progress

.opencode/skills/alpha-mask-creator/SKILL.md

@@ -0,0 +1,65 @@
+---
+name: alpha-mask-creator
+description: Takes an image and creates a high quality alpha mask of the requested entity. Typically invoked when the user asks to "create a specific alpha mask for"
+---
+# Alpha Mask Creator
+
+This skill creates alpha masks from images based on user requests. Use this when the user asks to "create a specific alpha mask for..." along with providing an image.
+
+## Workflow
+
+### 1. Draw Polygon Outline
+Use the `polygon-drawer` skill to draw a polygon outline around the identified object. Save the output to `./tmp/` with a sensible filename:
+- Format: `<original_name>.<object>.png`
+- Example: If the user wants a door mask from `room_010.png`, name it `room_010.door.png`
+
+### 2. Extract Mask
+Run the `extract_mask` script to generate the alpha mask:
+```bash
+source ./tools/venv/bin/activate
+python tools/extract_mask.py "the door with the red outline" ./tmp/<original_name>.<object>.png ./tmp/<original_name>.<object>.mask.png
+```
+
+### 3. Quality Check
+Examine the generated mask using the `image-inspector` agent:
+- Check if the mask is precise and accurate
+- Verify the object is properly isolated
+- Ensure no artifacts or incorrect areas
+
+**If the mask quality is poor:**
+- Retry the polygon-drawer step with different parameters or seed
+- Iterate until the mask is satisfactory
+
+### 5. Report Results
+Tell the user:
+- Location of the polygon outline image: `./tmp/<original_name>.<object>.png`
+- Location of the mask: `./tmp/<original_name>.<object>.mask.png`
+- Confirm the mask quality is acceptable
+
+## Naming Conventions
+
+| User Request | Original File | Output Files |
+|--------------|---------------|--------------|
+| Door | `room_010.png` | `room_010.door.png`, `room_010.door.mask.png` |
+| Tree | `forest_bg.png` | `forest_bg.tree.png`, `forest_bg.tree.mask.png` |
+| Character | `scene_005.png` | `scene_005.character.png`, `scene_005.character.mask.png` |
+
+## Tools Used
+
+- `image-inspector` - Analyze images to locate objects
+- `polygon-drawer` - Draw polygon outlines
+- `bash` - Run extract_mask script
+- `image-expert` - Verify mask quality
+
+## Example Interaction
+
+**User:** "Create a specific alpha mask for the door in this image" [attaches `cottage_exterior.png`]
+
+**You:**
+1. Analyze the image to locate the door
+2. Draw polygon around the door → `./tmp/cottage_exterior.door.png`
+3. Extract mask → `./tmp/cottage_exterior.door.mask.png`
+4. Verify mask quality
+5. Report: "Mask created successfully! Files saved to:
+   - Polygon outline: `./tmp/cottage_exterior.door.png`
+   - Alpha mask: `./tmp/cottage_exterior.door.mask.png`"

.opencode/skills/kq4-polygon-drawer/SKILL.md

@@ -0,0 +1,67 @@
+# Polygon Drawer Skill
+
+This skill uses @tools/draw_polygon.py to draw polygons on any image. It's useful for highlighting areas, marking objects, creating masks, or annotating images with colored shapes.
+
+## Usage Examples
+
+### Draw a red box around an object
+```
+Draw a red box around the door in image.png
+# Executes: python tools/draw_polygon.py tmp/annotated_image.png "0.3,0.2 0.7,0.2 0.7,0.8 0.3,0.8" --color red --save tmp/annotated_image.png
+# Opens: /home/noti/dev/ai-game-2/tmp/annotated_image.png
+```
+
+### Highlight an area in green
+```
+Highlight the pool in green using a polygon
+# Executes: python tools/draw_polygon.py tmp/pool_highlight.png "0.2,0.5 0.8,0.5 0.8,0.9 0.2,0.9" --color green --fill --save tmp/pool_highlight.png
+# Opens: /home/noti/dev/ai-game-2/tmp/pool_highlight.png
+```
+
+### Draw a blue circle (approximated with polygon)
+```
+Draw a blue circle around the tree
+# Executes: python tools/draw_polygon.py tmp/tree_circle.png "0.4,0.3 0.6,0.3 0.6,0.5 0.4,0.5" --color blue --save tmp/tree_circle.png
+# Opens: /home/noti/dev/ai-game-2/tmp/tree_circle.png
+```
+
+### Fill a region with semi-transparent overlay
+```
+Fill the sky area with semi-transparent yellow
+# Executes: python tools/draw_polygon.py tmp/sky_overlay.png "0,0 1,0 1,0.6 0,0.6" --color yellow --fill --save tmp/sky_overlay.png
+# Opens: /home/noti/dev/ai-game-2/tmp/sky_overlay.png
+```
+
+### Mark a specific point with a small polygon
+```
+Mark the character location with a red polygon
+# Executes: python tools/draw_polygon.py tmp/character_mark.png "0.5,0.4 0.52,0.38 0.54,0.4 0.52,0.42" --color red --save tmp/character_mark.png
+# Opens: /home/noti/dev/ai-game-2/tmp/character_mark.png
+```
+
+## How It Works
+
+1. **Analyze** the image to understand what needs highlighting
+2. **Extract** polygon coordinates based on visual patterns or user input
+3. **Execute** the draw_polygon.py script with percentage coordinates (0.0-1.0)
+4. **Save** the result to ./tmp/ and display the file path
+5. **Open** the output for you to view
+
+## Coordinate System
+
+Coordinates use **percentage mode** (0.0-1.0) by default:
+- `0.0,0.0` = top-left corner
+- `1.0,1.0` = bottom-right corner
+- `0.5,0.5` = center of image
+
+## Common Use Cases
+
+- Highlighting objects or regions of interest
+- Creating masks for image processing
+- Marking locations for documentation
+- Annotating images with colored boundaries
+- Visualizing spatial relationships
+
+## Output
+
+All annotated images are saved to `./tmp/` and the full path is displayed so you can open them in your image viewer.

.opencode/skills/polygon-drawer/SKILL.md

@@ -0,0 +1,79 @@
+---
+name: polygon-drawer
+description: Takes an image and adds a polygon to it at the user's request. Typically the user will ask to add a polygon around a particular object
+---
+# Polygon Drawer Skill
+
+This skill uses @tools/draw_polygon.py to help you draw polygons on any image. It's useful for highlighting areas, drawing boxes, marking regions, or annotating images with colored shapes.
+
+## Usage Examples
+
+### Draw a red box around an object
+```
+Draw a red box around the door in <image>
+```
+
+### Highlight a specific area in green
+```
+Highlight the pool area in <image> with a green polygon
+```
+
+### Draw a blue rectangle with percentage coordinates
+```
+Draw a blue polygon on <image> using coordinates: 0.5,0.3 0.7,0.3 0.7,0.6 0.5,0.6
+```
+
+### Mark a triangular region with fill
+```
+Draw a semi-transparent yellow triangle on <image>: 0.5,0.1 0.9,0.9 0.1,0.9 --fill
+```
+
+### Draw a custom-colored polygon with thick lines
+```
+Draw a cyan polygon on <image>: 0.2,0.2 0.8,0.2 0.8,0.8 0.2,0.8 --color cyan --thickness 5
+```
+
+### Use pixel coordinates (absolute mode)
+```
+Draw a red box using pixel coordinates: 100,50 400,50 400,300 100,300 --absolute
+```
+
+Make sure to quote the coordinates when calling the script: "100,50 400,50 400,300 100,300"
+
+## How It Works
+
+1. **Analyze** the image to identify what needs highlighting
+2. **Calculate** polygon coordinates (default: percentage mode 0.0-1.0)
+3. **Execute** `@tools/draw_polygon.py` with the calculated points
+4. **Output** the resulting image to `./tmp/` directory
+5. **Display** the save path so you can open it
+
+## Coordinate Systems
+
+- **Percentage mode** (default): 0.0-1.0 coordinates where (0,0) is top-left and (1,1) is bottom-right
+- **Absolute mode**: Use `--absolute` flag for actual pixel coordinates
+
+## Output Location
+By default, the image is just shown to the user without writing anything. 
+
+If the user asks to save it, use the --save flag.
+
+All generated images are saved to:
+```
+./tmp/
+```
+
+The skill will output the full path to the saved image, e.g., `./tmp/output_1234567890.png`
+
+## Color Options
+
+- Named colors: red, green, blue, yellow, cyan, magenta, white, black, orange, purple, brown, pink, gray, grey, lime, navy, teal, maroon
+- Hex codes: #FF0000, #00FF00FF, etc.
+
+## Flags
+
+- `--fill`: Fill polygon with semi-transparent color
+- `--absolute`: Use pixel coordinates instead of percentages
+- `--color <name>`: Specify polygon color (default: red)
+- `--thickness <pixels>`: Line thickness (default: 4)
+- `--save <path>`: Override default save location (still goes to ./tmp/)

AGENTS.md

@@ -8,6 +8,10 @@ This is a **Godot 4.6** game project (King's Quest IV remake). The project uses:
 - NavigationServer2D for pathfinding
 - Custom script builder pattern for game scripts
 
+## Tools
+ * Helper scripts (python) exist in ./tools
+ * a virtual env exists there, too. When you need to run one, you should source ./tools/venv/bin/activate
+
 
 ## Build & Development Commands
 
@@ -290,3 +294,4 @@ scene.start_main_script(scene.ScriptBuilder.init(scene.ScriptBuilder.walk_path(e
 func _on_exit_interacted() -> void:
     $target_scene.default_script(self)
 ```
+

asset-work/kq4_062_bedroom/caption_1_3877392774_with_box.png.import

@@ -0,0 +1,40 @@
+[remap]
+
+importer="texture"
+type="CompressedTexture2D"
+uid="uid://w8p5drokk4sv"
+path="res://.godot/imported/caption_1_3877392774_with_box.png-49e8cf7c8aa8b4112e682ca98e672cf8.ctex"
+metadata={
+"vram_texture": false
+}
+
+[deps]
+
+source_file="res://asset-work/kq4_062_bedroom/caption_1_3877392774_with_box.png"
+dest_files=["res://.godot/imported/caption_1_3877392774_with_box.png-49e8cf7c8aa8b4112e682ca98e672cf8.ctex"]
+
+[params]
+
+compress/mode=0
+compress/high_quality=false
+compress/lossy_quality=0.7
+compress/uastc_level=0
+compress/rdo_quality_loss=0.0
+compress/hdr_compression=1
+compress/normal_map=0
+compress/channel_pack=0
+mipmaps/generate=false
+mipmaps/limit=-1
+roughness/mode=0
+roughness/src_normal=""
+process/channel_remap/red=0
+process/channel_remap/green=1
+process/channel_remap/blue=2
+process/channel_remap/alpha=3
+process/fix_alpha_border=true
+process/premult_alpha=false
+process/normal_map_invert_y=false
+process/hdr_as_srgb=false
+process/hdr_clamp_exposure=false
+process/size_limit=0
+detect_3d/compress_to=1

tmp/caption_1_1979017646_generated.house.mask.png.import

@@ -0,0 +1,40 @@
+[remap]
+
+importer="texture"
+type="CompressedTexture2D"
+uid="uid://cqq5llgcudt15"
+path="res://.godot/imported/caption_1_1979017646_generated.house.mask.png-405a06af69107de7cf56d2c8cd43a11a.ctex"
+metadata={
+"vram_texture": false
+}
+
+[deps]
+
+source_file="res://tmp/caption_1_1979017646_generated.house.mask.png"
+dest_files=["res://.godot/imported/caption_1_1979017646_generated.house.mask.png-405a06af69107de7cf56d2c8cd43a11a.ctex"]
+
+[params]
+
+compress/mode=0
+compress/high_quality=false
+compress/lossy_quality=0.7
+compress/uastc_level=0
+compress/rdo_quality_loss=0.0
+compress/hdr_compression=1
+compress/normal_map=0
+compress/channel_pack=0
+mipmaps/generate=false
+mipmaps/limit=-1
+roughness/mode=0
+roughness/src_normal=""
+process/channel_remap/red=0
+process/channel_remap/green=1
+process/channel_remap/blue=2
+process/channel_remap/alpha=3
+process/fix_alpha_border=true
+process/premult_alpha=false
+process/normal_map_invert_y=false
+process/hdr_as_srgb=false
+process/hdr_clamp_exposure=false
+process/size_limit=0
+detect_3d/compress_to=1

tmp/caption_1_1979017646_generated.house.png.import

@@ -0,0 +1,40 @@
+[remap]
+
+importer="texture"
+type="CompressedTexture2D"
+uid="uid://bjjc74ww1ynu2"
+path="res://.godot/imported/caption_1_1979017646_generated.house.png-ee332416aaeec8be4ca29e4afab44dae.ctex"
+metadata={
+"vram_texture": false
+}
+
+[deps]
+
+source_file="res://tmp/caption_1_1979017646_generated.house.png"
+dest_files=["res://.godot/imported/caption_1_1979017646_generated.house.png-ee332416aaeec8be4ca29e4afab44dae.ctex"]
+
+[params]
+
+compress/mode=0
+compress/high_quality=false
+compress/lossy_quality=0.7
+compress/uastc_level=0
+compress/rdo_quality_loss=0.0
+compress/hdr_compression=1
+compress/normal_map=0
+compress/channel_pack=0
+mipmaps/generate=false
+mipmaps/limit=-1
+roughness/mode=0
+roughness/src_normal=""
+process/channel_remap/red=0
+process/channel_remap/green=1
+process/channel_remap/blue=2
+process/channel_remap/alpha=3
+process/fix_alpha_border=true
+process/premult_alpha=false
+process/normal_map_invert_y=false
+process/hdr_as_srgb=false
+process/hdr_clamp_exposure=false
+process/size_limit=0
+detect_3d/compress_to=1

tmp/caption_1_454377357_generated.door.mask.png.import

@@ -0,0 +1,40 @@
+[remap]
+
+importer="texture"
+type="CompressedTexture2D"
+uid="uid://bh0a1gngcfhlt"
+path="res://.godot/imported/caption_1_454377357_generated.door.mask.png-aec0dc7b07204882097c739e77da3d17.ctex"
+metadata={
+"vram_texture": false
+}
+
+[deps]
+
+source_file="res://tmp/caption_1_454377357_generated.door.mask.png"
+dest_files=["res://.godot/imported/caption_1_454377357_generated.door.mask.png-aec0dc7b07204882097c739e77da3d17.ctex"]
+
+[params]
+
+compress/mode=0
+compress/high_quality=false
+compress/lossy_quality=0.7
+compress/uastc_level=0
+compress/rdo_quality_loss=0.0
+compress/hdr_compression=1
+compress/normal_map=0
+compress/channel_pack=0
+mipmaps/generate=false
+mipmaps/limit=-1
+roughness/mode=0
+roughness/src_normal=""
+process/channel_remap/red=0
+process/channel_remap/green=1
+process/channel_remap/blue=2
+process/channel_remap/alpha=3
+process/fix_alpha_border=true
+process/premult_alpha=false
+process/normal_map_invert_y=false
+process/hdr_as_srgb=false
+process/hdr_clamp_exposure=false
+process/size_limit=0
+detect_3d/compress_to=1

tmp/caption_1_454377357_generated.door.png.import

@@ -0,0 +1,40 @@
+[remap]
+
+importer="texture"
+type="CompressedTexture2D"
+uid="uid://cgx781rh8x160"
+path="res://.godot/imported/caption_1_454377357_generated.door.png-bc97e1ab49f2ba5ecd73d4604203e223.ctex"
+metadata={
+"vram_texture": false
+}
+
+[deps]
+
+source_file="res://tmp/caption_1_454377357_generated.door.png"
+dest_files=["res://.godot/imported/caption_1_454377357_generated.door.png-bc97e1ab49f2ba5ecd73d4604203e223.ctex"]
+
+[params]
+
+compress/mode=0
+compress/high_quality=false
+compress/lossy_quality=0.7
+compress/uastc_level=0
+compress/rdo_quality_loss=0.0
+compress/hdr_compression=1
+compress/normal_map=0
+compress/channel_pack=0
+mipmaps/generate=false
+mipmaps/limit=-1
+roughness/mode=0
+roughness/src_normal=""
+process/channel_remap/red=0
+process/channel_remap/green=1
+process/channel_remap/blue=2
+process/channel_remap/alpha=3
+process/fix_alpha_border=true
+process/premult_alpha=false
+process/normal_map_invert_y=false
+process/hdr_as_srgb=false
+process/hdr_clamp_exposure=false
+process/size_limit=0
+detect_3d/compress_to=1

tmp/caption_1_454377357_generated.door.v2.mask.png.import

@@ -0,0 +1,40 @@
+[remap]
+
+importer="texture"
+type="CompressedTexture2D"
+uid="uid://b82q5606fv2ta"
+path="res://.godot/imported/caption_1_454377357_generated.door.v2.mask.png-3db723937261dbc5639bd97d6f39ddb8.ctex"
+metadata={
+"vram_texture": false
+}
+
+[deps]
+
+source_file="res://tmp/caption_1_454377357_generated.door.v2.mask.png"
+dest_files=["res://.godot/imported/caption_1_454377357_generated.door.v2.mask.png-3db723937261dbc5639bd97d6f39ddb8.ctex"]
+
+[params]
+
+compress/mode=0
+compress/high_quality=false
+compress/lossy_quality=0.7
+compress/uastc_level=0
+compress/rdo_quality_loss=0.0
+compress/hdr_compression=1
+compress/normal_map=0
+compress/channel_pack=0
+mipmaps/generate=false
+mipmaps/limit=-1
+roughness/mode=0
+roughness/src_normal=""
+process/channel_remap/red=0
+process/channel_remap/green=1
+process/channel_remap/blue=2
+process/channel_remap/alpha=3
+process/fix_alpha_border=true
+process/premult_alpha=false
+process/normal_map_invert_y=false
+process/hdr_as_srgb=false
+process/hdr_clamp_exposure=false
+process/size_limit=0
+detect_3d/compress_to=1

tmp/caption_1_454377357_generated.door.v2.png.import

@@ -0,0 +1,40 @@
+[remap]
+
+importer="texture"
+type="CompressedTexture2D"
+uid="uid://doo1jdbpi336u"
+path="res://.godot/imported/caption_1_454377357_generated.door.v2.png-2fdfe23cc08c0c7abc1681d07849c9fc.ctex"
+metadata={
+"vram_texture": false
+}
+
+[deps]
+
+source_file="res://tmp/caption_1_454377357_generated.door.v2.png"
+dest_files=["res://.godot/imported/caption_1_454377357_generated.door.v2.png-2fdfe23cc08c0c7abc1681d07849c9fc.ctex"]
+
+[params]
+
+compress/mode=0
+compress/high_quality=false
+compress/lossy_quality=0.7
+compress/uastc_level=0
+compress/rdo_quality_loss=0.0
+compress/hdr_compression=1
+compress/normal_map=0
+compress/channel_pack=0
+mipmaps/generate=false
+mipmaps/limit=-1
+roughness/mode=0
+roughness/src_normal=""
+process/channel_remap/red=0
+process/channel_remap/green=1
+process/channel_remap/blue=2
+process/channel_remap/alpha=3
+process/fix_alpha_border=true
+process/premult_alpha=false
+process/normal_map_invert_y=false
+process/hdr_as_srgb=false
+process/hdr_clamp_exposure=false
+process/size_limit=0
+detect_3d/compress_to=1

tmp/stone_box.png.import

@@ -0,0 +1,40 @@
+[remap]
+
+importer="texture"
+type="CompressedTexture2D"
+uid="uid://tc8kp6g4d6v3"
+path="res://.godot/imported/stone_box.png-23873de51cd956b464cbf5c40e614970.ctex"
+metadata={
+"vram_texture": false
+}
+
+[deps]
+
+source_file="res://tmp/stone_box.png"
+dest_files=["res://.godot/imported/stone_box.png-23873de51cd956b464cbf5c40e614970.ctex"]
+
+[params]
+
+compress/mode=0
+compress/high_quality=false
+compress/lossy_quality=0.7
+compress/uastc_level=0
+compress/rdo_quality_loss=0.0
+compress/hdr_compression=1
+compress/normal_map=0
+compress/channel_pack=0
+mipmaps/generate=false
+mipmaps/limit=-1
+roughness/mode=0
+roughness/src_normal=""
+process/channel_remap/red=0
+process/channel_remap/green=1
+process/channel_remap/blue=2
+process/channel_remap/alpha=3
+process/fix_alpha_border=true
+process/premult_alpha=false
+process/normal_map_invert_y=false
+process/hdr_as_srgb=false
+process/hdr_clamp_exposure=false
+process/size_limit=0
+detect_3d/compress_to=1

tmp/stone_mask.png.import

@@ -0,0 +1,40 @@
+[remap]
+
+importer="texture"
+type="CompressedTexture2D"
+uid="uid://d1h745821w85b"
+path="res://.godot/imported/stone_mask.png-2f0bfa74533a45031cbfc6d9f501b5d2.ctex"
+metadata={
+"vram_texture": false
+}
+
+[deps]
+
+source_file="res://tmp/stone_mask.png"
+dest_files=["res://.godot/imported/stone_mask.png-2f0bfa74533a45031cbfc6d9f501b5d2.ctex"]
+
+[params]
+
+compress/mode=0
+compress/high_quality=false
+compress/lossy_quality=0.7
+compress/uastc_level=0
+compress/rdo_quality_loss=0.0
+compress/hdr_compression=1
+compress/normal_map=0
+compress/channel_pack=0
+mipmaps/generate=false
+mipmaps/limit=-1
+roughness/mode=0
+roughness/src_normal=""
+process/channel_remap/red=0
+process/channel_remap/green=1
+process/channel_remap/blue=2
+process/channel_remap/alpha=3
+process/fix_alpha_border=true
+process/premult_alpha=false
+process/normal_map_invert_y=false
+process/hdr_as_srgb=false
+process/hdr_clamp_exposure=false
+process/size_limit=0
+detect_3d/compress_to=1

tmp/stone_outline.png.import

@@ -0,0 +1,40 @@
+[remap]
+
+importer="texture"
+type="CompressedTexture2D"
+uid="uid://cdo0hg0am6cus"
+path="res://.godot/imported/stone_outline.png-33fd18a74617b08b2436a4a9d98a33b6.ctex"
+metadata={
+"vram_texture": false
+}
+
+[deps]
+
+source_file="res://tmp/stone_outline.png"
+dest_files=["res://.godot/imported/stone_outline.png-33fd18a74617b08b2436a4a9d98a33b6.ctex"]
+
+[params]
+
+compress/mode=0
+compress/high_quality=false
+compress/lossy_quality=0.7
+compress/uastc_level=0
+compress/rdo_quality_loss=0.0
+compress/hdr_compression=1
+compress/normal_map=0
+compress/channel_pack=0
+mipmaps/generate=false
+mipmaps/limit=-1
+roughness/mode=0
+roughness/src_normal=""
+process/channel_remap/red=0
+process/channel_remap/green=1
+process/channel_remap/blue=2
+process/channel_remap/alpha=3
+process/fix_alpha_border=true
+process/premult_alpha=false
+process/normal_map_invert_y=false
+process/hdr_as_srgb=false
+process/hdr_clamp_exposure=false
+process/size_limit=0
+detect_3d/compress_to=1

tools/draw_polygon.py

@@ -0,0 +1,212 @@
+#!/usr/bin/env python3
+"""Draw a polygon on an image using Pillow."""
+
+import argparse
+import sys
+from pathlib import Path
+
+try:
+    from PIL import Image, ImageDraw, ImageFilter
+except ImportError:
+    print("Error: Pillow is required. Install with: pip install Pillow")
+    sys.exit(1)
+
+
+COLOR_PALETTE = {
+    "red": "#FF0000",
+    "green": "#00FF00",
+    "blue": "#0000FF",
+    "yellow": "#FFFF00",
+    "cyan": "#00FFFF",
+    "magenta": "#FF00FF",
+    "white": "#FFFFFF",
+    "black": "#000000",
+    "orange": "#FFA500",
+    "purple": "#800080",
+    "brown": "#A52A2A",
+    "pink": "#FFC0CB",
+    "gray": "#808080",
+    "grey": "#808080",
+    "lime": "#00FF00",
+    "navy": "#000080",
+    "teal": "#008080",
+    "maroon": "#800000",
+}
+
+
+def parse_color(color_str: str) -> str:
+    """Parse color string to hex."""
+    color_lower = color_str.lower()
+    if color_lower in COLOR_PALETTE:
+        return COLOR_PALETTE[color_lower]
+    if color_str.startswith("#") and len(color_str) in [7, 9]:
+        return color_str
+    raise ValueError(
+        f"Invalid color '{color_str}'. Use a name (e.g., red, blue) or hex code (e.g., #FF0000)."
+    )
+
+
+def parse_points(points_str: str) -> list[tuple[float, float]]:
+    """Parse points string into list of (x, y) tuples."""
+    points = []
+    for point in points_str.split():
+        if "," not in point:
+            raise ValueError(f"Invalid point format: '{point}'. Use 'x,y' format.")
+        try:
+            x, y = point.split(",")
+            points.append((float(x.strip()), float(y.strip())))
+        except ValueError:
+            raise ValueError(f"Invalid point format: '{point}'. Use 'x,y' format.")
+    return points
+
+
+def convert_to_pixels(
+    points: list[tuple[float, float]], width: int, height: int, absolute: bool
+) -> list[tuple[int, int]]:
+    """Convert points to pixel coordinates."""
+    if absolute:
+        return [(int(x), int(y)) for x, y in points]
+    else:
+        return [
+            (int(x * width), int(y * height))
+            for x, y in points
+            if 0 <= x <= 1 and 0 <= y <= 1
+        ]
+
+
+def draw_polygon_on_image(
+    image_path: Path,
+    points: list[tuple[float, float]],
+    color: str,
+    thickness: int,
+    fill: bool,
+    absolute: bool,
+    save_path: Path | None,
+) -> None:
+    """Draw polygon on image."""
+    image = Image.open(image_path).convert("RGBA")
+    width, height = image.size
+
+    pixel_points = convert_to_pixels(points, width, height, absolute)
+
+    if len(pixel_points) < 3:
+        raise ValueError("Need at least 3 points to draw a polygon.")
+
+    draw = ImageDraw.Draw(image)
+
+    hex_color = parse_color(color)
+    draw.polygon(pixel_points, fill=hex_color if fill else None, outline=hex_color, width=thickness)
+
+    if fill:
+        alpha = 128
+        fill_color = hex_color[:7] + f"{alpha:02X}"
+        draw.polygon(pixel_points, fill=fill_color, outline=hex_color, width=thickness)
+
+    if save_path:
+        image.save(save_path)
+        print(f"Saved: {save_path}")
+    else:
+        image.show()
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        description="Draw a polygon on an image using Pillow.",
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+        epilog="""
+Examples:
+  # Draw a triangle using percentages (default)
+  python draw_polygon.py scene.png "0.5,0.1 0.9,0.9 0.1,0.9"
+
+  # Draw with pixel coordinates (absolute mode)
+  python draw_polygon.py scene.png "100,50 400,50 250,300" --absolute
+
+  # Save to file with blue color
+  python draw_polygon.py scene.png "0.2,0.2 0.8,0.2 0.8,0.8 0.2,0.8" --color blue --save output.png
+
+  # Fill polygon semi-transparently
+  python draw_polygon.py scene.png "0.3,0.3 0.7,0.3 0.7,0.7 0.3,0.7" --fill --color red
+
+  # Draw with custom hex color and thickness
+  python draw_polygon.py scene.png "0.1,0.1 0.9,0.1 0.9,0.9 0.1,0.9" --color #00FF00 --thickness 5
+
+Coordinate Formats:
+  Percentage (default): 0.0 to 1.0, where 0.0,0.0 is top-left and 1.0,1.0 is bottom-right
+  Absolute (with --absolute): Actual pixel coordinates from 0 to image width/height
+
+Color Formats:
+  Named colors: red, green, blue, yellow, cyan, magenta, white, black, orange, purple, brown, pink, gray, grey, lime, navy, teal, maroon
+  Hex codes: #FF0000, #00FF00FF, etc.
+        """,
+    )
+
+    parser.add_argument(
+        "image",
+        type=Path,
+        help="Path to input image file",
+    )
+    parser.add_argument(
+        "points",
+        type=str,
+        help="Space-separated list of x,y coordinates (e.g., '0.1,0.1 0.9,0.1 0.5,0.9'). Use percentages (0.0-1.0) by default, or --absolute for pixel coordinates.",
+    )
+    parser.add_argument(
+        "--color",
+        type=str,
+        default="red",
+        help="Polygon color (default: red). Use named colors (red, blue, green, etc.) or hex codes (#FF0000).",
+    )
+    parser.add_argument(
+        "--absolute",
+        action="store_true",
+        help="Use pixel coordinates instead of percentages",
+    )
+    parser.add_argument(
+        "--save",
+        type=Path,
+        default=None,
+        help="Save output to file path instead of displaying",
+    )
+    parser.add_argument(
+        "--thickness",
+        type=int,
+        default=2,
+        help="Line thickness in pixels (default: 2)",
+    )
+    parser.add_argument(
+        "--fill",
+        action="store_true",
+        help="Fill polygon with semi-transparent color",
+    )
+
+    args = parser.parse_args()
+
+    if not args.image.exists():
+        print(f"Error: Image not found: {args.image}", file=sys.stderr)
+        sys.exit(1)
+
+    try:
+        points = parse_points(args.points)
+        if len(points) < 3:
+            print(f"Error: Need at least 3 points. Got {len(points)}.", file=sys.stderr)
+            sys.exit(1)
+
+        draw_polygon_on_image(
+            image_path=args.image,
+            points=points,
+            color=args.color,
+            thickness=args.thickness,
+            fill=args.fill,
+            absolute=args.absolute,
+            save_path=args.save,
+        )
+    except ValueError as e:
+        print(f"Error: {e}", file=sys.stderr)
+        sys.exit(1)
+    except Exception as e:
+        print(f"Error: {e}", file=sys.stderr)
+        sys.exit(1)
+
+
+if __name__ == "__main__":
+    main()

tools/extract_mask.py

@@ -125,7 +125,7 @@ def extract_mask(
     with open(workflow_path, "r") as f:
         workflow = json.load(f)
 
-    prompt_text = f"Create a black and white alpha mask of {subject}"
+    prompt_text = f"Create a black and white alpha mask of {subject}, leaving everything else black"
 
     print(f"Encoding input image...")
     base64_image = encode_image_base64(input_image)

tools/make_ora.py

@@ -0,0 +1,154 @@
+import argparse
+import os
+import zipfile
+import tempfile
+import shutil
+import numpy as np
+from PIL import Image
+import xml.etree.ElementTree as ET
+
+NUM_LAYERS = 4
+MASKS_PER_LAYER = 3
+NOISE_SCALES = [4, 16, 64]
+
+
+def noise_mask(w, h, scale):
+    sw = max(1, w // scale)
+    sh = max(1, h // scale)
+
+    noise = np.random.rand(sh, sw)
+    img = Image.fromarray((noise * 255).astype(np.uint8), "L")
+    img = img.resize((w, h), Image.BILINEAR)
+
+    return img
+
+
+def build_stack_xml(w, h, groups):
+    image = ET.Element(
+        "image",
+        {"version": "0.0.3", "w": str(w), "h": str(h)}
+    )
+
+    root_stack = ET.SubElement(image, "stack")
+
+    for group in groups:
+
+        group_el = ET.SubElement(
+            root_stack,
+            "stack",
+            {"name": group["name"]}
+        )
+
+        for layer in group["layers"]:
+            ET.SubElement(
+                group_el,
+                "layer",
+                {
+                    "name": layer["name"],
+                    "src": layer["src"],
+                    "opacity": "1.0"
+                }
+            )
+
+    return ET.tostring(image, encoding="utf-8", xml_declaration=True)
+
+
+def build_ora(input_png, output_ora):
+
+    base = Image.open(input_png).convert("RGBA")
+    w, h = base.size
+
+    tmp = tempfile.mkdtemp()
+
+    try:
+
+        data_dir = os.path.join(tmp, "data")
+        thumb_dir = os.path.join(tmp, "Thumbnails")
+
+        os.makedirs(data_dir)
+        os.makedirs(thumb_dir)
+
+        groups = []
+
+        for i in range(NUM_LAYERS):
+
+            group_layers = []
+
+            base_path = f"data/layer_{i}.png"
+            base.save(os.path.join(tmp, base_path))
+
+            group_layers.append({
+                "name": "base",
+                "src": base_path
+            })
+
+            for j in range(MASKS_PER_LAYER):
+
+                mask = noise_mask(w, h, NOISE_SCALES[j])
+
+                mask_path = f"data/layer_{i}_mask_{j}.png"
+                mask.save(os.path.join(tmp, mask_path))
+
+                group_layers.append({
+                    "name": f"mask {j}",
+                    "src": mask_path
+                })
+
+            groups.append({
+                "name": f"group {i}",
+                "layers": group_layers
+            })
+
+        stack_xml = build_stack_xml(w, h, groups)
+
+        with open(os.path.join(tmp, "stack.xml"), "wb") as f:
+            f.write(stack_xml)
+
+        with open(os.path.join(tmp, "mimetype"), "w") as f:
+            f.write("image/openraster")
+
+        base.save(os.path.join(tmp, "mergedimage.png"))
+
+        thumb = base.copy()
+        thumb.thumbnail((256, 256))
+        thumb.save(os.path.join(thumb_dir, "thumbnail.png"))
+
+        with zipfile.ZipFile(output_ora, "w") as z:
+
+            z.write(
+                os.path.join(tmp, "mimetype"),
+                "mimetype",
+                compress_type=zipfile.ZIP_STORED
+            )
+
+            for root, _, files in os.walk(tmp):
+                for file in files:
+
+                    if file == "mimetype":
+                        continue
+
+                    full = os.path.join(root, file)
+                    rel = os.path.relpath(full, tmp)
+
+                    z.write(full, rel)
+
+    finally:
+        shutil.rmtree(tmp)
+
+
+def main():
+
+    parser = argparse.ArgumentParser(
+        description="Generate ORA with grouped layers and mask layers"
+    )
+
+    parser.add_argument("input_png")
+    parser.add_argument("output_ora")
+
+    args = parser.parse_args()
+
+    build_ora(args.input_png, args.output_ora)
+
+
+if __name__ == "__main__":
+    main()

tools/requirements.txt

@@ -1,3 +1,4 @@
 opencv-python
 numpy
 shapely
+Pillow>=10.0.0