Re: [Bf-committers] Patches Submitted
On Wed, Jul 20, 2011 at 3:37 PM, Scott Giese scott.gi...@comcast.net wrote: Hi Gang, FYI. I submitted 3 patches for your review. I'm new to the list and I wanted to give back to the Blender community. 28030 SCONS Build: Build Date reflects http://projects.blender.org/tracker/index.php?func=detailaid=28030group_i d=9atid=127 1 instead of actual date of build 28031 Minor typo in Blenlib http://projects.blender.org/tracker/index.php?func=detailaid=28031group_i d=9atid=127 28032 Python Mathutils: Matrix Multiplication Error http://projects.blender.org/tracker/index.php?func=detailaid=28032group_i d=9atid=127 Great work guys! Appreciate the great product. Scott Thanks for the fixes, committed all patches however you're changes to mathutils effectively only change the order of multiplication, http://projects.blender.org/tracker/index.php?func=detailaid=28032group_id=9atid=127 In you're example print (m1 * m2) Change to... print (m2 * m1) This is a bit confusing because in C we have mul_m4_m4m4(m1, m2); which is the equivalent to m2 * m1 in python. A while back Benoit Bolsee was concerned our matrix multiplication order was wrong so we switched it (between 2.4x and 2.5x) What makes you think the order in blender is wrong? what's you're reference? Just checked and we're currently doing matrix multiplication differently to numpy which doesn't bode well :S - test: # --- snip m1 = ((0.0, 0.0, 1.0, 0.0), (-1.0, 0.0, 0.0, 0.0), (0.0, -1.0, 0.0, 0.0), (0.6, 0.0, -0.05, 1.0)) m2 = ((1.0, 0.0, 0.0, 0.0), (0.0, 1.0, 0.0, 0.0), (0.0, 0.0, 1.0, 0.0), (0.0, -0.02, -0.1, 1.0)) from numpy import matrix n_m1 = matrix(m1) n_m2 = matrix(m2) print(\nnumpy\n%r % (n_m1 * n_m2)) from mathutils import Matrix b_m1 = Matrix(m1) b_m2 = Matrix(m2) print(\nmathutils\n%r % (b_m1 * b_m2)) # --- output numpy matrix([[ 0. , 0. , 1. , 0. ], [-1. , 0. , 0. , 0. ], [ 0. , -1. , 0. , 0. ], [ 0.6 , -0.02, -0.15, 1. ]]) mathutils Matrix((0.0, 0.0, 1.0, 0.0), (-1.0, 0.0, 0.0, 0.0), (0.0, -1.0, 0.0, 0.0), (0.62, 0.1, -0.05, 1.0)) # --- switch m1/m2 order for both mathutils and numpy. re-run numpy matrix([[ 0. , 0. , 1. , 0. ], [-1. , 0. , 0. , 0. ], [ 0. , -1. , 0. , 0. ], [ 0.62, 0.1 , -0.05, 1. ]]) mathutils Matrix((0.0, 0.0, 1.0, 0.0), (-1.0, 0.0, 0.0, 0.0), (0.0, -1.0, 0.0, 0.0), (0.6, -0.0, -0.15, 1.0)) ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
Re: [Bf-committers] Patches Submitted
I've needed to try and figure this one out too so here's my two cents and some random matrices. Since Blender uses column major order matrices these printouts are actually visually transposed to normal math, actual matrix columns are shown horizontally in the inner brackets and rows are vertical. So the 2.5 in a*b ((a*b)[2][0] - column 2, row 0) is a result of dot produt (a row 0, b column 2) as it should be - dot( (1.0, 0.0, 1.0, 0.0), (0.5, 1.0, 2.0, 0.0) ). a Matrix((1.0, 0.0, 0.0, 0.0), (0.0, 1.0, 0.0, 0.0), (1.0, 0.0, -1.0, 0.0), (0.0, 0.0, 0.0, 1.0)) b Matrix((1.0, 0.0, 0.0, 0.0), (0.0, 1.0, 0.0, 0.0), (0.5, 1.0, 2.0, 0.0), (0.0, 0.0, 0.0, 1.0)) a*b Matrix((1.0, 0.0, 0.0, 0.0), (0.0, 1.0, 0.0, 0.0), (2.5, 1.0, -2.0, 0.0), (0.0, 0.0, 0.0, 1.0)) b*a Matrix((1.0, 0.0, 0.0, 0.0), (0.0, 1.0, 0.0, 0.0), (0.5, -1.0, -2.0, 0.0), (0.0, 0.0, 0.0, 1.0)) One nice thing about this is that the columns are the axises (0,1,2) and translation (3) of a matrix, so a camera direction for example would be -a[2][0:3]. 2011/7/20 Campbell Barton ideasma...@gmail.com On Wed, Jul 20, 2011 at 3:37 PM, Scott Giese scott.gi...@comcast.net wrote: Hi Gang, FYI. I submitted 3 patches for your review. I'm new to the list and I wanted to give back to the Blender community. 28030 SCONS Build: Build Date reflects http://projects.blender.org/tracker/index.php?func=detailaid=28030group_i d=9atid=127 1 instead of actual date of build 28031 Minor typo in Blenlib http://projects.blender.org/tracker/index.php?func=detailaid=28031group_i d=9atid=127 28032 Python Mathutils: Matrix Multiplication Error http://projects.blender.org/tracker/index.php?func=detailaid=28032group_i d=9atid=127 Great work guys! Appreciate the great product. Scott Thanks for the fixes, committed all patches however you're changes to mathutils effectively only change the order of multiplication, http://projects.blender.org/tracker/index.php?func=detailaid=28032group_id=9atid=127 In you're example print (m1 * m2) Change to... print (m2 * m1) This is a bit confusing because in C we have mul_m4_m4m4(m1, m2); which is the equivalent to m2 * m1 in python. A while back Benoit Bolsee was concerned our matrix multiplication order was wrong so we switched it (between 2.4x and 2.5x) What makes you think the order in blender is wrong? what's you're reference? Just checked and we're currently doing matrix multiplication differently to numpy which doesn't bode well :S - test: # --- snip m1 = ((0.0, 0.0, 1.0, 0.0), (-1.0, 0.0, 0.0, 0.0), (0.0, -1.0, 0.0, 0.0), (0.6, 0.0, -0.05, 1.0)) m2 = ((1.0, 0.0, 0.0, 0.0), (0.0, 1.0, 0.0, 0.0), (0.0, 0.0, 1.0, 0.0), (0.0, -0.02, -0.1, 1.0)) from numpy import matrix n_m1 = matrix(m1) n_m2 = matrix(m2) print(\nnumpy\n%r % (n_m1 * n_m2)) from mathutils import Matrix b_m1 = Matrix(m1) b_m2 = Matrix(m2) print(\nmathutils\n%r % (b_m1 * b_m2)) # --- output numpy matrix([[ 0. , 0. , 1. , 0. ], [-1. , 0. , 0. , 0. ], [ 0. , -1. , 0. , 0. ], [ 0.6 , -0.02, -0.15, 1. ]]) mathutils Matrix((0.0, 0.0, 1.0, 0.0), (-1.0, 0.0, 0.0, 0.0), (0.0, -1.0, 0.0, 0.0), (0.62, 0.1, -0.05, 1.0)) # --- switch m1/m2 order for both mathutils and numpy. re-run numpy matrix([[ 0. , 0. , 1. , 0. ], [-1. , 0. , 0. , 0. ], [ 0. , -1. , 0. , 0. ], [ 0.62, 0.1 , -0.05, 1. ]]) mathutils Matrix((0.0, 0.0, 1.0, 0.0), (-1.0, 0.0, 0.0, 0.0), (0.0, -1.0, 0.0, 0.0), (0.6, -0.0, -0.15, 1.0)) ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
[Bf-committers] Proposal: better envmap scripting
Hello all. I'm currently writing an add-on to ease the export of renders and environment maps to the Source game engine, and I'm finding my options when it comes to envmaps extremely limited. I'd like to add the following functions: bpy.types.EnvironmentMap.save( filename, layout=( (0,0),(0,1),(0,2),(1,0),(1,1),(1,2) ) ) Saves the envmap with the scene render settings. layout determines the location of each face in the output image; 1 == EnvironmentMap.resolution. bpy.types.EnvironmentMap.render(ignore_cache=False) Renders the envmap if it is stale. Ignores the camera and indeed goes ahead without one. Might render other envmaps for recursion. ignore_cache bypasses the normal envmap cache checks. bpy.types.EnvironmentMap.clear() Same as bpy.ops.texture.envmap_clear(), but without the overhead of setting up then reverting the context. Is there any reason why the above couldn't be added to trunk? ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
[Bf-committers] Material Nodes: No normals in worldspace?
In recent time I worked with node-materials quite often and anything seamed to be fine. But then I came across the very simple task that needs the actual normals of the object surface in worldspace (z-up, x-right, y-backside). But there is no input for such normals. All I have found are the normals in viewspace and the view-vector of the camera (also in viewspace). Since both are only relative to each other, there is no way to get the worldspace normal of an object just by using this pair. That means, that there is no way to influence the shading depending on the orientation (not location) of the faces itself. For example you would like to give any upward looking face an own material and fade it out to a second material, as the angle gets bigger. This can be achieved within viewspace, but only if you never intend to rotate the camera. Tilt it 180° to left or right and the bottom of all objects will show the first material instead of the second. So my questions are: a) Is there a way to access the normals in worldspace/localspace within material nodes? b) If not: Why was it never implemented? Seams to be very useful: Snow depending on surface angle; Trees covered with moss at the rain side, ice at the wings of a plane, etc.) c) Is there any workaround that would also work on animated meshes? Best wishes from Tobias Oelgarte ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
[Bf-committers] The dilation I am seeing in Blender is imo not good. I thought I'd propose an alternative (if there's support)?
The dilation I am seeing in Blender is imo not good. I thought I'd propose an alternative (if there's support)? Here's a close-up of the dilation blender does -- http://jbit.net/~sparky/blender_dial/bakezoom_BI_dial.png Here's a close-up of the dilation I was able to do outside of blender using diff. code -- http://jbit.net/~sparky/blender_dial/bakezoom_other_dial.png The bumped visual you get using the first one is this (ugly filtering scar) -- http://jbit.net/~sparky/blender_dial/scr_BI_dial.png The visual you get using the other method is this (significantly more subtle) -- http://jbit.net/~sparky/blender_dial/scr_other_dial.png The blend file to produce the texture is up there -- http://jbit.net/~sparky/blender_dial/bmake.blend This is the full-res baked result you get with dilation using blender -- http://jbit.net/~sparky/blender_dial/bake_BI.png and this is using the alternative dilation -- http://jbit.net/~sparky/blender_dial/bake_other_dial.png I dumped everything here http://jbit.net/~sparky/blender_dial/ incl. a blender file. Let me know what you think. Cheers, Morten. ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
[Bf-committers] Simple Todos-Straighten
I have created/finished the straighten uvedit tool mentioned at http://wiki.blender.org/index.php/Dev:2.5/Source/Development/Todo/Simple_Todos Please check out Patch [#28045] Straighten tool from Simple Todos ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
Re: [Bf-committers] Patches Submitted
My reasoning for assuming this was a bug: 1. Previously working scripts (2.49) broke. 2. m1 *= m2 is equivalent to m1 = m1 * m2, where m1 represents the Base matrix and m2 represents the Influencing matrix. This is more intuitive to me. There is no equivalent shorthand for m1 = m2 * m1. e.g. Apply a series of transforms to produce a single transformation matrix contextMatrix = mathutils.Matrix() for part in parts: contextMatrix *= data.tran.matrix[part.matrix_index] contextMatrix *= data.tran.matrix[part.parent.matrix_index] ... newObject.matrix_basis = contextMatrix 3. Treating leftMatrix as the Base and rightMatrix as the Influencing facilitates a hypothetical method of varying argument counts. e.g. resultMatrix = Matrix.Combine(baseMatrix, translateMatrix, rotationMatrix, scaleMatrix, ...) The above outlines my thought process. I was not aware that the change was intentional. In light of the ... stop breaking APIs? discussion, we may be better served by leaving it as-is. Scott -Original Message- From: Campbell Barton [mailto:ideasma...@gmail.com] Sent: Wednesday, July 20, 2011 2:16 AM To: bf-blender developers Subject: Re: [Bf-committers] Patches Submitted On Wed, Jul 20, 2011 at 3:37 PM, Scott Giese scott.gi...@comcast.net wrote: Hi Gang, FYI. I submitted 3 patches for your review. I'm new to the list and I wanted to give back to the Blender community. 28030 SCONS Build: Build Date reflects http://projects.blender.org/tracker/index.php?func=detailaid=28030group_i d=9atid=127 1 instead of actual date of build 28031 Minor typo in Blenlib http://projects.blender.org/tracker/index.php?func=detailaid=28031group_i d=9atid=127 28032 Python Mathutils: Matrix Multiplication Error http://projects.blender.org/tracker/index.php?func=detailaid=28032group_i d=9atid=127 Great work guys! Appreciate the great product. Scott Thanks for the fixes, committed all patches however you're changes to mathutils effectively only change the order of multiplication, http://projects.blender.org/tracker/index.php?func=detailaid=28032group_id =9atid=127 In you're example print (m1 * m2) Change to... print (m2 * m1) This is a bit confusing because in C we have mul_m4_m4m4(m1, m2); which is the equivalent to m2 * m1 in python. A while back Benoit Bolsee was concerned our matrix multiplication order was wrong so we switched it (between 2.4x and 2.5x) What makes you think the order in blender is wrong? what's you're reference? Just checked and we're currently doing matrix multiplication differently to numpy which doesn't bode well :S - test: # --- snip m1 = ((0.0, 0.0, 1.0, 0.0), (-1.0, 0.0, 0.0, 0.0), (0.0, -1.0, 0.0, 0.0), (0.6, 0.0, -0.05, 1.0)) m2 = ((1.0, 0.0, 0.0, 0.0), (0.0, 1.0, 0.0, 0.0), (0.0, 0.0, 1.0, 0.0), (0.0, -0.02, -0.1, 1.0)) from numpy import matrix n_m1 = matrix(m1) n_m2 = matrix(m2) print(\nnumpy\n%r % (n_m1 * n_m2)) from mathutils import Matrix b_m1 = Matrix(m1) b_m2 = Matrix(m2) print(\nmathutils\n%r % (b_m1 * b_m2)) # --- output numpy matrix([[ 0. , 0. , 1. , 0. ], [-1. , 0. , 0. , 0. ], [ 0. , -1. , 0. , 0. ], [ 0.6 , -0.02, -0.15, 1. ]]) mathutils Matrix((0.0, 0.0, 1.0, 0.0), (-1.0, 0.0, 0.0, 0.0), (0.0, -1.0, 0.0, 0.0), (0.62, 0.1, -0.05, 1.0)) # --- switch m1/m2 order for both mathutils and numpy. re-run numpy matrix([[ 0. , 0. , 1. , 0. ], [-1. , 0. , 0. , 0. ], [ 0. , -1. , 0. , 0. ], [ 0.62, 0.1 , -0.05, 1. ]]) mathutils Matrix((0.0, 0.0, 1.0, 0.0), (-1.0, 0.0, 0.0, 0.0), (0.0, -1.0, 0.0, 0.0), (0.6, -0.0, -0.15, 1.0)) ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
