Miskeenity Canvas

Cross platform 2d infinite canvas app inspired by dingboard.com.

Web version available to try here (chromium browsers only): https://sava41.github.io/miskeenity-canvas/

Features

Paint

Add Text

Add and Edit Images

AI Image Segmentation (Desktop Only)

Requirements:

• CMake 3.28 or later
• Python 3.8 or newer
• cpp compiler

Build for Desktop:

> cmake -B build -DCMAKE_BUILD_TYPE=<type>
> cmake --build build

currently there are issues building on macos

Using SAM for Image Segmentation

To use AI image segmentation you need to download the models and place them in the miskeenity canvas executable folder. These are included in the release downloads or can be generated from the official SAM repository. sam_vit_h_4b8939.onnx can be exported using the steps found here https://github.com/facebookresearch/segment-anything#onnx-export. The preprocess model sam_preprocess.onnx needs to be exported using an unofficial script such as the one found here (thank you dinglufe) https://github.com/dinglufe/segment-anything-cpp-wrapper/blob/main/export_pre_model.py

Before using the models also make sure the ONNX runtime dynamic library is present in executable folder.

Build WASM/HTML version via Emscripten (Linux, macOS)

Setup the emscripten SDK as described here:

https://emscripten.org/docs/getting_started/downloads.html#installation-instructions

> git clone https://github.com/emscripten-core/emsdk.git
> cd emsdk
> git pull
> ./emsdk install latest
> ./emsdk activate latest
> cd ..

> emsdk/upstream/emscripten/emcmake cmake . -B embuild -DCMAKE_BUILD_TYPE=<type>
> cmake --build embuild

Planned Features

• canvas history (undo/redo)
• copy/paste/clipboard
• keyboard shortcuts
• rendering improvements (anti-aliasing, better font rendering)