[Bf-cycles] ray epsilon and transparent objects

[Mohamed Sakr]

the links for image and blend files in case the last message was too large.
http://www.pasteall.org/pic/88987
http://www.pasteall.org/blend/36383

On Wed, Jun 3, 2015 at 6:09 PM, Mohamed Sakr <3dsakr at gmail.com> wrote:

> here is a test scene
> and the image, with a marker over error parts.
>
>
> On Wed, Jun 3, 2015 at 6:05 PM, Sergey Sharybin <sergey.vfx at gmail.com>
> wrote:
>
>> I guess it is about storing "near clip" (aka ray->tnear from some other
>> render engines terminology). We indeed don't have because we're pushing ray
>> forward on surface bounce. See ray_offset() functions for example. It has
>> own upsides and downsides.
>>
>> There's a patch [1] where i've been experimenting with changing it to
>> ray->tnear approach in order to solve some numerical instabilities in the
>> watertight intersection code, still some work to be done there tho.
>>
>> In any case, as Antony already pointed out sharing .blend file which
>> demonstrates the issue makes investigation much simplier.
>>
>> [1] https://developer.blender.org/D1212
>>
>>
>>
>> On Wed, Jun 3, 2015 at 8:47 PM, Antony Riakiotakis <kalast at gmail.com>
>> wrote:
>>
>>> An image and blend file would go a long way in making things clearer.
>>> _______________________________________________
>>> Bf-cycles mailing list
>>> Bf-cycles at blender.org
>>> http://lists.blender.org/mailman/listinfo/bf-cycles
>>>
>>
>>
>>
>> --
>> With best regards, Sergey Sharybin
>>
>> _______________________________________________
>> Bf-cycles mailing list
>> Bf-cycles at blender.org
>> http://lists.blender.org/mailman/listinfo/bf-cycles
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.blender.org/pipermail/bf-cycles/attachments/20150603/c717a8ef/attachment.htm