- Background image would jitter around on ATI card when zoomed in too
far. It appears that the driver is was multiplying the image width
by the zoom factor at some point and clamping this number, before
clipping the visible image. Somehow this then fed back in to clipping
the zoom factor.
Fix is to only tell GL to draw the smallest number of pixels (width
and height) that would be visible on the screen. Since this is not
a generally bad thing to do applied fix for all users of
glaDrawPixelsSafe.
- restored 'Warp' (Shift+W). Should work like before, including type mode
and holding ctrl/shift modifiers
- added CTRL event in queue, to only redraw when you press it. looks nicer.
+ Refined the headerprint for Translation. Now prints only the needed info for constraint in the constraint's space (ie: if you're moving 1 unit along the local X axis, regardless of it's orientation, it will print "D: 1.000 along local X")
Still need to make numinput work like that (typing a number with a local axis constraint would move along that axis. There's some base code already though, just need a finishing touch, but it's late now)
+ Optimised PET calculations by using the TD_NOACTION flag (actually, that might have been in the last commit).
+ Added a float axismtx[3][3] member to TransData to store the orientation of the element (useful for local axis constrainst which, in edit could be moving along normals and the like).
- Fixed scaling in edit mode (was doing some matrix multiplications in the wrong order, only visible when using a constraint)
- Fixed the constraint projection matrix. It didn't work for planar constraint if the constraint space wasn't global (in a nutshell, it produced weird results for local space planes).
- Some potential bugs fixed (Note to Ton: added an ext pointer in TransInfo to point to the TransDataExtension block. With the sort done after allocation, the first td pointer doesn't necesarely point at the start of the ext block, so we needed another to free it correctly).
- Got rid of some remaining test with G.obedit.
- Moved constraint reset from init to post trans code (Ton, that means you can create constraints before calling transform, like for the menus for example).
NOTE:
I was getting some random segfault with the new headerprint code. Very random, couldn't reproduce with a debug version. I did some initialisation that might have been missing (though doubtful that's what caused the crashes). Was linked to using constraint though not caused by them. Probably due to some dumb late coding error.
Bumpmaps for skin on Env's dinosaurs appeared to be less nice in 2.36. This
was caused by the bugfix to make bumpmapping correct for rotations, which
should only work for flat/cube mapping. It also rotated it for sphere/tube
though, which gives less interesting bumps.
So; now the correction is skipped for tube/sphere mapping bumping.
Wood/marble now have three waveforms to choose from: Sine, Saw and Triangle.
The Saw wave allows for much more realistic wood, especially in combination
with a ColorBand. A blender3d.org release page is being constructed about it.
Added: commit in editmesh_add.c to remove circle warning in face-select mode.
Works with edit data and in object mode with single selections (only one object selected).
Also started adding constraint stuff in headerprints. Only for Translation for now.
Pressing X,Y,Z (and the Ctrl versions) toggle between global, local and off like it used to do.
- made generic 'calc distance' function for it
- added generic call to sort TransData with selection first, for speedup
of propmode calculus
- removed most propmode exceptions from code, only used for counting now
- all editmode transdata conversion function structured identical
The dynamical allocated arrays (rwenzlaff, april 03) for render vertices/
faces/halos were freed in a way that requires the last element of array to
be NULL.
This wasn't coded however, causing memory errors in exceptional cases;
like when amount of faces/vertices is exact between 261888 and 262144. :)
Found (old) issue in render... the "VlakRen" (render face) has an Object
pointer which can point to a duplicated one (dupliframes/dupliverts option),
which is freed before actual render starts... this nicely crashes in Windows.
I now store the original Object pointer, which can lead to texture
orientation errors in some situations (like normal-bump stuff).
Real solution is recode of duplicator system!
Environment maps type "Load" don't need to have an "Ob" object set to
render. Without object it renders applied envmaps like reflection maps.
Please note that envmaps then still "mirror" realistic, provided the
original object the envmap was rendered with was not rotated.
Made a generic snapGrid function. Now, each transform needs to define it's snapping parameters in their init. This is needed because of the constraint branching.
Renamed the constraints defines to CON_*
Renamed the fonctions used to interactively select a constraint (with MMB). The previous naming scheme was obscure at best.
Added a CON_SELECT flag when selecting interactively. (Following the idea of the patch joeedh submitted).
New behavior when selecting a constraint interactively, not only does it highlight the axis that will be selected, it also makes it the current constraint.
Holding down Ctrl when selecting a constraint with MMB now does the same as hitting Ctrl-AXIS. That is, it locks that axis (planar constraint on the normal plane).
- - - - -
To all those who posted suggestions, I'm not forgetting them:
- broken / desoto with the plane draw thingy,
- RobertT's push/pull transformation (that was from before the new code even)
- UnNamed's MMB interactive idea as we discussed on IRC
and all the others that I've noted down somewhere.
Next for today, I'll have a look at local constraint (object's axis).
If the startup file ".Blanguages" is not found, Blender now will only printf
a warning if started in debug mode (-d). It used to popup a menu, even
before the UI was initialized, causing annoyance... it's irrelevant info.
Pressing AKEY in channel buttons (names) of IpoWindow did not show all or
hide all anymore. Was due to commit last july, which enabled unlimited
amount of channels.
- Gert de Roost reported an inconsistency between nmesh.getMode and .setMode. Now .setMode() optionally accepts an int value, as returned by getMode().
- Campbell Barton pointed that object.getData(name_only=True) was by mistake returning the obj name, not the obdata name, as it should. Fixed now.
- small doc updates
Thanks both for the reports.
In user terms: the motion on screen of the selection follows the motion of the mouse pointer.
Gives some errors when the constraint plane is nearly perpendicular to the view port though.
Added a debug print function for 4D vectors to arithb.c
Optimised the 3D -> view projection functions in view.c (a bit).
- added texture space grab/scale (TKEY objectmode)
- made new transform work with menus (meaning, dropping dreaded while-hold)
To Martin & other while-hold lovers: this needs to be carefully thought over
and designed. I prefer to look on this within context of making transform
fully tablet/pen friendly, as option. Aim now for me is still: get transform
back to work! :)
Brought back functionality to work with "Ipo Keys" (Kkey in 3d window). New
transform code is looking more messy now... its design was not purposed to
clean on this level. Acceptable for this stage however.
Also renamed old variables *tob into *td.
Changed the meaning of mtx and smtx for objects. It's now uniform accross the board.
Added TD_OBJECT as a TransData flag. Objects is requiring exception code here and there, use a flag instead of G.obedit. Will document what the exceptions are for people who wants to add more transformations.
Split TransData in two.
Note to Ton:
You'll have to change posemode to use the new structure. I've left some variables as part of TransData to have a compilable version in CVS, remove them when you stop using them.
Check MetaElement and Object conversion for example on how to use TransDataExtension (though I'm sure you can figure it out by yourself).
And that's it. Not much coding time in the week days. :\
