Rearranging UI, working on depth

.gitignore

@@ -164,4 +164,5 @@ cython_debug/
 /src/__pycache__
 /utils/__pycache__
 /__pycache__
-/temp_models
+/temp_models
+/.vscode/settings.json

app.py

@@ -1,3 +1,5 @@
+import gradio as gr
+import spaces
 import os
 # Import constants
 import numpy as np
@@ -7,7 +9,7 @@ from typing import Optional, Union, List, Tuple
 from PIL import Image, ImageFilter
 import cv2
 import utils.constants as constants
-import gradio as gr
+
 
 from haishoku.haishoku import Haishoku
 
@@ -15,10 +17,10 @@ from tempfile import NamedTemporaryFile
 import atexit
 import random
 #import accelerate 
-from transformers import AutoTokenizer, DPTImageProcessor, DPTForDepthEstimation
+from transformers import AutoTokenizer , DPTImageProcessor, DPTForDepthEstimation
 from pathlib import Path
 import logging
-#logging.getLogger("transformers.modeling_utils").setLevel(logging.ERROR)
+logging.getLogger("transformers.modeling_utils").setLevel(logging.ERROR)
 import gc
 
 IS_SHARED_SPACE = constants.IS_SHARED_SPACE
@@ -32,7 +34,6 @@ from utils.color_utils import (
     update_color_opacity,
 )
 from utils.misc import (get_filename, pause, convert_ratio_to_dimensions) #install_cuda_toolkit,install_torch, _get_output, setup_runtime_env)
-#from utils.depth_estimation import generate_depth_button_click
 
 from utils.image_utils import (
     change_color,
@@ -87,7 +88,8 @@ PIPELINE_CLASSES = {
     "FluxControlPipeline": FluxControlPipeline
 }
 
-import spaces
+#from utils.depth_estimation import generate_depth_and_3d
+
 
 input_image_palette = []
 current_prerendered_image = gr.State("./images/images/Beeuty-1.png")
@@ -297,7 +299,7 @@ class Condition(object):
         type_id = torch.ones_like(ids[:, :1]) * self.type_id
         return tokens, ids, type_id
 
-@spaces.GPU(progress=gr.Progress(track_tqdm=True))
+@spaces.GPU(duration=150, progress=gr.Progress(track_tqdm=True))
 def generate_image_lowmem(
     text,
     neg_prompt=None,
@@ -342,7 +344,7 @@ def generate_image_lowmem(
         if pipeline_name == "FluxPipeline":
             pipe.enable_model_cpu_offload()
             pipe.vae.enable_slicing()
-            pipe.vae.enable_tiling()
+            #pipe.vae.enable_tiling()
         else:
             pipe.enable_model_cpu_offload()
 
@@ -593,10 +595,10 @@ def generate_ai_image_local (
             return tmp.name
     except Exception as e:
         print(f"Error generating AI image: {e}")
-        gc.collect()
+        #gc.collect()
         return None
 
-@spaces.GPU(duration=140)
+#@spaces.GPU(duration=140)
 def generate_input_image_click(map_option, prompt_textbox_value, negative_prompt_textbox_value, model_textbox_value, randomize_seed=True, seed=None, use_conditioned_image=False, strength=0.5, image_format="16:9", scale_factor=(8/3), progress=gr.Progress(track_tqdm=True)):
     if randomize_seed:
         seed = random.randint(0, constants.MAX_SEED)
@@ -651,7 +653,7 @@ def generate_input_image_click(map_option, prompt_textbox_value, negative_prompt
         upscaled_image.save(tmp_upscaled.name, format="PNG")
         constants.temp_files.append(tmp_upscaled.name)
         print(f"Upscaled image saved to {tmp_upscaled.name}")
-    gc.collect()
+    #gc.collect()
     # Return the path of the upscaled image
     return tmp_upscaled.name
 
@@ -689,12 +691,13 @@ def add_border(image, mask_width, mask_height, blank_color):
 
 ################################## DEPTH ESTIMATION ##################################
 
+# Load models once during module import
+image_processor = DPTImageProcessor.from_pretrained("Intel/dpt-large",)
+depth_model = DPTForDepthEstimation.from_pretrained("Intel/dpt-large", ignore_mismatched_sizes=True)
 
-#@spaces.GPU()
+@spaces.GPU()
 def estimate_depth(image):
-    # Load models once during module import
-    image_processor = DPTImageProcessor.from_pretrained("Intel/dpt-large")
-    depth_model = DPTForDepthEstimation.from_pretrained("Intel/dpt-large", ignore_mismatched_sizes=True)
+
 
     # Ensure image is in RGB mode
     if image.mode != "RGB":
@@ -736,7 +739,7 @@ def estimate_depth(image):
 
     return depth_pil, output
 
-#@spaces.GPU()
+@spaces.GPU()
 def create_3d_model(rgb_image, depth_array, voxel_size_factor=0.01):
     import open3d as o3d
     depth_o3d = o3d.geometry.Image(depth_array.astype(np.float32))
@@ -745,7 +748,7 @@ def create_3d_model(rgb_image, depth_array, voxel_size_factor=0.01):
     rgbd_image = o3d.geometry.RGBDImage.create_from_color_and_depth(
         rgb_o3d,
         depth_o3d,
-        convert_rgb_to_intensity=False
+        convert_rgb_to_intensity=True
     )
 
     # Create a point cloud from the RGBD image
@@ -782,6 +785,7 @@ def generate_depth_and_3d(input_image_path, voxel_size_factor):
     model_path = create_3d_model(resized_image, depth_array, voxel_size_factor=voxel_size_factor)
     return depth_image, model_path
 
+
 def generate_depth_button_click(depth_image_source, voxel_size_factor, input_image, output_image, overlay_image, bordered_image_output):
     if depth_image_source == "Input Image":
         image_path = input_image
@@ -798,7 +802,7 @@ def generate_depth_button_click(depth_image_source, voxel_size_factor, input_ima
 def getVersions():
     return versions_html()
 
-generate_input_image_click.zerogpu = True
+#generate_input_image_click.zerogpu = True
 #generate_depth_button_click.zerogpu = True
 #def main(debug=False):
 title = "HexaGrid Creator"
@@ -1020,17 +1024,16 @@ with gr.Blocks(css_paths="style_20250128.css", title=title, theme='Surn/beeuty')
                 with gr.Column(scale=2):
                         with gr.Accordion("Template Image Styles", open = False):
                             with gr.Row():
-                                # Gallery from PRE_RENDERED_IMAGES GOES HERE
-                                prerendered_image_gallery = gr.Gallery(label="Image Gallery", show_label=True, value=build_prerendered_images(constants.pre_rendered_maps_paths), elem_id="gallery", elem_classes="solid", type="filepath", columns=[3], rows=[3], preview=False ,object_fit="contain", height="auto", format="png",allow_preview=False)
-                            with gr.Row():
-                                image_guidance_stength = gr.Slider(label="Image Guidance Strength (prompt percentage)", minimum=0, maximum=1.0, value=0.8, step=0.01, interactive=True)
+                                with gr.Column(scale=2):
+                                    # Gallery from PRE_RENDERED_IMAGES GOES HERE
+                                    prerendered_image_gallery = gr.Gallery(label="Image Gallery", show_label=True, value=build_prerendered_images(constants.pre_rendered_maps_paths), elem_id="gallery", elem_classes="solid", type="filepath", columns=[3], rows=[3], preview=False ,object_fit="contain", height="auto", format="png",allow_preview=False)
                                 with gr.Column():
+                                    image_guidance_stength = gr.Slider(label="Image Guidance Strength (prompt percentage)", minimum=0, maximum=1.0, value=0.8, step=0.01, interactive=True)                                
                                     replace_input_image_button = gr.Button(
                                         "Replace Input Image",
                                         elem_id="prerendered_replace_input_image_button",
                                         elem_classes="solid"
-                                    )
-                                with gr.Column():
+                                    )                                
                                     generate_input_image_from_gallery = gr.Button(
                                         "Generate AI Image from Gallery",
                                         elem_id="generate_input_image_from_gallery",
@@ -1212,21 +1215,21 @@ with gr.Blocks(css_paths="style_20250128.css", title=title, theme='Surn/beeuty')
 
 
 
-#if __name__ == "__main__":
-logging.basicConfig(
-    format="[%(levelname)s] %(asctime)s %(message)s", level=logging.INFO
-)
-logging.info("Environment Variables: %s" % os.environ)
-    # if _get_output(["nvcc", "--version"]) is None:
-    #     logging.info("Installing CUDA toolkit...")
-    #     install_cuda_toolkit()
-    # else:
-    #     logging.info("Detected CUDA: %s" % _get_output(["nvcc", "--version"]))
-
-    # logging.info("Installing CUDA extensions...")
-    # setup_runtime_env()
-    #main(os.getenv("DEBUG") == "1")
-    #main()
-
-hexaGrid.queue(default_concurrency_limit=1,max_size=12,api_open=False)
-hexaGrid.launch(allowed_paths=["assets","/","./assets","images","./images", "./images/prerendered"], favicon_path="./assets/favicon.ico", max_file_size="10mb")
+if __name__ == "__main__":
+    logging.basicConfig(
+        format="[%(levelname)s] %(asctime)s %(message)s", level=logging.INFO
+    )
+    logging.info("Environment Variables: %s" % os.environ)
+        # if _get_output(["nvcc", "--version"]) is None:
+        #     logging.info("Installing CUDA toolkit...")
+        #     install_cuda_toolkit()
+        # else:
+        #     logging.info("Detected CUDA: %s" % _get_output(["nvcc", "--version"]))
+
+        # logging.info("Installing CUDA extensions...")
+        # setup_runtime_env()
+        #main(os.getenv("DEBUG") == "1")
+        #main()
+
+    hexaGrid.queue(default_concurrency_limit=1,max_size=12,api_open=False)
+    hexaGrid.launch(allowed_paths=["assets","/","./assets","images","./images", "./images/prerendered"], favicon_path="./assets/favicon.ico", max_file_size="10mb")

utils/constants.py

@@ -46,7 +46,7 @@ PROMPTS = {
     "BorderBlack": "Top-down view of a hexagon-based alien map with black borders. Features rivers, mountains, volcanoes, and snow at top and bottom. Colors: light blue, green, tan, brown. No reflections or shadows. Partial hexes on edges are black.",
     "Earth": "Top-down view of a world map with rivers, mountains, volcanoes, and snow at top and bottom. Colors: light blue, green, tan, brown. No reflections or shadows. Partial edge hexes are black. Overhead view.",
     "Beeuty": "Top-down view of a table map with honeycomb_shapes, lakes, dense forests, magical flora, and hex_grids. Map for tabletop gaming with clarity and strategic elements. Colors: yellow, green, purple, brown. Partial hexes on edges are black.",
-    "Scyfi City": "Top-down view of a futuristic urban battlefield map with lakes, forests, ruined buildings, and city streets. Emphasizes clarity and strategy for tabletop games. Colors: teal, dark green, violet, brown. Partial edge hexes are black. Viewed from above.",
+    "Scifi City": "Top-down view of a futuristic urban battlefield map with lakes, forests, ruined buildings, and city streets. Emphasizes clarity and strategy for tabletop games. Colors: teal, dark green, violet, brown. Partial edge hexes are black. Viewed from above.",
     "Alien Landscape": "Top-down view of a barren alien world map made from hexagon pieces. Features light blue rivers, brown mountains, red volcanoes, and white snow at top and bottom. Colors: light blue, green, tan, brown. Partial hexes on edges are black.",
     "Alien World": "Top-down view of an alien world map built from hexagon pieces. Includes rivers, mountains, volcanoes, and snowy areas. Colors: light blue, green, tan, brown. Partial edge hexes are black. Overhead view.",
     "Mystic Forest": "Top-down view of a mystic forest map with lakes, dense forests, magical flora, and hex grids. Designed for clarity in tabletop gaming. Colors: light blue, green, purple, brown. Partial hexes on edges are black.",