipe, hf_writer=None):
                 bas_plane_near,
                 bas_plane_far,
                 bas_embossing,
-                bas_denoise_steps,
+                bas_num_inference_steps,
                 bas_ensemble_size,
-                bas_processing_res,
+                bas_processing_resolution,
                 bas_size_longest_px,
                 bas_size_longest_cm,
                 bas_filter_size,
@@ -1036,8 +1067,8 @@ def run_demo_server(pipe, hf_writer=None):
 
 
 def main():
-    CHECKPOINT = "prs-eth/marigold-lcm-v1-0"
-    CROWD_DATA = "crowddata-marigold-lcm-v1-0-space-v1-0"
+    CHECKPOINT = "prs-eth/marigold-depth-lcm-v1-0"
+    CROWD_DATA = "crowddata-marigold-depth-lcm-v1-0-space-v1-0"
 
     os.system("pip freeze")
 
@@ -1046,7 +1077,14 @@ def main():
 
     device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
 
-    pipe = MarigoldDepthConsistencyPipeline.from_pretrained(CHECKPOINT)
+    pipe = MarigoldDepthPipeline.from_pretrained(
+        CHECKPOINT, variant="fp16", torch_dtype=torch.float16
+    ).to(device)
+    pipe.vae_tiny = diffusers.AutoencoderTiny.from_pretrained(
+        "madebyollin/taesd", torch_dtype=torch.float16
+    ).to(device)
+    pipe.set_progress_bar_config(disable=True)
+
     try:
         import xformers
 
@@ -1054,12 +1092,10 @@ def main():
     except:
         pass  # run without xformers
 
-    pipe = pipe.to(device)
-
     hf_writer = None
-    if "HF_TOKEN_LOGIN" in os.environ:
+    if "HF_TOKEN_LOGIN_WRITE_CROWD" in os.environ:
         hf_writer = HuggingFaceDatasetSaver(
-            os.getenv("HF_TOKEN_LOGIN"),
+            os.getenv("HF_TOKEN_LOGIN_WRITE_CROWD"),
             CROWD_DATA,
             private=True,
             info_filename="dataset_info.json",

extrude.py

@@ -20,11 +20,13 @@
 
 import math
 import os
+import zipfile
 
 import numpy as np
 import pygltflib
 import trimesh
-from PIL import Image, ImageFilter
+from PIL import Image
+from scipy.ndimage import median_filter
 
 
 def quaternion_multiply(q1, q2):
@@ -103,6 +105,7 @@ def glb_add_lights(path_input, path_output):
 def extrude_depth_3d(
     path_rgb,
     path_depth,
+    path_out_base=None,
     output_model_scale=100,
     filter_size=3,
     coef_near=0.0,
@@ -114,6 +117,7 @@ def extrude_depth_3d(
     vertex_colors=True,
     scene_lights=True,
     prepare_for_3d_printing=False,
+    zip_outputs=False,
 ):
     f_far_inner = -emboss
     f_far_outer = f_far_inner - f_back
@@ -121,12 +125,11 @@ def extrude_depth_3d(
     f_near = max(f_near, f_far_inner)
 
     depth_image = Image.open(path_depth)
-    assert depth_image.mode == "I", depth_image.mode
-    depth_image = depth_image.filter(ImageFilter.MedianFilter(size=filter_size))
 
     w, h = depth_image.size
     d_max = max(w, h)
     depth_image = np.array(depth_image).astype(np.double)
+    depth_image = median_filter(depth_image, size=filter_size)
     z_min, z_max = np.min(depth_image), np.max(depth_image)
     depth_image = (depth_image.astype(np.double) - z_min) / (z_max - z_min)
     depth_image[depth_image < coef_near] = coef_near
@@ -341,14 +344,57 @@ def extrude_depth_3d(
         )
         mesh.apply_transform(rotation_mat)
 
-    path_out_base = os.path.splitext(path_depth)[0].replace("_16bit", "")
+    if path_out_base is None:
+        path_out_base = os.path.splitext(path_depth)[0].replace("_16bit", "")
     path_out_glb = path_out_base + ".glb"
     path_out_stl = path_out_base + ".stl"
+    path_out_obj = path_out_base + ".obj"
 
     mesh.export(path_out_glb, file_type="glb")
     if scene_lights:
         glb_add_lights(path_out_glb, path_out_glb)
-
     mesh.export(path_out_stl, file_type="stl")
-
-    return path_out_glb, path_out_stl
+    mesh.export(path_out_obj, file_type="obj")
+
+    if zip_outputs:
+        with zipfile.ZipFile(path_out_glb + ".zip", "w", zipfile.ZIP_DEFLATED) as zipf:
+            arcname = os.path.basename(os.path.splitext(path_out_glb)[0]) + ".glb"
+            zipf.write(path_out_glb, arcname=arcname)
+            path_out_glb = path_out_glb + ".zip"
+        with zipfile.ZipFile(path_out_stl + ".zip", "w", zipfile.ZIP_DEFLATED) as zipf:
+            arcname = os.path.basename(os.path.splitext(path_out_stl)[0]) + ".stl"
+            zipf.write(path_out_stl, arcname=arcname)
+            path_out_stl = path_out_stl + ".zip"
+        with zipfile.ZipFile(path_out_obj + ".zip", "w", zipfile.ZIP_DEFLATED) as zipf:
+            arcname = os.path.basename(os.path.splitext(path_out_obj)[0]) + ".obj"
+            zipf.write(path_out_obj, arcname=arcname)
+            path_out_obj = path_out_obj + ".zip"
+
+    return path_out_glb, path_out_stl, path_out_obj
+
+
+if __name__ == "__main__":
+    img_rgb = "files/basrelief/einstein.jpg"
+    img_depth = "gradio_cached_examples/examples_image/Depth outputs/54d74157894322bdc77c/einstein_depth_16bit.png"
+    Image.open(img_rgb).resize((512, 512), Image.LANCZOS).save(
+        "einstein_3d_tex_512.jpg"
+    )
+    Image.open(img_depth).convert(mode="F").resize((512, 512), Image.BILINEAR).convert(
+        "I"
+    ).save("einstein_3d_depth_512.png")
+    extrude_depth_3d(
+        "einstein_3d_tex_512.jpg",
+        "einstein_3d_depth_512.png",
+        path_out_base="einstein_3d_out",
+        output_model_scale=100,
+        filter_size=3,
+        coef_near=0.0,
+        coef_far=0.5,
+        emboss=0.5,
+        f_thic=0.05,
+        f_near=-0.25,
+        f_back=0.01,
+        vertex_colors=True,
+        scene_lights=True,
+        prepare_for_3d_printing=True,
+    )

gradio_cached_examples/examples_bas/3D model outputs high-res/{f4a69712004b188f4bb0/food_depth_512.glb → 0f57994f5d6ac12c1020/food_depth_512.glb.zip}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f0a1784597fc64d8e06cf549fef351b23de5f075e04758d8ed26b1e97aefdeb1
-size 7053032
+oid sha256:2343e8ab8c40539f3d4cdb6bde4af13964b989af33d1993eb2c3edd9e3822950
+size 2089619

gradio_cached_examples/examples_bas/3D model outputs high-res/0fbd5fe3c8b8dd4f05d1/einstein_depth_512.stl

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:2da8767fd0fafe985b0deeb832254b26e93cfdbdb0e7040aaa3233048c566227
-size 26419484

gradio_cached_examples/examples_bas/3D model outputs high-res/127d9bcaf03fa5f41dd3/food_depth_512.stl.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:95a5d9c1fd336e1e73ed3bcd388ec980da90b4ba601ce999839ab44baca5abb3
+size 6527220

gradio_cached_examples/examples_bas/3D model outputs high-res/49a7818bed0f17ce01cf/food_depth_512.stl

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:9accc5a6bada4c60a89263e885b2d11a03a9482447ede5ab5fa72458e2522137
-size 17630084

gradio_cached_examples/examples_bas/3D model outputs high-res/96a98e08d96fd47e5cc6/einstein_depth_512.obj.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a5b62efa853cbba11eba28f99307b40dfca300ec8047f2d778db92b3d3417d81
+size 5977474

gradio_cached_examples/examples_bas/3D model outputs high-res/a17995f3d4750a0e0bbc/food_depth_512.obj.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2fe92c1762da36b2d9fbe1419d5f91fc9cc64f739de6dc2b122fb1f1c6ca3e7e
+size 4044632

gradio_cached_examples/examples_bas/3D model outputs high-res/b0b93bdcbedf077307ba/coin_depth_512.stl.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac5e87a94275d6df74d5a7a42672cb97fbf29a3c2a6caf8dbc3b979641fea560
+size 7434540

gradio_cached_examples/examples_bas/3D model outputs high-res/b69b15d737b2de6938a5/einstein_depth_512.glb

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:aaf9049d52123504ee44e1af8426f9e081fc67450a3e7aefd7c8556e14ee7fee
-size 10568768

gradio_cached_examples/examples_bas/3D model outputs high-res/c0b00e36b9db31f4e9fd/coin_depth_512.stl

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:398abd4c61d86dc8bc5ab4ac5d6ed0aad50658512048176b647d8df72ae4ddf9
-size 26162484

gradio_cached_examples/examples_bas/3D model outputs high-res/c7499e9097e58b706e51/einstein_depth_512.glb.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e2b7641a5d61736e66868e895d1c97b0f15458d694e1164cc484bdea961a309f
+size 2977145

gradio_cached_examples/examples_bas/3D model outputs high-res/ebe8a8d03fbc1a1fc2bd/coin_depth_512.glb.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:abb6853f1086ad891ffd7407f3b913f2019f6c28256065f1ae94145c374ec220
+size 3521996

gradio_cached_examples/examples_bas/3D model outputs high-res/ee9ee048f590c0c9a2c8/einstein_depth_512.stl.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e5bf0df4560b958ab8f45cda1cd86a989d0506fbc1ab4c7ff37037a200ac4fd4
+size 8353042

gradio_cached_examples/examples_bas/3D model outputs high-res/ef353a417723b89274d9/coin_depth_512.glb

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:e2b98bc390dde108fc557d0a855c1ad9605e79ec6fde24122b74b63685ea5c45
-size 10465992

gradio_cached_examples/examples_bas/3D model outputs high-res/fbaa26ffc2eb3654c177/coin_depth_512.obj.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6bce06188d1a22e3a6b6b339cca62a2ea59da00925ea6c19bc580bf56250323a
+size 5631807

gradio_cached_examples/examples_bas/3D preview low-res relief highlight/{adb58f0f05db6046581f → 78ff2a583036eab8fe9b}/coin_depth_256.glb

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b589d787251bd3117f5eeecead011950015ba0f4b42dc3b26f8af8f52177b91a
-size 2369860
+oid sha256:1ae49b26ba4187dc3704cc00040c5671f1088a131f4796768fa86ec3eb67c1a1
+size 2369864

gradio_cached_examples/examples_bas/3D preview low-res relief highlight/{393dc8309ae2f31f47e9 → 8feb5fe1e8941c880c40}/food_depth_256.glb

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4956ff55699585fd7bd703785cc894f35ec6a6c19595b7e483ccecf1377203fd
-size 1598964
+oid sha256:fb7df74b49b34e916945c7bd31b45d6359db2ebb3d0bbb0e2fdd9d9222fb0816
+size 1598968

gradio_cached_examples/examples_bas/3D preview low-res relief highlight/{d2ca1f18e3da0b06142e → bb26fd8a9d7890806329}/einstein_depth_256.glb

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:18b05dfa7db4ce4fef14121a95abd57adcef5df5eeca23a44bc5f9e42bc3901f
-size 2397704
+oid sha256:88372996a41fccdc6e03b77edb2542fb75aaa1623b25c3cbf4a65de61131d0d3
+size 2397708

gradio_cached_examples/examples_bas/log.csv

@@ -1,4 +1,4 @@
 "3D preview (low-res, relief highlight)",3D model outputs (high-res),flag,username,timestamp
-"{""path"":""gradio_cached_examples/examples_bas/3D preview low-res relief highlight/adb58f0f05db6046581f/coin_depth_256.glb"",""url"":""/file=/tmp/gradio/784992f04cfa1fd8ac2d92cd4d58b141f3433c58/coin_depth_256.glb"",""size"":null,""orig_name"":""coin_depth_256.glb"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}}","[{""path"":""gradio_cached_examples/examples_bas/3D model outputs high-res/ef353a417723b89274d9/coin_depth_512.glb"",""url"":""/file=/tmp/gradio/b64f0ea070ce8490cd696bc76acb7bbf918d3183/coin_depth_512.glb"",""size"":10465992,""orig_name"":""coin_depth_512.glb"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}},{""path"":""gradio_cached_examples/examples_bas/3D model outputs high-res/c0b00e36b9db31f4e9fd/coin_depth_512.stl"",""url"":""/file=/tmp/gradio/15659bed2799af75f4eb1c16b4f9b0d2af113235/coin_depth_512.stl"",""size"":26162484,""orig_name"":""coin_depth_512.stl"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}}]",,,2024-04-03 18:29:48.978017
-"{""path"":""gradio_cached_examples/examples_bas/3D preview low-res relief highlight/d2ca1f18e3da0b06142e/einstein_depth_256.glb"",""url"":""/file=/tmp/gradio/59b193d76e6d22d077eceffbc7653b132c85b8b3/einstein_depth_256.glb"",""size"":null,""orig_name"":""einstein_depth_256.glb"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}}","[{""path"":""gradio_cached_examples/examples_bas/3D model outputs high-res/b69b15d737b2de6938a5/einstein_depth_512.glb"",""url"":""/file=/tmp/gradio/2495ecf670703eea800c67256f1c14df947c133c/einstein_depth_512.glb"",""size"":10568768,""orig_name"":""einstein_depth_512.glb"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}},{""path"":""gradio_cached_examples/examples_bas/3D model outputs high-res/0fbd5fe3c8b8dd4f05d1/einstein_depth_512.stl"",""url"":""/file=/tmp/gradio/324a085b1d9c80ff7084169f2fa9a88caf67084c/einstein_depth_512.stl"",""size"":26419484,""orig_name"":""einstein_depth_512.stl"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}}]",,,2024-04-03 18:29:51.356218
-"{""path"":""gradio_cached_examples/examples_bas/3D preview low-res relief highlight/393dc8309ae2f31f47e9/food_depth_256.glb"",""url"":""/file=/tmp/gradio/067615320ce5aadeb06abed3ca571ed420986df5/food_depth_256.glb"",""size"":null,""orig_name"":""food_depth_256.glb"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}}","[{""path"":""gradio_cached_examples/examples_bas/3D model outputs high-res/f4a69712004b188f4bb0/food_depth_512.glb"",""url"":""/file=/tmp/gradio/ed41062301027eb8fffedbecde30e5ce2e0123bc/food_depth_512.glb"",""size"":7053032,""orig_name"":""food_depth_512.glb"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}},{""path"":""gradio_cached_examples/examples_bas/3D model outputs high-res/49a7818bed0f17ce01cf/food_depth_512.stl"",""url"":""/file=/tmp/gradio/9eb6407ecff9a420865feebadad805e4c1bc23d8/food_depth_512.stl"",""size"":17630084,""orig_name"":""food_depth_512.stl"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}}]",,,2024-04-03 18:29:54.313883
+"{""path"": ""gradio_cached_examples/examples_bas/3D preview low-res relief highlight/78ff2a583036eab8fe9b/coin_depth_256.glb"", ""url"": ""/file=/tmp/gradio/75f3f8661319bbfa07a73e3c6aad5381af2bb662/coin_depth_256.glb"", ""size"": null, ""orig_name"": ""coin_depth_256.glb"", ""mime_type"": null, ""is_stream"": false, ""meta"": {""_type"": ""gradio.FileData""}}","[{""path"": ""gradio_cached_examples/examples_bas/3D model outputs high-res/ebe8a8d03fbc1a1fc2bd/coin_depth_512.glb.zip"", ""url"": ""/file=/tmp/gradio/8d501a8d9351d4ac105f330018213f88c8ab46e2/coin_depth_512.glb.zip"", ""size"": 3521996, ""orig_name"": ""coin_depth_512.glb.zip"", ""mime_type"": null, ""is_stream"": false, ""meta"": {""_type"": ""gradio.FileData""}}, {""path"": ""gradio_cached_examples/examples_bas/3D model outputs high-res/b0b93bdcbedf077307ba/coin_depth_512.stl.zip"", ""url"": ""/file=/tmp/gradio/67269f9efe556613199b0e4e587db71ecaf273e7/coin_depth_512.stl.zip"", ""size"": 7434540, ""orig_name"": ""coin_depth_512.stl.zip"", ""mime_type"": null, ""is_stream"": false, ""meta"": {""_type"": ""gradio.FileData""}}, {""path"": ""gradio_cached_examples/examples_bas/3D model outputs high-res/fbaa26ffc2eb3654c177/coin_depth_512.obj.zip"", ""url"": ""/file=/tmp/gradio/3e0dc5d754e113c9553b04644fe50279a65ad85d/coin_depth_512.obj.zip"", ""size"": 5631807, ""orig_name"": ""coin_depth_512.obj.zip"", ""mime_type"": null, ""is_stream"": false, ""meta"": {""_type"": ""gradio.FileData""}}]",,,2024-06-02 01:52:32.122923
+"{""path"": ""gradio_cached_examples/examples_bas/3D preview low-res relief highlight/bb26fd8a9d7890806329/einstein_depth_256.glb"", ""url"": ""/file=/tmp/gradio/d72948c76f50b644cfd2bc4f2eb68ec4d1dfd6b3/einstein_depth_256.glb"", ""size"": null, ""orig_name"": ""einstein_depth_256.glb"", ""mime_type"": null, ""is_stream"": false, ""meta"": {""_type"": ""gradio.FileData""}}","[{""path"": ""gradio_cached_examples/examples_bas/3D model outputs high-res/c7499e9097e58b706e51/einstein_depth_512.glb.zip"", ""url"": ""/file=/tmp/gradio/1d0bac2ed809c2c3d7ffecf0a40ca8f8398521da/einstein_depth_512.glb.zip"", ""size"": 2977145, ""orig_name"": ""einstein_depth_512.glb.zip"", ""mime_type"": null, ""is_stream"": false, ""meta"": {""_type"": ""gradio.FileData""}}, {""path"": ""gradio_cached_examples/examples_bas/3D model outputs high-res/ee9ee048f590c0c9a2c8/einstein_depth_512.stl.zip"", ""url"": ""/file=/tmp/gradio/3eb2701225d3b2362b3fc702c7b02bf2d9072308/einstein_depth_512.stl.zip"", ""size"": 8353042, ""orig_name"": ""einstein_depth_512.stl.zip"", ""mime_type"": null, ""is_stream"": false, ""meta"": {""_type"": ""gradio.FileData""}}, {""path"": ""gradio_cached_examples/examples_bas/3D model outputs high-res/96a98e08d96fd47e5cc6/einstein_depth_512.obj.zip"", ""url"": ""/file=/tmp/gradio/063a0b23c678e16f689b0bd8e3b6784e084a283f/einstein_depth_512.obj.zip"", ""size"": 5977474, ""orig_name"": ""einstein_depth_512.obj.zip"", ""mime_type"": null, ""is_stream"": false, ""meta"": {""_type"": ""gradio.FileData""}}]",,,2024-06-02 01:52:37.086710
+"{""path"": ""gradio_cached_examples/examples_bas/3D preview low-res relief highlight/8feb5fe1e8941c880c40/food_depth_256.glb"", ""url"": ""/file=/tmp/gradio/32d5f81d5bff33dcbf73bf5b7f46e23f9f73e0e3/food_depth_256.glb"", ""size"": null, ""orig_name"": ""food_depth_256.glb"", ""mime_type"": null, ""is_stream"": false, ""meta"": {""_type"": ""gradio.FileData""}}","[{""path"": ""gradio_cached_examples/examples_bas/3D model outputs high-res/0f57994f5d6ac12c1020/food_depth_512.glb.zip"", ""url"": ""/file=/tmp/gradio/c7790f4982c192c3fe02d99143d293d0c6c28bc3/food_depth_512.glb.zip"", ""size"": 2089619, ""orig_name"": ""food_depth_512.glb.zip"", ""mime_type"": null, ""is_stream"": false, ""meta"": {""_type"": ""gradio.FileData""}}, {""path"": ""gradio_cached_examples/examples_bas/3D model outputs high-res/127d9bcaf03fa5f41dd3/food_depth_512.stl.zip"", ""url"": ""/file=/tmp/gradio/cdfe46cad85481cdf6472efdddfdc6cca3fad3f4/food_depth_512.stl.zip"", ""size"": 6527220, ""orig_name"": ""food_depth_512.stl.zip"", ""mime_type"": null, ""is_stream"": false, ""meta"": {""_type"": ""gradio.FileData""}}, {""path"": ""gradio_cached_examples/examples_bas/3D model outputs high-res/a17995f3d4750a0e0bbc/food_depth_512.obj.zip"", ""url"": ""/file=/tmp/gradio/9f3cac3ebaef08375d5558783cb66009842b5a38/food_depth_512.obj.zip"", ""size"": 4044632, ""orig_name"": ""food_depth_512.obj.zip"", ""mime_type"": null, ""is_stream"": false, ""meta"": {""_type"": ""gradio.FileData""}}]",,,2024-06-02 01:52:42.365773

gradio_cached_examples/examples_video/Depth outputs/{207a3fa61126cf3d1981 → 41ffc6734af27742f841}/obama_depth_16bit.zip

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c42c9011121d927ca26e0abfc62f3724fbfc8a6848b522010f2e6ed98210a7ce
-size 38156752
+oid sha256:0e475e04f4b079263eefb9e90b01c303b02a361a927d9d8617b8c9875a43166a
+size 32516136

gradio_cached_examples/examples_video/Depth outputs/{93ea09fe13fa3ed86f2d → 4d0f30c84219f7e28c5b}/cab_depth_16bit.zip

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d5890dc53ca881d7327a9e70c8ac895ce2b21389af7da22d91060dd059663ab3
-size 39185578
+oid sha256:98662bffb5c1e4c7600e6b6bf05a66fc1e0fa393a95d6bee349076e4364c97cc
+size 41584240

gradio_cached_examples/examples_video/{Output video depth red-near blue-far/a9ebfdb1e7b3856d929a → Depth outputs/7e50f61bc97230e10bc9}/cab_depth_colored.mp4

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:81b3d4916e5b594174a49a7b064e4490b39dbed443f9d8ecea347c433284ff54
-size 473925
+oid sha256:a26735b01e362daef276832e1d6604a6dfbf7239ead96b98de490e39fca4dbc5
+size 672175

gradio_cached_examples/examples_video/Depth outputs/{0467a73b0d8d3ade99a7 → abdbb71b14bbbd9e179e}/elephant_depth_colored.mp4

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9dfb3689c97f04cabf39b4aa8a06ad4ce8fb424a353bf30d1641fb28f79c4d48
-size 755285
+oid sha256:81ffe1eaf172ce2767fae276c9ffb272ba858b540e50c326556bc80cc6b0e6bf
+size 950445

gradio_cached_examples/examples_video/Depth outputs/{93d8ed44918458cb56b7 → c3260a8a07bb43e647b8}/elephant_depth_16bit.zip

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:46f56972e6aced965f89c3c2057534fe88fb3898ae4f7fcd3af4e5cd9a8527d6
-size 41877515
+oid sha256:a4feebb5dc02d71b499a33bf18040c60dedf8563413087ee17d71f7edd394e0c
+size 50743181

gradio_cached_examples/examples_video/Depth outputs/{df8f2c089d19ebdb0eef → ffbf959599ab71f64cb8}/obama_depth_colored.mp4

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:86a007b675d5e294a7e8b036dea45a1034a0580f387353feec599920eb560ba7
-size 430863
+oid sha256:61a04be91e28ad207d3813b4cae689d9d2510adf88a9826bd6b55a7b93934422
+size 505939

gradio_cached_examples/examples_video/Output video depth red-near blue-far/{2301d8bcf6db4f0588c7 → 575c12e3b6ef94b4620e}/obama_depth_colored.mp4

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:86a007b675d5e294a7e8b036dea45a1034a0580f387353feec599920eb560ba7
-size 430863
+oid sha256:61a04be91e28ad207d3813b4cae689d9d2510adf88a9826bd6b55a7b93934422
+size 505939

gradio_cached_examples/examples_video/{Depth outputs/c12711ccd5a2c49cf18c → Output video depth red-near blue-far/6002bb1e4b17b4f06de1}/cab_depth_colored.mp4

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:81b3d4916e5b594174a49a7b064e4490b39dbed443f9d8ecea347c433284ff54
-size 473925
+oid sha256:a26735b01e362daef276832e1d6604a6dfbf7239ead96b98de490e39fca4dbc5
+size 672175

gradio_cached_examples/examples_video/Output video depth red-near blue-far/{89a2cce251acce733da7 → d0eb4f4347977106f1f2}/elephant_depth_colored.mp4

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9dfb3689c97f04cabf39b4aa8a06ad4ce8fb424a353bf30d1641fb28f79c4d48
-size 755285
+oid sha256:81ffe1eaf172ce2767fae276c9ffb272ba858b540e50c326556bc80cc6b0e6bf
+size 950445

gradio_cached_examples/examples_video/log.csv

@@ -1,4 +1,4 @@
 "Output video depth (red-near, blue-far)",Depth outputs,flag,username,timestamp
-"{""video"":{""path"":""gradio_cached_examples/examples_video/Output video depth red-near blue-far/a9ebfdb1e7b3856d929a/cab_depth_colored.mp4"",""url"":""/file=/tmp/gradio/baf61a4ecb34b14c80bd9dcc84bd792760ea4970/cab_depth_colored.mp4"",""size"":null,""orig_name"":""cab_depth_colored.mp4"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}},""subtitles"":null}","[{""path"":""gradio_cached_examples/examples_video/Depth outputs/c12711ccd5a2c49cf18c/cab_depth_colored.mp4"",""url"":""/file=/tmp/gradio/baf61a4ecb34b14c80bd9dcc84bd792760ea4970/cab_depth_colored.mp4"",""size"":473925,""orig_name"":""cab_depth_colored.mp4"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}},{""path"":""gradio_cached_examples/examples_video/Depth outputs/93ea09fe13fa3ed86f2d/cab_depth_16bit.zip"",""url"":""/file=/tmp/gradio/d12046d2cec66d2c8f5b425605842521d6dfcf16/cab_depth_16bit.zip"",""size"":39185578,""orig_name"":""cab_depth_16bit.zip"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}}]",,,2024-04-03 18:28:18.946260
-"{""video"":{""path"":""gradio_cached_examples/examples_video/Output video depth red-near blue-far/89a2cce251acce733da7/elephant_depth_colored.mp4"",""url"":""/file=/tmp/gradio/33743a70fe7f2a295fbfcea57161cfd65b59f525/elephant_depth_colored.mp4"",""size"":null,""orig_name"":""elephant_depth_colored.mp4"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}},""subtitles"":null}","[{""path"":""gradio_cached_examples/examples_video/Depth outputs/0467a73b0d8d3ade99a7/elephant_depth_colored.mp4"",""url"":""/file=/tmp/gradio/33743a70fe7f2a295fbfcea57161cfd65b59f525/elephant_depth_colored.mp4"",""size"":755285,""orig_name"":""elephant_depth_colored.mp4"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}},{""path"":""gradio_cached_examples/examples_video/Depth outputs/93d8ed44918458cb56b7/elephant_depth_16bit.zip"",""url"":""/file=/tmp/gradio/b6bee3a1450eb3169246820b83b4697fd13bdc5c/elephant_depth_16bit.zip"",""size"":41877515,""orig_name"":""elephant_depth_16bit.zip"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}}]",,,2024-04-03 18:28:58.771952
-"{""video"":{""path"":""gradio_cached_examples/examples_video/Output video depth red-near blue-far/2301d8bcf6db4f0588c7/obama_depth_colored.mp4"",""url"":""/file=/tmp/gradio/4fe8eb9aba002687469b3311ff39a0018692b10c/obama_depth_colored.mp4"",""size"":null,""orig_name"":""obama_depth_colored.mp4"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}},""subtitles"":null}","[{""path"":""gradio_cached_examples/examples_video/Depth outputs/df8f2c089d19ebdb0eef/obama_depth_colored.mp4"",""url"":""/file=/tmp/gradio/4fe8eb9aba002687469b3311ff39a0018692b10c/obama_depth_colored.mp4"",""size"":430863,""orig_name"":""obama_depth_colored.mp4"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}},{""path"":""gradio_cached_examples/examples_video/Depth outputs/207a3fa61126cf3d1981/obama_depth_16bit.zip"",""url"":""/file=/tmp/gradio/476f799e9f8c4000140a77684a2f34cbe86586f2/obama_depth_16bit.zip"",""size"":38156752,""orig_name"":""obama_depth_16bit.zip"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}}]",,,2024-04-03 18:29:43.295723
+"{""video"":{""path"":""gradio_cached_examples/examples_video/Output video depth red-near blue-far/6002bb1e4b17b4f06de1/cab_depth_colored.mp4"",""url"":""/file=/tmp/gradio/78d7a698603ea6a8b5e1a07b5613c044ee89b3cf/cab_depth_colored.mp4"",""size"":null,""orig_name"":""cab_depth_colored.mp4"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}},""subtitles"":null}","[{""path"":""gradio_cached_examples/examples_video/Depth outputs/7e50f61bc97230e10bc9/cab_depth_colored.mp4"",""url"":""/file=/tmp/gradio/78d7a698603ea6a8b5e1a07b5613c044ee89b3cf/cab_depth_colored.mp4"",""size"":551342,""orig_name"":""cab_depth_colored.mp4"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}},{""path"":""gradio_cached_examples/examples_video/Depth outputs/4d0f30c84219f7e28c5b/cab_depth_16bit.zip"",""url"":""/file=/tmp/gradio/4f2596bb97d63c19425cd4f24502bf4549549c68/cab_depth_16bit.zip"",""size"":41330021,""orig_name"":""cab_depth_16bit.zip"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}}]",,,2024-03-28 17:10:29.241744
+"{""video"":{""path"":""gradio_cached_examples/examples_video/Output video depth red-near blue-far/d0eb4f4347977106f1f2/elephant_depth_colored.mp4"",""url"":""/file=/tmp/gradio/e8f142b72d6cbe92e6f5a28df8e89b604dfa7138/elephant_depth_colored.mp4"",""size"":null,""orig_name"":""elephant_depth_colored.mp4"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}},""subtitles"":null}","[{""path"":""gradio_cached_examples/examples_video/Depth outputs/abdbb71b14bbbd9e179e/elephant_depth_colored.mp4"",""url"":""/file=/tmp/gradio/e8f142b72d6cbe92e6f5a28df8e89b604dfa7138/elephant_depth_colored.mp4"",""size"":846854,""orig_name"":""elephant_depth_colored.mp4"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}},{""path"":""gradio_cached_examples/examples_video/Depth outputs/c3260a8a07bb43e647b8/elephant_depth_16bit.zip"",""url"":""/file=/tmp/gradio/92a046ee5518a817550aab4a4cf07df5a1a71ad5/elephant_depth_16bit.zip"",""size"":42873909,""orig_name"":""elephant_depth_16bit.zip"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}}]",,,2024-03-28 17:11:09.183092
+"{""video"":{""path"":""gradio_cached_examples/examples_video/Output video depth red-near blue-far/575c12e3b6ef94b4620e/obama_depth_colored.mp4"",""url"":""/file=/tmp/gradio/429cce73a4a67c3c47ae93357bc8d0bea33789cb/obama_depth_colored.mp4"",""size"":null,""orig_name"":""obama_depth_colored.mp4"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}},""subtitles"":null}","[{""path"":""gradio_cached_examples/examples_video/Depth outputs/ffbf959599ab71f64cb8/obama_depth_colored.mp4"",""url"":""/file=/tmp/gradio/429cce73a4a67c3c47ae93357bc8d0bea33789cb/obama_depth_colored.mp4"",""size"":509224,""orig_name"":""obama_depth_colored.mp4"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}},{""path"":""gradio_cached_examples/examples_video/Depth outputs/41ffc6734af27742f841/obama_depth_16bit.zip"",""url"":""/file=/tmp/gradio/368fa8ae89ce36b3b6fb0d6044510e5cacef38fd/obama_depth_16bit.zip"",""size"":42499726,""orig_name"":""obama_depth_16bit.zip"",""mime_type"":null,""is_stream"":false,""meta"":{""_type"":""gradio.FileData""}}]",,,2024-03-28 17:11:54.147899

requirements.txt

@@ -1,10 +1,10 @@
-accelerate==0.25.0
+accelerate==0.30.1
 aiofiles==23.2.1
-aiohttp==3.9.3
+aiohttp==3.9.5
 aiosignal==1.3.1
 altair==5.3.0
-annotated-types==0.6.0
-anyio==4.3.0
+annotated-types==0.7.0
+anyio==4.4.0
 async-timeout==4.0.3
 attrs==23.2.0
 Authlib==1.3.0
@@ -12,115 +12,120 @@ certifi==2024.2.2
 cffi==1.16.0
 charset-normalizer==3.3.2
 click==8.0.4
-cmake==3.29.0.1
-contourpy==1.2.0
-cryptography==42.0.5
+contourpy==1.2.1
+cryptography==42.0.7
 cycler==0.12.1
-dataclasses-json==0.6.4
-datasets==2.18.0
+dataclasses-json==0.6.6
+datasets==2.19.1
 Deprecated==1.2.14
-diffusers==0.27.2
+diffusers==0.28.0
 dill==0.3.8
-exceptiongroup==1.2.0
-fastapi==0.110.0
+dnspython==2.6.1
+email_validator==2.1.1
+exceptiongroup==1.2.1
+fastapi==0.111.0
+fastapi-cli==0.0.4
 ffmpy==0.3.2
-filelock==3.13.3
-fonttools==4.50.0
+filelock==3.14.0
+fonttools==4.53.0
 frozenlist==1.4.1
-fsspec==2024.2.0
-gradio==4.21.0
-gradio_client==0.12.0
-gradio_imageslider==0.0.18
+fsspec==2024.3.1
+gradio==4.32.2
+gradio_client==0.17.0
+gradio_imageslider==0.0.20
 h11==0.14.0
 httpcore==1.0.5
+httptools==0.6.1
 httpx==0.27.0
-huggingface-hub==0.22.1
-idna==3.6
-imageio==2.34.0
-imageio-ffmpeg==0.4.9
+huggingface-hub==0.23.0
+idna==3.7
+imageio==2.34.1
+imageio-ffmpeg==0.5.0
 importlib_metadata==7.1.0
 importlib_resources==6.4.0
-itsdangerous==2.1.2
-Jinja2==3.1.3
-jsonschema==4.21.1
+itsdangerous==2.2.0
+Jinja2==3.1.4
+jsonschema==4.22.0
 jsonschema-specifications==2023.12.1
 kiwisolver==1.4.5
-lit==18.1.2
 markdown-it-py==3.0.0
 MarkupSafe==2.1.5
-marshmallow==3.21.1
+marshmallow==3.21.2
 matplotlib==3.8.2
 mdurl==0.1.2
 mpmath==1.3.0
 multidict==6.0.5
 multiprocess==0.70.16
 mypy-extensions==1.0.0
-networkx==3.2.1
+networkx==3.3
 numpy==1.26.4
-nvidia-cublas-cu11==11.10.3.66
-nvidia-cuda-cupti-cu11==11.7.101
-nvidia-cuda-nvrtc-cu11==11.7.99
-nvidia-cuda-runtime-cu11==11.7.99
-nvidia-cudnn-cu11==8.5.0.96
-nvidia-cufft-cu11==10.9.0.58
-nvidia-curand-cu11==10.2.10.91
-nvidia-cusolver-cu11==11.4.0.1
-nvidia-cusparse-cu11==11.7.4.91
-nvidia-nccl-cu11==2.14.3
-nvidia-nvtx-cu11==11.7.91
-orjson==3.10.0
+nvidia-cublas-cu12==12.1.3.1
+nvidia-cuda-cupti-cu12==12.1.105
+nvidia-cuda-nvrtc-cu12==12.1.105
+nvidia-cuda-runtime-cu12==12.1.105
+nvidia-cudnn-cu12==8.9.2.26
+nvidia-cufft-cu12==11.0.2.54
+nvidia-curand-cu12==10.3.2.106
+nvidia-cusolver-cu12==11.4.5.107
+nvidia-cusparse-cu12==12.1.0.106
+nvidia-nccl-cu12==2.19.3
+nvidia-nvjitlink-cu12==12.5.40
+nvidia-nvtx-cu12==12.1.105
+orjson==3.10.3
 packaging==24.0
-pandas==2.2.1
-pillow==10.2.0
+pandas==2.2.2
+pillow==10.3.0
 protobuf==3.20.3
 psutil==5.9.8
-pyarrow==15.0.2
+pyarrow==16.0.0
 pyarrow-hotfix==0.6
 pycparser==2.22
-pydantic==2.6.4
-pydantic_core==2.16.3
+pydantic==2.7.2
+pydantic_core==2.18.3
 pydub==0.25.1
 pygltflib==1.16.1
-Pygments==2.17.2
+Pygments==2.18.0
 pyparsing==3.1.2
 python-dateutil==2.9.0.post0
+python-dotenv==1.0.1
 python-multipart==0.0.9
 pytz==2024.1
 PyYAML==6.0.1
-referencing==0.34.0
-regex==2023.12.25
+referencing==0.35.1
+regex==2024.5.15
 requests==2.31.0
 rich==13.7.1
-rpds-py==0.18.0
-ruff==0.3.4
-safetensors==0.4.2
+rpds-py==0.18.1
+ruff==0.4.7
+safetensors==0.4.3
 scipy==1.11.4
 semantic-version==2.10.0
 shellingham==1.5.4
 six==1.16.0
 sniffio==1.3.1
-spaces==0.25.0
-starlette==0.36.3
-sympy==1.12
+spaces==0.28.3
+starlette==0.37.2
+sympy==1.12.1
 tokenizers==0.15.2
 tomlkit==0.12.0
 toolz==0.12.1
-torch==2.0.1
-tqdm==4.66.2
+torch==2.2.0
+tqdm==4.66.4
 transformers==4.36.1
 trimesh==4.0.5
-triton==2.0.0
-typer==0.12.0
-typer-cli==0.12.0
-typer-slim==0.12.0
+triton==2.2.0
+typer==0.12.3
 typing-inspect==0.9.0
-typing_extensions==4.10.0
+typing_extensions==4.11.0
 tzdata==2024.1
+ujson==5.10.0
 urllib3==2.2.1
-uvicorn==0.29.0
+uvicorn==0.30.0
+uvloop==0.19.0
+watchfiles==0.22.0
 websockets==11.0.3
 wrapt==1.16.0
-xformers==0.0.21
+xformers==0.0.24
 xxhash==3.4.1
 yarl==1.9.4
-zipp==3.18.1
+zipp==3.19.1

requirements_min.txt

@@ -1,14 +1,14 @@
-gradio==4.21.0
-gradio-imageslider==0.0.18
+gradio>=4.32.1
+gradio-imageslider>=0.0.20
 pygltflib==1.16.1
 trimesh==4.0.5
 imageio
 imageio-ffmpeg
 Pillow
 
-spaces==0.25.0
-accelerate==0.25.0
-diffusers==0.27.2
+spaces
+accelerate
+diffusers>=0.28.0
 matplotlib==3.8.2
 scipy==1.11.4
 torch==2.0.1