LVS - my personal Overlua follower

[Youkakun]

For those who didn't discovered it in Aegisub forum already, i present my personal follower of Overlua here: LVS.

Personally i use it for typesetting, karaoke effects, image creation and some video effects.
Opening AvsPMod -> writing the Avisynth script for video & LVS loading -> writing the LVS input script while refreshing AvsPMod to see the results -> encode Avisynth script.
It's some Lua programming with a graphics kit, but all-in-all easier and more powerful than Overlua (and active).
In comparison, some advantages:

• More up-to-date Lua (v5.2) with utf-8 support on windows
• Simpler and larger API
• Better text support (with windows fonts)
• Advanced ASS parser
• Fast image convolution with arbitrary kernels
• PNG im- & export
• Autoload folder for standard included scripts
• Mesh gradient
• ...

I've already thought about writing a GUI, but using it with AvsP is comfortable enough. Maybe it could be the core of NyuFX 2... i don't know.
Other plugin interfaces shouldn't be a problem and i've already implented a ffmpeg demuxer [1][2] in NyuFX 1.6, so this project could have a future

I tried to write a lot of documentation, like easy usage examples, html documents and many comments in source code.
Feedback would be nice and contributors are welcome.

[fgghjjkll]

Interesting project. I will probably look into this soon.
Though it is easier than OverLua, do you still need to be a masochist to use it effectively?

[Youkakun]

Not really.
A big set of functions, good documentations and lot of commented examples should make it much much easier to learn and use than Overlua. A folder for autoloading scripts serves as external library place, so users can extend the environment and share this way their LVS extensions.

[Youkakun]

LVS took his way to version v1.3, became more powerful + fast, bugs were fixed and tests made by creating typesets and karaoke effects.

One script for a simple karaoke effect could look like this:

-- Parse ASS file and fill global tables
ass.load("my_ktime.ass")

-- Roumaji & Kanji
local function roumaji_kanji(ctx, ms, line)
	-- Iterate through sylables
	for _, syl in ipairs(line.styleref.alignment < 7 and line.chars or line.syls) do	-- Use characters in case of kanjis
		-- Set color dependent on sylable activity
		ctx:set_source(ms >= syl.start_time and ms < syl.end_time and g2du.red or g2du.yellow)
		-- Draw sylable text
		ctx:path_add_text(syl.x, syl.y, syl.text, ass.unpack_font(line.styleref))
		ctx:fill()
		ctx:path_clear()
	end
end

-- Subtitle
local function subtitle(ctx, line)
	-- Draw line text
	ctx:path_add_text(line.x, line.y, line.text, ass.unpack_font(line.styleref))
	ctx:set_source(g2du.yellow)
	ctx:fill()
	ctx:path_clear()
end

-- Process frames
function GetFrame(frame, frame_i)
	-- Create drawing context for frame
	local ctx = frame:get_context()
	-- Get frame time
	local ms = frame_i / VIDEO_FPS * 1000
	-- Look for frame-related ASS lines
	for _, line in ipairs(lines) do
		if ms >= line.start_time and ms < line.end_time then
			-- Draw ASS line
			if line.styleref.alignment > 3 then
				roumaji_kanji(ctx, ms, line)
			else
				subtitle(ctx, line)
			end
		end
	end
end

It's just a color change of active sylables, but imagine to use gradients and pictures, invert frame colors, deform text like clay, use various convolution filters, work with audio data, etc.

More informations:
Online docs
Samples