archive/blender-archive
Archived
1
1
Fork 0
Code Pull Requests Packages Activity

saveRenderedImage is broken? - isnt working anymore for some resion. switched back to renderAnim.

Browse Source 
More efficient texture usage, packer now rotates the convex hull of the UV's for each image to fit the most image into the smallest rectangle.
This commit is contained in:
Campbell Barton
2006-05-28 10:48:50 +00:00
parent 98b2e98c79
commit e7767e39bb
1 changed files with 112 additions and 16 deletions
Download Patch File Download Diff File Expand all files Collapse all files

128
release/scripts/uv_auto_layout_tex.py
View File

@@ -43,19 +43,82 @@ This is usefull for game models where 1 image is faster then many, and saves the
# Function to find all the images we use
import Blender as B
import boxpack2d
from Blender.Mathutils import Vector
from Blender.Mathutils import Vector, RotationMatrix
from Blender.Scene import Render
import BPyMathutils
'''
RotMatStepRotation = []
rot_angle = 22.5
while rot_angle > 0.1:
RotMatStepRotation.append([\
(rot_angle, RotationMatrix( rot_angle, 2)),\
(-rot_angle, RotationMatrix( -rot_angle, 2))])
rot_angle = rot_angle/2.0
'''
def pointBoundsArea(points):
x= [p.x for p in points] # Lazy use of LC's
y= [p.y for p in points]
return (max(x)-min(x)) * (max(y)-min(y))
def bestBoundsRotation(current_points):
current_area= pointBoundsArea(current_points)
main_rot_angle= 0.0
rot_angle= 45.0
while rot_angle > 0.1:
mat_pos= RotationMatrix( rot_angle, 2)
mat_neg= RotationMatrix( -rot_angle, 2)
new_points_pos= [v*mat_pos for v in current_points]
new_points_neg= [v*mat_neg for v in current_points]
area_pos= pointBoundsArea(new_points_pos)
area_neg= pointBoundsArea(new_points_neg)
# Works!
#print 'Testing angle', rot_angle, current_area, area_pos, area_neg
best_area= min(area_pos, area_neg, current_area)
if area_pos == best_area:
current_area= area_pos
current_points= new_points_pos
main_rot_angle+= rot_angle
elif area_neg == best_area:
current_area= area_neg
current_points= new_points_neg
main_rot_angle-= rot_angle
rot_angle *= 0.5
# Return the optimal rotation.
return main_rot_angle
BIGNUM= 1<<30
class faceGroup(object):
__slots__= 'xmax', 'ymax', 'xmin', 'ymin',\
'image', 'faces', 'box_pack', 'size'\
'image', 'faces', 'box_pack', 'size', 'ang', 'rot_mat', 'cent'\
def __init__(self, mesh_list, image, size, PREF_IMAGE_MARGIN):
self.image= image
self.size= size
# Find the best rotation.
all_points= [Vector(uv) for me in mesh_list for f in me.faces for uv in f.uv if f.image==image]
boundry_indicies= BPyMathutils.convexHull(all_points)
bountry_points= [all_points[i] for i in boundry_indicies]
# Yay this works.
self.ang= bestBoundsRotation(bountry_points)
self.rot_mat= RotationMatrix(self.ang, 2), RotationMatrix(-self.ang, 2),
#print 'ANGLE', image.name, ang
# Add to our face group and set bounds.
xmin=ymin= BIGNUM
# Find the centre
xmin=ymin= BIGNUM
xmax=ymax= -BIGNUM
self.faces= []
for me in mesh_list:
@@ -63,15 +126,33 @@ class faceGroup(object):
if f.image==image:
self.faces.append(f)
for uv in f.uv:
#uv= uv * self.rot_mat
xmax= max(xmax, uv.x)
xmin= min(xmin, uv.x)
ymax= max(ymax, uv.y)
ymin= min(ymin, uv.y)
self.cent= Vector((xmax-xmin)/2, (ymax+ymin)/2 )
# now get the bounds after rotation about the cent.
xmin=ymin= BIGNUM
xmax=ymax= -BIGNUM
for f in self.faces:
for uv in f.uv:
uv= ((uv-self.cent) * self.rot_mat[0]) + self.cent
xmax= max(xmax, uv.x)
xmin= min(xmin, uv.x)
ymax= max(ymax, uv.y)
ymin= min(ymin, uv.y)
# The box pack list is to be passed to the external function "boxpack2d"
# format is ID, w,h
# Store the bounds, impliment the margin.
# The bounds rect will need to be rotated to the rotation angle.
self.xmax= xmax + (PREF_IMAGE_MARGIN/size[0])
self.xmin= xmin - (PREF_IMAGE_MARGIN/size[0])
self.ymax= ymax + (PREF_IMAGE_MARGIN/size[1])
@@ -116,8 +197,9 @@ class faceGroup(object):
for f in self.faces:
for uv in f.uv:
uv.x= offset_x+ (((uv.x-self.xmin) * self.size[0])/width)
uv.y= offset_y+ (((uv.y-self.ymin) * self.size[1])/height)
uv_rot= ((uv-self.cent) * self.rot_mat[0]) + self.cent
uv.x= offset_x+ (((uv_rot.x-self.xmin) * self.size[0])/width)
uv.y= offset_y+ (((uv_rot.y-self.ymin) * self.size[1])/height)
def auto_layout_tex(mesh_list, scn, PREF_IMAGE_PATH, PREF_IMAGE_SIZE, PREF_KEEP_ASPECT, PREF_IMAGE_MARGIN): #, PREF_SIZE_FROM_UV=True):
@@ -156,13 +238,18 @@ def auto_layout_tex(mesh_list, scn, PREF_IMAGE_PATH, PREF_IMAGE_SIZE, PREF_KEEP_
# RENDER THE FACES.
render_scn= B.Scene.New()
render_scn.makeCurrent()
render_context= render_scn.getRenderingContext()
render_context.endFrame(1)
render_context.startFrame(1)
render_context.currentFrame(1)
render_context.setRenderPath(PREF_IMAGE_PATH)
# Set the render context
PREF_IMAGE_PATH_EXPAND= B.sys.expandpath(PREF_IMAGE_PATH+'#') + '.png'
PREF_IMAGE_PATH_EXPAND= B.sys.expandpath(PREF_IMAGE_PATH) + '.png'
# TEST THE FILE WRITING.
'''
try:
# Can we write to this file???
f= open(PREF_IMAGE_PATH_EXPAND, 'w')
@@ -170,12 +257,10 @@ def auto_layout_tex(mesh_list, scn, PREF_IMAGE_PATH, PREF_IMAGE_SIZE, PREF_KEEP_
except:
B.Draw.PupMenu('Error: Could not write to path|' + PREF_IMAGE_PATH_EXPAND)
return
'''
render_context= render_scn.getRenderingContext()
render_context.imageSizeX(PREF_IMAGE_SIZE)
render_context.imageSizeY(PREF_IMAGE_SIZE)
render_context.startFrame(1)
render_context.endFrame(1)
render_context.enableOversampling(True)
render_context.setOversamplingLevel(16)
render_context.setRenderWinSize(100)
@@ -184,7 +269,7 @@ def auto_layout_tex(mesh_list, scn, PREF_IMAGE_PATH, PREF_IMAGE_SIZE, PREF_KEEP_
render_context.enableSky() # No alpha needed.
render_context.enableRGBColor()
Render.EnableDispView() # Broken??
#Render.EnableDispView() # Broken??
# New Mesh and Object
render_mat= B.Material.New()
@@ -193,7 +278,7 @@ def auto_layout_tex(mesh_list, scn, PREF_IMAGE_PATH, PREF_IMAGE_SIZE, PREF_KEEP_
render_me= B.Mesh.New()
render_me.verts.extend([Vector(0,0,0)]) # Stupid, dummy vert, preverts errors.
render_me.verts.extend([Vector(0,0,0)]) # Stupid, dummy vert, preverts errors. when assigning UV's/
render_ob= B.Object.New('Mesh')
render_ob.link(render_me)
render_scn.link(render_ob)
@@ -265,17 +350,28 @@ def auto_layout_tex(mesh_list, scn, PREF_IMAGE_PATH, PREF_IMAGE_SIZE, PREF_KEEP_
target_face.uv[0].y= target_face.uv[3].y= fg.ymin
target_face.uv[1].y= target_face.uv[2].y= fg.ymax
for uv in target_face.uv:
uv_rot= ((uv-fg.cent) * fg.rot_mat[1]) + fg.cent
uv.x= uv_rot.x
uv.y= uv_rot.y
# VCOLS
# Set them white.
for c in target_face.col:
c.r= c.g= c.b= 255
render_context.render()
render_context.saveRenderedImage(PREF_IMAGE_PATH_EXPAND)
#render_context.render()
render_context.renderAnim()
Render.CloseRenderWindow()
#print 'attempting to save an image', PREF_IMAGE_PATH_EXPAND
#render_context.saveRenderedImage(PREF_IMAGE_PATH_EXPAND)
#if not B.sys.exists(PREF_IMAGE_PATH):
# raise 'Error!!!'
Render.CloseRenderWindow()
# NOW APPLY THE SAVED IMAGE TO THE FACES!
#print PREF_IMAGE_PATH_EXPAND