[Bf-committers] [Bf-blender-cvs] SVN commit: /data/svn/bf-blender [27124] trunk/blender/source/blender/ blenkernel/intern/fcurve.c: getting double frames problem, set the epsilon to 100th of a frame rather then 100, 000th.

[Mike Belanger]

I can't imagine using a strip scaled 10,000 x.   
I'll add it to the wiki as a warning.  But not everyone reads the wiki before diving in.  

I wonder why this can't just be restricted?  I.e. >
There's a maximum scale of say, 100x, or whatever's within practical ranges.

 
Mike Belanger ( Mikahl )
www.watchmike.ca





On 2010-02-25, at 3:36 AM, Nathan Vegdahl wrote:

> If people end up having real use-cases for scaling NLA strips so
> extremely, then we can always revisit this and try to fix it in a more
> sophisticated way.  But I have a hard time imagining real use-cases
> for that, so I suspect it will be a non-issue.
> 
> --Nathan
> 
> On Wed, Feb 24, 2010 at 9:25 PM, Joshua Leung <aligorith at gmail.com> wrote:
>> Hi,
>> 
>> I thought I'd just mention here a possible issue that arises (or arised in
>> the past) with a coarse epsilon value like this here, in case anyone has to
>> track this down again at some point.
>> 
>> Basically, past a certain point, if NLA strips get scaled, and then the user
>> tries to add more keyframes, epsilon values like this simply won't suffice.
>> However, I strongly DO NOT RECOMMEND to let your NLA strips get into such a
>> state (i.e. you scale you NLA strips, then intend to go back to the action
>> and add more keyframes to the end (as opposed to simply tweaking the
>> existing ones)), since this is not the best way to do things. This comes
>> from seeing many files come and go in the bug tracker that were scaled to
>> ridiculous amounts (scale factor around 10000, all keyframes stored in the
>> space of 1 frame or less in non-nla-scaled-time), though part of that was
>> also the result of the buggy design of the old NLA system.
>> 
>> In short, you (as in the animators) have been warned ;)
>> </end rant>
>> 
>> Regards,
>> Aligorith
>> 
>> On Thu, Feb 25, 2010 at 6:14 AM, Campbell Barton <ideasman42 at gmail.com>wrote:
>> 
>>> Revision: 27124
>>> 
>>> http://projects.blender.org/plugins/scmsvn/viewcvs.php?view=rev&root=bf-blender&revision=27124
>>> Author:   campbellbarton
>>> Date:     2010-02-24 18:14:16 +0100 (Wed, 24 Feb 2010)
>>> 
>>> Log Message:
>>> -----------
>>> getting double frames problem, set the epsilon to 100th of a frame rather
>>> then 100,000th.
>>> 
>>> Modified Paths:
>>> --------------
>>>    trunk/blender/source/blender/blenkernel/intern/fcurve.c
>>> 
>>> Modified: trunk/blender/source/blender/blenkernel/intern/fcurve.c
>>> ===================================================================
>>> --- trunk/blender/source/blender/blenkernel/intern/fcurve.c     2010-02-24
>>> 15:56:27 UTC (rev 27123)
>>> +++ trunk/blender/source/blender/blenkernel/intern/fcurve.c     2010-02-24
>>> 17:14:16 UTC (rev 27124)
>>> @@ -326,7 +326,7 @@
>>>  }
>>> 
>>>  /* threshold for binary-searching keyframes - threshold here should be
>>> good enough for now, but should become userpref */
>>> -#define BEZT_BINARYSEARCH_THRESH       0.00001f
>>> +#define BEZT_BINARYSEARCH_THRESH       0.01f /* was 0.00001, but giving
>>> errors */
>>> 
>>>  /* Binary search algorithm for finding where to insert BezTriple. (for use
>>> by insert_bezt_fcurve)
>>>  * Returns the index to insert at (data already at that index will be
>>> offset if replace is 0)
>>> 
>>> 
>>> _______________________________________________
>>> Bf-blender-cvs mailing list
>>> Bf-blender-cvs at blender.org
>>> http://lists.blender.org/mailman/listinfo/bf-blender-cvs
>>> 
>> _______________________________________________
>> Bf-committers mailing list
>> Bf-committers at blender.org
>> http://lists.blender.org/mailman/listinfo/bf-committers
>> 
> _______________________________________________
> Bf-committers mailing list
> Bf-committers at blender.org
> http://lists.blender.org/mailman/listinfo/bf-committers