evinedev / artal Goto Github PK

I think it would make sense to split the API in a few components, like so:

• 'raw' loader - loads PSD file into a rather flat table structure, loads images as ImageData.
  this shouldn't do 'semantic parsing' - e.g. combining / renaming open-close/group layers etc. but stay true to the PSD format for the most part. It should also be safe to run in a thread (= no love.graphics access) and configurable to skip loading imagedatas, or load only the composed info.
• parser - takes the 'raw' output and reorganizes it into a tree that resembles the layer structure in a modern photoshop version - with group nodes and rasterized layers as leaves. Turns the ImageDatas into Images. The tree logic would resemble this code.
• renderer code - provides ways to render layers, layer groups and whole files correctly. I have a complete collection of photoshop blendmode shaders here, but not all of them have been tested. The renderer code would resemble this code.