- User Since
- Aug 13 2010, 4:07 PM (353 w, 5 d)
- "When a texture coordinate shader node refers to an object and the scene is made a full copy of, the material itself gets duplicated, but the nodes still refer to an object in a previous scene, not the duplicate."
Isn't this a bug?
Maybe got it -
- When a texture coordinate shader node refers to an object and the scene is made a full copy of, the material itself gets duplicated, but the nodes still refer to an object in a previous scene, not the duplicate.
Tue, May 23
Open the attached file and re-save it. The file has only one object present (as seen in the viewport and outliner), the rest were deleted. But a bunch of objects remain saved in the file as can be seen when trying to append or browsing data blocks.
Wed, Apr 26
Apr 14 2017
Indeed. On linux, the daily build works fine, the 2.78c crashes.
Still though - daily build keeps non-linked objects present in the file after saving.
Mar 14 2017
-Change file path for file output in node editor.
-Check use curves in scene properties.
-Press render again.
-Resulting output files are different.
Mar 13 2017
Just run into this. So basically - viewport preview doesn't work for semi transparent objects as what you see is not what you get. I understand that probably output can be changed into what is shown in viewport theoretically, but I wasn't able to do that in compositor. I can get a same image as if viewport would be rendered over some color, but not if keeping the alpha (adding un/premultiply seems to fix alpha issue, but brightens the colors, that doesn't mach the viewport again).
Mar 10 2017
Ideally a "find missing files" function, could have in additional to "Find All" option to replace the packed files too, so that only known not to exist files could be unpacked after. But that's definitely a feature request...
Nov 18 2016
Nov 11 2016
As I said, I understand that, that's why I asked if it is possible to let user know in the status bar, when the auto-save is being executed or at least when file writing fails.
Nov 9 2016
Oct 3 2016
Sorry, a proper one
Aug 1 2016
May 25 2016
May 24 2016
May 22 2016
So it is usable and the actual value is retrievable. Not the end of the world then. Thanks.
Though is there a reason to rounding it this way? I mean it's a direct input value, not a generated one, thus any weird number gets entered exactly as intended.
Strangely that works. However, try entering 0.012, 0.017, 0.01111 - it returns 0.01, 0.02 etc.
May 21 2016
May 18 2016
Feb 6 2016
Feb 4 2016
That's the point - objects visibility state icon is not the only criteria for it to actually be visible. Local viewport layers do that too, hence it's not enough to iterate through that to get if user actually is aware of the object. Didn't find a proper way to do that in an hour long googling.
Thanks, that clarifies the reason it's not showing. Had to deal with same problem on what objects python considers visible. However that's not exactly right behavior. Though I agree that it's hard to figure out a quick fix without crippling something else...
Maybe that's far fetched idea in terms of coding, but how about considering all viewports currently displayed inclusively, instead of scene, to define visibility state? That would fix both this issue and accidental changes to actually invisible objects via some scripts that depend on visibility list too.
Sorry, accidentally erased the link, here:
Feb 3 2016
Jul 27 2015
Maybe I mis-described something. With the present setup in the file (no need to create new constraints), when parented, object changes position, though constraint is not even position related.
Jul 26 2015
There is no set inverse button in the Copy rotation constraint.
Jul 25 2015
Jul 22 2015
Apr 29 2015
Sorry, it seems I deleted reference to file after uploading, here it is:
Apr 12 2015
Mar 23 2015
But is there a reason for separating camera scale factor? I mean the same image hardly can be used in both camera and orthogonal views, thus there shouldn't be collision of use. The default scale factor of 5 might be a bit misleading for camera though.
Jan 14 2015
I was thinking the same at first, but with limit of only 3 bounces almost all rays reach their limit. So then there either is some mythical condition when shader takes seriously more to compute single bounce or the ray doesn't really get terminated after the limit in some cases. I'm only speculating of course. Wish there'd be a way to see what actually happens :D
Jan 11 2015
@Zauber Paracelsus (zauberexonar) This saved me lots of headache inventing bizarre workaround shading trees :) Thanks.
Though the description seems totally different, and bounces are set to flat 3 for everything, indeed, suggested increase of roughness to 0.001 seems to decrease render time drastically. Might too high refraction precision be a clue to the problem?
Dec 18 2014
I see. So if I get it right the data basically is there but it has no v.goup definition. (I presume so because as soon as I manually create correctly named group for each object everything seems to work).
Could the solution be then to have a function that just copies v.group list from active object (or does it once while linking)? I guess it won't be a problem for me to python that, but it probably is a frustration for lots of people.
Sorry, accidentally pressed Submit.
Dec 1 2014
Sep 25 2014
I have the versions with mesh before and after this started to occur, but unfortunately the mesh is under strict NDA.
Almost exclusive operations performed were vertex merge -> at last/at center, but in this case I don't see particular changes between the two meshes. Is there a way to select the problematic edge so I can look for differences?
Sep 23 2014
Unfortunately can't comment on patch status, but at moment I'm using image viewer region -> change viewed frame, force-refresh and read it from that. Quite awkward and slow (though probably slowest part is transfer of pixels to python). Simply setting needed frame would be useful for sure.
Sep 5 2014
Jul 4 2014
Isn't that's why there is a link viewport with render layers button? To control whether viewport affects render layes.
Noo - it's other way around! In Cycles it works works great. It's a headache to deal with in internal :)
Viewport shouldn't override anything, especially if it affects final render.
Imagine rendering multipass with multiple scenes, where there are different layers in different scenes to be selected - hitting F12 becomes impossible then.
Jun 27 2014
May 24 2014
My mistake - there should be at least two keyframes of not the same value to see the effect.
May 12 2014
May 8 2014
It probably requires quite a bit of GPU memory.
But CPU freezes too for me.
May 3 2014
Apr 23 2014
But it doesn't seems like a distinct from parenting feature either :)
Apr 22 2014
Really would like to see this happen. I had to go through considerable efforts in connecting with external software on a few projects during last couple of years because of relatively simple fluid sims were required, yet grid fluid system, though good to have, proved to be limited in too many aspects for production use.
Apr 16 2014
As of yet - same here - not reproducible with latest build.
Perhaps it's windows only?
Cor me crashes are quite consistent on both win8 and vista.
A bit different messages, but crashes in both CPU and GPU modes.
Mar 12 2014
For those running into this issue: Scale up the scene. White lines will become thinner and almost disappear depending on scale.
Mar 10 2014
It might be placebo, but It seems in an RC (vc2013) it is harder to reproduce, still happens at about 1 in 30 mouse transitions on w8.1 and 1 in 50 on vista machine.
I guess this could be closed until I get some reproducible evidence.
Sorry, somehow forgot to mention it's about Cycles.
Mar 9 2014
Mar 8 2014
Feb 12 2014
Can't comment on matter if it is possible, but BlenderCAM is a really useful piece of software for people getting into CNC and other forms of "real world" 3D. Definitely deserves attention as a huge unused Blender potential.
Nov 25 2013
I'd be happy to test the patch if I'd know how to get a working version with it :)
You are right - non indented empty lines do not break Python flow. Still I'm sure most people would prefer to have those tabs in empty lines, not to have to put them manually when they decide to add code to that line. But that's a matter of preference rather than a bug, so I guess it's not for me to say :)
Empty newlines should be affected if they are inside selection, otherwise such tabbing will break python indentation flow.
Last line of selection on the other hand, if has no selected characters, usually is viewed as not selected (for good reasons).
Nov 21 2013
For people googling here accidentally:
image.reload() frees memory!
Maybe there is something a script could do explicitly to prevent crashing? Manually freeing all the cache?
I'm using image editor to read video frames via python for 3D scanning and it crashes after about 300 frames.