diff --git a/lua.c b/lua.c
index f754987..0c9b9da 100644
--- a/lua.c
+++ b/lua.c
@@ -83,6 +83,7 @@ struct icelua_entry icelua_client[] = {
 	{icelua_fn_client_model_render_bone_global, "model_render_bone_global"},
 	{icelua_fn_client_model_render_bone_local, "model_render_bone_local"},
 	{icelua_fn_client_img_blit, "img_blit"},
+	{icelua_fn_client_img_blit_to, "img_blit_to"},
 	{NULL, NULL}
 };
 
diff --git a/lua_image.h b/lua_image.h
index 6f394a8..14e546c 100644
--- a/lua_image.h
+++ b/lua_image.h
@@ -151,3 +151,31 @@ int icelua_fn_common_img_get_dims(lua_State *L)
 	
 	return 2;
 }
+
+int icelua_fn_client_img_blit_to(lua_State *L)
+{
+	int top = icelua_assert_stack(L, 4, 9);
+	int dx, dy, bw, bh, sx, sy;
+	uint32_t color;
+	
+	img_t *dest = lua_touserdata(L, 1);
+	if(dest == NULL || dest->udtype != UD_IMG)
+		return luaL_error(L, "not an image");
+	img_t *source = lua_touserdata(L, 2);
+	if(source == NULL || source->udtype != UD_IMG)
+		return luaL_error(L, "not an image");
+	
+	dx = lua_tointeger(L, 3);
+	dy = lua_tointeger(L, 4);
+	bw = (top < 5 ? source->head.width : lua_tointeger(L, 5));
+	bh = (top < 6 ? source->head.height : lua_tointeger(L, 6));
+	sx = (top < 7 ? 0 : lua_tointeger(L, 7));
+	sy = (top < 8 ? 0 : lua_tointeger(L, 8));
+	color = (top < 9 ? 0xFFFFFFFF : (uint32_t)lua_tointeger(L, 9));
+	
+	render_blit_img(dest->pixels, dest->head.width, dest->head.height, 
+        dest->head.width,
+		source, dx, dy, bw, bh, sx, sy, color);
+	
+	return 0;
+}
diff --git a/pkg/base/icegui/widgets.lua b/pkg/base/icegui/widgets.lua
index 33b46df..5147bbe 100644
--- a/pkg/base/icegui/widgets.lua
+++ b/pkg/base/icegui/widgets.lua
@@ -1,12 +1,4 @@
--- look into actual drawing functionality!~
--- the client drawing code requires manual memory management:
--- we have to allocate a buffer and draw pixels to it
--- this code can't deal with that problem...
--- lib_gui will have to provide a layer that takes the abstract APIs here and adds drawing functions
--- on top. the abstract API can assist by adding a "dirty" flag so that cache management is straightforward.
-
 -- sketch listener and collision system:
---    rect and layers detection (derive layers from hierarchy)
 --    onDown onUp onMove onClick(down+up inside collision) onDrag(down inside collision, movement) onKeyboard
 --    when will mouse cursor be visible? important engine consideration!
 
diff --git a/pkg/base/lib_gui.lua b/pkg/base/lib_gui.lua
index 1a40d6c..fb858da 100644
--- a/pkg/base/lib_gui.lua
+++ b/pkg/base/lib_gui.lua
@@ -190,20 +190,27 @@ function gui_create_fixwidth_font(image, char_width, char_height, indexing_fn)
 	
 	-- print text with topleft at x, y, color c, string str
 	function this.print(x, y, c, str, buffer)
-		buffer = buffer or client
-		local i
 		for i=1,#str do
 			local idx = this.indexing_fn(string.byte(str, i))
-			buffer.img_blit(this.image, x, y, this.width, this.height, idx*this.width, 0, c)
+			if buffer == nil then
+				local temp = common.img_new(this.width, this.height)
+				client.img_blit_to(temp, this.image, 0, 0, this.width, this.height, 
+					idx*this.width, 0, c)
+				common.img_pixel_set(temp, 0, 0, 0xFFFF0000)
+				client.img_blit(temp, x, y)
+				common.img_free(temp)
+				--client.img_blit(this.image, x, y, this.width, this.height, idx*this.width, 0, c)
+			else
+				client.img_blit_to(buffer, this.image, x, y, this.width, this.height, idx*this.width, 0, c)
+			end
 			x = x + this.width
 		end
+		local i
 	end
 	
 	-- print a selection of precomputed text
 	function this.print_precomputed(data, offx, offy, buffer)
 		
-		buffer = buffer or client
-		
 		for y=1,#data do
 			for x=1,#data[y] do
 				local char = data[y][x][1]
@@ -211,8 +218,13 @@ function gui_create_fixwidth_font(image, char_width, char_height, indexing_fn)
 				local px = data[y][x][3] + offx
 				local py = data[y][x][4] + offy
 				local c = data[y][x][5]
-				buffer.img_blit(this.image, px, py, this.width, this.height, 
-					idx*this.width, 0, c)
+				if buffer == nil then
+					client.img_blit(this.image, px, py, this.width, this.height, 
+						idx*this.width, 0, c)
+				else
+					client.img_blit_to(buffer, this.image, px, py, this.width, this.height, 
+						idx*this.width, 0, c)
+				end
 			end
 		end
 		
@@ -354,24 +366,83 @@ function gui_create_scene(width, height)
 	end
 	
 	function scene.textfield(options)
-		-- store a text buffer in here and its last render state in here...
-		-- define the width and height to be the width and height of the text...
-		-- additional formatting?
 		
 		local this = scene.display_object(options)
 		
-		-- if we specified width and height in options, we are implicitly using wordwrapping?
-		-- no... 
+		this.wordwrap = options.wordwrap or true
+		this.color = options.color or 0xFF88FF88
+		this.autosize = options.autosize or true
+		this.font = options.font or font_mini
+		this.use_img = true
+		
+		-- so, we are computing the text around offset 0, 0
+		-- but when we go to display or collide with it, we're going to have to
+		-- apply the widget offsets on top.
+		
+		-- TODO: cursor + text selection collision, data structure, rendering
+		-- TODO: compute_unwrapped should allow new lines...
+		-- TODO: check to see if the draw-to-buffer method I made is actually any good
+		
+		function this.setter_keys.width(w)
+			if this.autosize == false then
+				rawset(this, 'width', w)
+			end
+		end
+		
+		function this.setter_keys.height(h)
+			if this.autosize == false then
+				rawset(this, 'height', h)
+			end
+		end
+		
+		function this._recalc_size()
+			if this.wordwrap == true then
+				this.text_cache = this.font.compute_wordwrap(this.width,
+					0, 0, this.color, this._text)
+			else
+				this.text_cache = this.font.compute_unwrapped(0, 0, 
+					this.color, this._text)
+			end
+			if this.autosize then
+				local dim = this.font.dimensions(this.text_cache)
+				rawset(this, 'width', dim.width)
+				rawset(this, 'height', dim.height)
+			end
+			this.dirty = true
+		end
+		
+		function this.getter_keys.text()
+			return this._text
+		end
+		
+		function this.setter_keys.text(str)
+			this._text = str
+			this._recalc_size()
+		end
+		
+		function this.draw_update()
+			this.font.print_precomputed(this.text_cache, 0, 0, this.img)
+			common.img_pixel_set(this.img, 0, 0, 0xFFFF0000)
+			common.img_pixel_set(this.img, 1, 0, 0xFFFF0000)
+			common.img_pixel_set(this.img, 0, 1, 0xFFFF0000)
+			common.img_pixel_set(this.img, 1, 1, 0xFFFF0000)
+		end
+		
+		this.text = options.text or ""
+		
+		return this
 		
 	end
 	
 	-- TEST CODE
-	--local frame = scene.rect_frame{width=320,height=320, x=width/2, y=height/2}
-	--local frame2 = scene.rect_frame{width=32,height=32, x=0, y=0}
-	--local frame3 = scene.rect_frame{width=32,height=32, x=64, y=96}
+	local frame = scene.rect_frame{width=320,height=320, x=width/2, y=height/2}
+	local frame2 = scene.rect_frame{width=32,height=32, x=0, y=0}
+	local frame3 = scene.rect_frame{width=32,height=32, x=64, y=96}
+	local text1 = scene.textfield{width=400,height=100, text="hello world"}
 	--root.add_child(frame)
 	--frame.add_child(frame2)
 	--frame.add_child(frame3)
+	root.add_child(text1)
 	
 	return scene