[Bf-vfx] Feature request - 'overscan' rendering option for better distortion compositing

[Francesco Paglia]

I found a simple rules that helped me a lot to solve this issue.
The problem that comes up is if you just change the render size you will
still maintain the same area rendered while you need an "extra area" that
have to be distorted.
The quick fix is to multiply for the same value the render size and the the
camera sensor.
For instance, if you have a 35 mm sensor and a 1920x1080 render view and
the undistorted plate result the 10% bigger set the render size to
2112x1188 (1920*110% X 1080*110%) and the sensor at 39,5 mm (35*110%)

Looking at the footage through the camera you wont see anymore the 3d
matching your video but once rendered keeping the center  of the image of
the rendered frame at the same position as the footage is you will have a
perfect match.

hope this help!
Francesco

2016-02-25 15:06 GMT+01:00 Troy Sobotka <troy.sobotka at gmail.com>:

> Tacking it on here for future reference, OpenImageIO has granular control
> over the display and data window. The documentation and threads highlight
> some of the nuances that might not be obvious at first glance.
>
> With respect,
> TJS
>
> On Wed, Feb 24, 2016, 10:06 PM Sean Kennedy <mack_dadd2 at hotmail.com>
> wrote:
>
>> I'm not in front of my computer at the moment, but I have an idea about
>> this.
>>
>> I'm thinking the trick to getting the overscan is in the compositor. For
>> the sake of discussion, let's say the footage tracked and undistorted is
>> 1920x1080, and undistorted, its size is 2100x1160.
>>
>> You would set your render size to be 2100x1160, so the 3d elements are
>> the overscan size. In the compositor, you would also work at the 2100x1160
>> size, which you could do by scaling up the footage to that absolute pixel
>> size, then putting the original footage at 1920x1080 over it. This
>> essentially sets the canvas size to 2100x1160, but you're still working
>> with your original size footage. Then add your undistort node, alpha over
>> your oversized 3d render, then reapply the distortion. This will crop in
>> your 3d, like it normally does, but your working at the overscan size, so
>> you may be ok once you crop it down to 1920x1080. You could add a crop node
>> and an output node to save out the correct 1920x1080 file. The default
>> render will have to be the oversized one, but just don't use that one.
>>
>> Again, there could be something I'm overlooking in this workflow, but
>> that's what I would try.
>>
>> And of course this is just a workaround, not a real solution.
>>
>> Sean
>>
>> Sent from my iPad
>>
>> On Feb 24, 2016, at 6:11 PM, "Stephen Hamacek" <s.hamacek at gmail.com>
>> wrote:
>>
>> Hi all, first post here so forgive me if I'm in the wrong place :)
>>
>> An issue that came up recently was the compositing situation where a CG
>> element needs to be placed over a footage plate, where the plate has been
>> undistorted for matchmoving etc. but then the original ('distorted')
>> footage is desired for the composite.
>>
>> The CG element can be distorted effectively with the distort node, but if
>> it ran off the edge of the rendered frame, that edge would then impinge on
>> the composite, cutting off the edges of the CG element.
>>
>> I'm wondering if there is some way of creating an 'overscan' option
>> somehow, maybe in the render settings, that could be linked to the
>> distortion settings, so that the CG is rendered slightly larger, so that
>> when it is distorted this issue does not arise.
>>
>> Here is a thread where this was discussed somewhat:
>>
>>
>> http://blenderartists.org/forum/showthread.php?390200-The-composition-of-distortion
>>
>> Would love to know your thoughts on this. It seems to be a common enough
>> issue with compositing which other software (e.g. Syntheyes) knows about,
>> but maybe I'm missing something in my understanding. Thanks!
>>
>> Stephen Hamacek (Hammers)
>>
>> _______________________________________________
>> Bf-vfx mailing list
>> Bf-vfx at blender.org
>> http://lists.blender.org/mailman/listinfo/bf-vfx
>>
>> _______________________________________________
>> Bf-vfx mailing list
>> Bf-vfx at blender.org
>> http://lists.blender.org/mailman/listinfo/bf-vfx
>>
>
> _______________________________________________
> Bf-vfx mailing list
> Bf-vfx at blender.org
> http://lists.blender.org/mailman/listinfo/bf-vfx
>
>


-- 
Francesco Paglia
Vfx and Production Supervisor

mobile  +39 347.82.12.473
e-mail   f.paglia.80 at gmail.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.blender.org/pipermail/bf-vfx/attachments/20160315/02f478db/attachment.htm