- User Since
- Sep 14 2008, 10:31 PM (440 w, 2 d)
Mon, Jan 30
unfortunately I didnt find the time to finish/clean this up (and I am not sure when I will find time) so I'm just pasting a very rough diff for others to pick up....
This just tries to solve a couple of issues I found with the clipping code:
- clip minx/miny not working (if center/cursor is 'in right side')
- clip doesnt allow for cursor being outside of UV bounds (or directly on the bounds like in this report) [the paste allows for this]
- when scaling and runnning into clipmax first, then scale further and run into clipmin after --> pop to clipmin
- also wasnt sure if it is a good idea to scale further on other axis after running into one clipping border [making the scale non-uniform in aspect] - which is the case in current code. There is a small snippet in the paste that could prevent this
Sat, Jan 28
I had a look at this just yesterday and have this working mostly already, only have to iron out a cornercase...
So assigning this to me [as I can finish it up this weekend] and hope you didnt already work on this.
If you already solved this though in the meantime feel free to comandeer back...
Jan 11 2017
first note: seems to only happen when 2d cursor is exactly on the bounds (or outside)
Will check logic in clipUVTransform() when time permits [could take some time though due to dayjob... so if somebody wants to jump in....]
@Vilko Lafliksyon (Eranekao): once this is committed, you'll see a message here [something like "Closed by commit ... "]
(you can then update&build immedialtely after or use builbot builds the day after or the official/stable release after this commit)
Does this answer your question?
Jan 10 2017
@Vilko Lafliksyon (Eranekao): this still waits for review... no garuantuee it will end up in master.
@reviewers: got a minute to check on this? thanx in advance...
Jan 7 2017
@Eonmach Realm (eonmachrealm): afaiks, armatures were already covered [couldnt add subdivision modifiers to them], but now lattices should also be fine... Feel free to report other bugs you encounter as well...
Jan 5 2017
Hm, no issues here (see T50376.blend), the modifier seems to do its thing based on the texture data.
At first glance This is the "SubdivisionSet" python operator not polling for the right object types, but it boils down to being able to call obj.modifiers.new("Subsurf", 'SUBSURF') without checking there for appropriate object types... will have a look...
mind having a look if this is a feasable fix? thanx in advance!
Jan 4 2017
this is what Mesh-Analysis (statvis) does: statvis_calc_distort() -- just to reference corresponding code for detecting non-flat faces that doesnt mark the faces from T50352: 3D-printing check misinterprates flat ngons as non-flat as non-flat...
proposed fix D2448: proposed fix for T50352
Can confirm this, will have a look...
First note: N-Panel > Mesh-Analysis doesnt color the faces in question as distorted [so C code does it 'right'], python operator seems to have the bug... (most likely unhandled flipped normals)
Nov 30 2016
regarding wording of tooltip: -1 for "surrounding" vs. "neighboring" (dont think this makes it more clear), suggest "...and try to reconnect detached nodes" instead
This commit only added menu entries for the given operators (not the operators themselves).
Oct 28 2016
quick note: every mode except "Percent" will run into clamping ("Clamp Overlap" option) real quick [but clamping will actually prevent the behaviour in the video]. "Percent" mode will actually work quite nicely though, no?
Sep 27 2016
Hi Jonathan & Julian, adding you as reviewers here because you had a look at D2234: T49347 part one and if we get a decission on this we can close T49347: Color for edges with bevel weight same as selected edges color...
Thanx for having a lookin advance!
Sep 25 2016
Sep 24 2016
k, have it working for the case of group instances in groups, will do some further thinking/checking and post patch later/tomorrow...
@Bastien Montagne (mont29) think this is your playground, but assigning to me cause I think I can get there and support this, will have a look (possibly this weekend). If it is real quick for you and you've got time, feel free to claim back...
Sep 23 2016
just very quick first look gives tons of these:
Just for others to see quickly (havent checked further)
btw: issues from report confirmed
@Antonio Vazquez (antoniov) thanx, np :)
@Matt (HyperSphere): thanx for the patch!
Sequencer is not really a part of blender that I am very familiar with, at first glance I can confirm the patch seems to solve issues mentioned in the report. Hope I can spend some more time checking more in depth...
author credits please ;)
only have very little time, but this seems to fix it for me... (also see comments in the paste)
only very limited time here, but:
it seems the meshdata is never really set to the object (thats why its empty -- which is not what create_floor() expects, thus giving empty lists to from_pydata() -- which is also bad...)
Sep 22 2016
@Brendon Murphy (meta-androcto): like I said, Carver MT is not buggy itself, it just shows a bug in the bmesh boolean modifier
@Matt (HyperSphere): didnt have a look at the issue yet, but you can post your diff here
Sep 21 2016
no commit access unfortunately
Hi merwin! almost forgot about this one. should be fine now?
Should I make a proper phabricator diff (instead of the paste)?
looks like an issue with blenders bmesh boolean modifier (script uses this internally) which Carver MT only exposes, the script itself is not really doing anything wrong...
this is more of a followup to followup to rB03073 (where this was done for the Outliner)
Changes to restrict flag will be lost on save/reload, so we should propably be strict and not allow these anywhere?
@YAFU (YAFU) : thanx for the files, can confirm issues in these, will have a look later
cannot reproduce this... color icon showing here in Properties Editor as well as Node Editor (for drag n drop)
tested this is current master and RC2 on linux
Totally agree about the duplicated standard colors that you tweeted about, as well as some others.
Anyways, could someone mark this as 'Accepted' then? Or just commit? Thanx in advance
Sep 20 2016
btw. while we are at it [and this is also UI-related]: what are your opinions on D2236: T49347 part two?
Sep 19 2016
Sep 17 2016
hi @Sergey Sharybin (sergey)! should we be strict and do something likethen [just to be consistent]?
Sep 16 2016
@Vilko Lafliksyon (Eranekao): dont speak too soon, this hasnt been commited yet ;) [those patches are in review, added you as a subscriber there so you can see progress]
I think this is a long-standing 'problem' (not sure how easy it would be to fix it once and for all -- or if there's another solution to this), but there is a workaround though of the top of my head:
allowed the toggle in D2237: possible fix for T49375
Hi @Erick Tukuniata (erickblender), I wasnt able to reproduce this (ortho and cursor depth seem to work fine here)
- where is your 3D cursor at?
- could you provide a .blend where this happens (3DCursor and view set exactly so the error happens)?
@Michael Otto (mayqel): yep if you dont have a normal connected it will use GPU_VIEW_NORMAL, if you've got somethink linked it multiplies GPU_VIEW_MATRIX with this
@Sergey Sharybin (sergey): cycles viewport seems fine to me, you mean BI/BGE use camera space normals whereas cycles uses world?
change came in rB002de
is this using the cycles Fresnel node? (not sure if this is even supposed to work with BI/GameEngine...)
made edge bevel weights also draw in vertex selection mode D2236: T49347 part two
Sep 15 2016
made this themable now D2234: T49347 part one
covered cases in scene: bake, render
covered cases in nodetree: CMP_NODE_OUTPUT_FILE and also the individual fileslots of that node
should be all cases now (at least I couldnt find more)
Sep 14 2016
@Vilko Lafliksyon (Eranekao): can have a look at that, too, guess it's fine to have just one report... will do tomorrow
note: bake uses this too (more others even?) So might have to consider a path like scene.render.bake.image_settings as well...
ImageFormatSettings were not setting their rna struct path, so a possible fix here D2228: possible fix for T49343
note: first occurance/autocomplete after a fresh restart of blender goes through without the error, only then (second/third) the error happens
Sep 13 2016
Sep 1 2016
Aug 28 2016
thanx for the notes! will come back to this the comming days...
Sorry, was having a quick look, but didnt have enough time to investigate this completely. Will come back to this [if noone else picks this up before I have the time]
Aug 27 2016
this apparently changed from 2.72 to 2.73 [looking into it now...]
cannot confirm here on linux (nvidia 970m with 367.35 drivers)
both current master and blender-2.78-testbuild2-linux-glibc219-x86_64 are not crashing here with specified file
note: your file uses MultiTexture materials, what you describe works in GLSL
not sure this is supposed to be supported in MultiTexture?
Aug 23 2016
Aug 11 2016
Jul 10 2016
@Campbell Barton (campbellbarton): think this would have a chance to go into 2.78? I've heard it being requested numerous times... Could you have a look? Thanx in advance!
Yep, looks like clouds procderural at scale 1 will give output values of .5 at corners [0,0], [0,1], [1,1], [1,0]
--> So no displacement there
file from report uses clouds at scale .25 which will give no displacement at all vertices even at subd level 2 or 3.