v360 in_pad

classic Classic list List threaded Threaded
7 messages Options
Reply | Threaded
Open this post in threaded view
|

v360 in_pad

Michael Koch
Hi,

I just saw that the in_pad parameter of the v360 filter was changed to a
maximum of 0.1. That's not enough if a double-fisheye camera has lenses
with 220° field of view. Please change the maximum to 0.25 or bigger.

Thanks,
Michael

_______________________________________________
ffmpeg-user mailing list
[hidden email]
https://ffmpeg.org/mailman/listinfo/ffmpeg-user

To unsubscribe, visit link above, or email
[hidden email] with subject "unsubscribe".
Reply | Threaded
Open this post in threaded view
|

Re: v360 in_pad

Paul B Mahol
On 4/29/20, Michael Koch <[hidden email]> wrote:
> Hi,
>
> I just saw that the in_pad parameter of the v360 filter was changed to a
> maximum of 0.1. That's not enough if a double-fisheye camera has lenses
> with 220° field of view. Please change the maximum to 0.25 or bigger.

Nope. use fixed paddding instead.

>
> Thanks,
> Michael
>
> _______________________________________________
> ffmpeg-user mailing list
> [hidden email]
> https://ffmpeg.org/mailman/listinfo/ffmpeg-user
>
> To unsubscribe, visit link above, or email
> [hidden email] with subject "unsubscribe".
_______________________________________________
ffmpeg-user mailing list
[hidden email]
https://ffmpeg.org/mailman/listinfo/ffmpeg-user

To unsubscribe, visit link above, or email
[hidden email] with subject "unsubscribe".
Reply | Threaded
Open this post in threaded view
|

Re: v360 in_pad

Michael Koch
Am 29.04.2020 um 16:45 schrieb Paul B Mahol:
> On 4/29/20, Michael Koch <[hidden email]> wrote:
>> Hi,
>>
>> I just saw that the in_pad parameter of the v360 filter was changed to a
>> maximum of 0.1. That's not enough if a double-fisheye camera has lenses
>> with 220° field of view. Please change the maximum to 0.25 or bigger.
> Nope. use fixed paddding instead.

what do you mean by "fixed padding"? There are only two options for
"dfisheye", in_pad and out_pad.
If a double-fisheye camera has lenses with 205° field of view, then for
stitching together the two hemispheres the in_pad value must be (205° -
180°) / 205° = 0.122. When you restrict the in_pad range to [0...0.1],
you are breaking something that did work perfectly before.

Michael

_______________________________________________
ffmpeg-user mailing list
[hidden email]
https://ffmpeg.org/mailman/listinfo/ffmpeg-user

To unsubscribe, visit link above, or email
[hidden email] with subject "unsubscribe".
Reply | Threaded
Open this post in threaded view
|

Re: v360 in_pad

Paul B Mahol
On 4/29/20, Michael Koch <[hidden email]> wrote:

> Am 29.04.2020 um 16:45 schrieb Paul B Mahol:
>> On 4/29/20, Michael Koch <[hidden email]> wrote:
>>> Hi,
>>>
>>> I just saw that the in_pad parameter of the v360 filter was changed to a
>>> maximum of 0.1. That's not enough if a double-fisheye camera has lenses
>>> with 220° field of view. Please change the maximum to 0.25 or bigger.
>> Nope. use fixed paddding instead.
>
> what do you mean by "fixed padding"? There are only two options for
> "dfisheye", in_pad and out_pad.
> If a double-fisheye camera has lenses with 205° field of view, then for
> stitching together the two hemispheres the in_pad value must be (205° -
> 180°) / 205° = 0.122. When you restrict the in_pad range to [0...0.1],
> you are breaking something that did work perfectly before.
>

Now you are required to manually enter input/output FoV instead of
manually calculating padding.
_______________________________________________
ffmpeg-user mailing list
[hidden email]
https://ffmpeg.org/mailman/listinfo/ffmpeg-user

To unsubscribe, visit link above, or email
[hidden email] with subject "unsubscribe".
Reply | Threaded
Open this post in threaded view
|

Re: v360 in_pad

Michael Koch
Am 30.04.2020 um 20:23 schrieb Paul B Mahol:

> On 4/29/20, Michael Koch <[hidden email]> wrote:
>> Am 29.04.2020 um 16:45 schrieb Paul B Mahol:
>>> On 4/29/20, Michael Koch <[hidden email]> wrote:
>>>> Hi,
>>>>
>>>> I just saw that the in_pad parameter of the v360 filter was changed to a
>>>> maximum of 0.1. That's not enough if a double-fisheye camera has lenses
>>>> with 220° field of view. Please change the maximum to 0.25 or bigger.
>>> Nope. use fixed paddding instead.
>> what do you mean by "fixed padding"? There are only two options for
>> "dfisheye", in_pad and out_pad.
>> If a double-fisheye camera has lenses with 205° field of view, then for
>> stitching together the two hemispheres the in_pad value must be (205° -
>> 180°) / 205° = 0.122. When you restrict the in_pad range to [0...0.1],
>> you are breaking something that did work perfectly before.
>>
> Now you are required to manually enter input/output FoV instead of
> manually calculating padding.

Thank you, I saw it already. Will test it as soon as it's available on
Zeranoe.

By the way, it would be nice if also the circular field of view could be
specified. Because in many cases v360 is used with fisheye lenses.
Circlular field of view means h_fov and v_fov are set to the same value.
But not as d_fov, where a multiplication by 1.41 is required.

Michael
_______________________________________________
ffmpeg-user mailing list
[hidden email]
https://ffmpeg.org/mailman/listinfo/ffmpeg-user

To unsubscribe, visit link above, or email
[hidden email] with subject "unsubscribe".
Reply | Threaded
Open this post in threaded view
|

Re: v360 in_pad

Nicolas George
In reply to this post by Michael Koch
Michael Koch (12020-04-29):
> what do you mean by "fixed padding"? There are only two options for
> "dfisheye", in_pad and out_pad.
> If a double-fisheye camera has lenses with 205° field of view, then for
> stitching together the two hemispheres the in_pad value must be (205° -
> 180°) / 205° = 0.122. When you restrict the in_pad range to [0...0.1], you
> are breaking something that did work perfectly before.

If 0.122 did indeed produce satisfactory results for you, then there is
no reason to forbid it.

Please send a patch to the devel mailing-list, where it can be
discussed.

Regards,

--
  Nicolas George

_______________________________________________
ffmpeg-user mailing list
[hidden email]
https://ffmpeg.org/mailman/listinfo/ffmpeg-user

To unsubscribe, visit link above, or email
[hidden email] with subject "unsubscribe".

signature.asc (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: v360 in_pad

Michael Koch
Am 30.04.2020 um 21:09 schrieb Nicolas George:
> Michael Koch (12020-04-29):
>> what do you mean by "fixed padding"? There are only two options for
>> "dfisheye", in_pad and out_pad.
>> If a double-fisheye camera has lenses with 205° field of view, then for
>> stitching together the two hemispheres the in_pad value must be (205° -
>> 180°) / 205° = 0.122. When you restrict the in_pad range to [0...0.1], you
>> are breaking something that did work perfectly before.
> If 0.122 did indeed produce satisfactory results for you, then there is
> no reason to forbid it.

The in_pad option is no longer required because now dfisheye has the
h_fov and v_fov options, which is the better solution because of two
reasons:
1. The same parameters were already available for (single-)fisheye (and
are working fine there).
2. h_fov and v_fov allow higher precision because the values are
floating point. The in_pad option was also float, but I think it was
rounded to an integer when the padding was applied to the image.

Michael

_______________________________________________
ffmpeg-user mailing list
[hidden email]
https://ffmpeg.org/mailman/listinfo/ffmpeg-user

To unsubscribe, visit link above, or email
[hidden email] with subject "unsubscribe".