On 3/9/18, Hazem Ashmawy <hazem.s.ashm...@gmail.com> wrote:
> On 3/9/18, Paul B Mahol <one...@gmail.com> wrote:
>> On 3/9/18, Hazem Ashmawy <hazem.s.ashm...@gmail.com> wrote:
>>> Add filter to convert between various panorama projections. It supports
>>> Equirectangular and Cubemaps (3x2 and 6x1 cubemap layouts).
>>>
>>> commit a8d80408bd9d99542cc29f30d7e6b00771846029
>>> Author: Hazem Ashmawy <hazem.s.ashm...@gmail.com>
>>> Date: Thu Mar 8 10:09:36 2018 +0200
>>>
>>> avfilter: add convertion to/from cubemap 6x1
>>>
>>> Signed-off-by: Hazem Ashmawy <hazem.s.ashm...@gmail.com>
>>>
>>> commit f9f6a6cc0ceb1a4e749041658b7a441696b82588
>>> Author: Paul B Mahol <one...@gmail.com>
>>> Date: Thu Dec 3 21:15:13 2015 +0100
>>>
>>> avfilter: add panorama filter
>>>
>>> Signed-off-by: Paul B Mahol <one...@gmail.com>
>>>
>>> Signed-off-by: Hazem Ashmawy <hazem.s.ashm...@gmail.com>
>>> ---
>>> libavfilter/Makefile | 1 +
>>> libavfilter/allfilters.c | 1 +
>>> libavfilter/vf_panorama.c | 553
>>> ++++++++++++++++++++++++++++++++++++++++++++++
>>> 3 files changed, 555 insertions(+)
>>> create mode 100644 libavfilter/vf_panorama.c
>>>
>>
>> Here is partial change for converting 3x2 cubemap as given by latest
>> youtube videos to
>> equirectangular projection:
>
> Should I modify cube faces order to match the one specified here too?
> https://github.com/google/spatial-media/blob/master/docs/spherical-video-v2-rfc.md#semantics-3
Yes, The 3x2 cubemap projection should be what google use.
There is also facebook own cubemap layout IIRC, but that one can came later.
>
>>
>> | diff --git a/libavfilter/vf_panorama.c b/libavfilter/vf_panorama.c
>> | index 7d9de5b62c..4d43fc0dd5 100644
>> | --- a/libavfilter/vf_panorama.c
>> | +++ b/libavfilter/vf_panorama.c
>> | @@ -394,9 +394,9 @@ static int config_output(AVFilterLink *outlink)
>> |
>> | phi_threshold = atan2(1., 1. / cos(theta_norm));
>> | if (phi > phi_threshold) {
>> | - face = DOWN;
>> | - } else if (phi < -phi_threshold) {
>> | face = TOP;
>> | + } else if (phi < -phi_threshold) {
>> | + face = DOWN;
>> | } else {
>> | ;
>> | }
>> | @@ -411,14 +411,14 @@ static int config_output(AVFilterLink *outlink)
>> | case RIGHT:
>> | locate(z, y, x, M_PI_2, rw, rh, &ox, &oy);
>> | break;
>> | - case TOP:
>> | - locate(y, z, x, M_PI, rw, rh, &ox, &oy);
>> | + case DOWN:
>> | + locate(y, z, x, M_PI_2, rw, rh, &ox, &oy);
>> | break;
>> | case BACK:
>> | - locate(x, y, z,-M_PI_2, rw, rh, &ox, &oy);
>> | + locate(x, y, z, 0, rw, rh, &ox, &oy);
>> | break;
>> | - case DOWN:
>> | - locate(y, x, z,-M_PI_2, rw, rh, &ox, &oy);
>> | + case TOP:
>> | + locate(y, x, z, M_PI , rw, rh, &ox, &oy);
>> | break;
>> | }
>> |
>>
>> Could you incorporate this?
>> And also update all other conversions so thay match with this one?
> Sure, I'll work on this.
Good, you will also need to add frame side data support... but that can wait.
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
