DASAI Mochi Expression Converter (u8g2 Bitmap Converter)

POV: I wanted my own DASAI mochi expression converter to make things easier. With this tool you can convert new expressions, compare with working ones, and generally make the workflow smoother when drawing / iterating on expressions.

This app converts a 128×64 black/white image into an Arduino PROGMEM bitmap array that matches u8g2.drawXBMP byte packing.

Check my post to figure out more: https://marcellomorettoni.wordpress.com/2025/12/25/can-you-use-createstudio-for-arduino-projects/

---

Repository tour

File	What it’s for
single_frame_converter.py	Minimal single-frame converter. Upload one PNG/JPG (resized to 128×64), tweak the frame index, and copy the generated const unsigned char ... PROGMEM bitmap. Useful for quick “does this frame pack correctly?” checks.
video_to_bitmap_gui.py	Full-featured Gradio app with tabs for single frames and whole-video imports. Includes threshold and inversion controls, packing/preview options (so you can pick whether white or black is “on”), hex snippets, bit statistics, and a C-array renderer for pasted code. The “Video to Animation” tab exports every frame as PNG plus a .c file with all arrays.
dasai-mochi-gen.py	THIS IS THE MOST IMPORTANT ONE - Animation packager for Arduino. Upload intro, idle, and expression videos (or add your own named extras) and it emits a face.ino file containing bitmap arrays and animation tables. BUILT FOR XIAO RP2040
adrafruit-translator.py	SSD1306 “emulator” that draws/animates eyes (blink, happy, sad, angry, tired, heart, gaze shifts, sleeping) and exports them as an MP4. Lets you preview motion without flashing a board.
eye-generator.py	Eye transition animator. Lets you design a start pose and an end pose (openness, lids, brow, tilt, gaze, spacing, roundness) and exports the interpolated MP4 so you can preview a custom eye movement before baking it into your Arduino assets.
sad.png	Example 128×64 expression you can load into the converters to verify alignment and contrast.

---

Example image (128×64 canvas)

Use this repo’s example as a reference for sizing and style:

Important: The expression canvas is 128×64. You can upload other sizes (it will resize), but you’ll get the best results if you draw on a true 128×64 canvas to avoid distortion.

---

How the converters work

When you upload an image, the UI produces:

1. Binarized (packed) preview (thresholded to pure B/W)
2. OLED preview that mirrors how u8g2.drawXBMP will render it
3. Arduino bitmap code as a const unsigned char ... PROGMEM array

White pixels are treated as ON, black pixels as OFF.

• conv.py always packs white = ON.
• conv3.py lets you choose which color is ON for both packing and previewing (so you can mirror OLED behaviors that use inverted colors).

---

Running the apps locally

All scripts are standalone; run only the one you need. They all launch a Gradio UI in the browser (by default on http://127.0.0.1:7860).

Install requirements first:

pip install gradio numpy pillow opencv-python imageio

Then launch one of the tools:

• Single-frame converter (fastest path):

  python conv.py

• Converter + debugger + video importer:

  python conv3.py

• Arduino animation generator (face.ino builder):

  python dasay-mochi-gen.py

• SSD1306 → MP4 eye animator:

  python adrafruit-translator.py

• Eye transition animator (start/end pose to MP4):

  python eye-generator.py

If you run into port conflicts, add --server.port 7861 (or any free port) to the python ... command.

---

Tips for best results

• Start with true 128×64 artwork to avoid rescaling artifacts.
• Use high-contrast black/white images; avoid mid-gray edges that might wobble across a threshold.
• In conv3.py, toggle Invert or Packing: WHITE = ON if your OLED uses opposite polarity.
• For animations, keep frame counts modest—Arduino flash fills up quickly. Use the hex snippet + bit stats panels to spot heavy frames.

---