[Bf-committers] [GSoC] Improved threadability of Compositor nodes

[Matheus Sousa Faria]

Hi,

I've submitted my first draft to you through the GSoC website. I am still
working on it, but I submitted because I want to know if I am on the right
track. Please, let me know of any changes.

Thanks,
Matheus de Sousa Faria

On Sun, Mar 20, 2016 at 5:04 AM Jeroen Bakker <j.bakker at atmind.nl> wrote:

> Hello Matheus,
>
> On 03/20/2016 06:03 AM, Matheus Sousa Faria wrote:
> > Hey,
> >
> > Thanks for the answers! I've been studying the nodes and the techniques
> to
> > finish my proposal. And I have more questions:
> >
> >     1. What is the difference between directional and vector blur? I saw
> in
> >     the practice the difference, but I want to know conceptually, why to
> >     keep both nodes if they create the same effect (Motion Blur)?
> Well both were designed for the same use-case. Mimicking motion
> artifacts of typical video footage. Directional blur is a limitted, but
> faster, vector blur has more options, other limitations around the
> border of the image and is slower.
>
> Most of the time in compositing artists can chose which node supports
> their purpose the best. To my knowledge the vector blur node is used a
> lot, the directional blur not that much.
> >     2. I was doing some the tests with the glare node and this question
> came
> >     to my mind: What is the difference of fog glow and blur + mix? I
> generated
> >     the same scene using the glare node, and using the blur node with
> the mix
> >     to combine the original image with the blurred one. The result seems
> the
> >     same to me.
> Fog uses a convolve kernel with a Hartley Transform. The convolve kernel
> is more precise than the normal blurs. But I can imagine you can get
> visually the same results when using the blur node. It could be written
> as a Bokeh (which is a convolve kernel itself) and use the
> BokehBlurOperation. But not sure about the details as it also does some
> post color corrections...
>
> >     3. And about the technic used in the Ghost Glare, is there any
> reference
> >     to it? I mean a paper or something, the closest thing that I found
> about
> >     this effect are the crepuscular rays[1].
> Not sure if that exists. It was written back in 2006? Perhaps Ton knows
> if a specific paper was being used for the algorithm.
>
> >     4. Do you recommend any reference (papers, books, etc) about glares?
> The
> >     things that I found were about the common glare/glow effect[2].
> >
> >
> > [1] http://http.developer.nvidia.com/GPUGems3/gpugems3_ch13.html
> > [2] http://http.developer.nvidia.com/GPUGems/gpugems_ch21.html
> I do know of some gaming/animation papers that creates very good result.
>   - not realistic:
> http://john-chapman-graphics.blogspot.nl/2013/02/pseudo-lens-flare.html
>   - relastic but only for point based light sources:
> http://resources.mpi-inf.mpg.de/lensflareRendering/
>
> I most of the time use the latter one for references. It is a difficult
> algorithm, but has very good results. Might be a bit slow on the cpu as
> it generates meshes for rendering.
>
> Jeroen
> _______________________________________________
> Bf-committers mailing list
> Bf-committers at blender.org
> http://lists.blender.org/mailman/listinfo/bf-committers
>
-- 
Matheus de Sousa Faria
Software Engineering Student