IMX frame interval monitor (FIM) available on ipu{1,2}_ic_prp{enc,vf} capture nodes?

IMX frame interval monitor (FIM) available on ipu{1,2}_ic_prp{enc,vf} capture nodes?

[Tim Harvey]

Steve,

Some time ago you mentioned that adding FIM capability
(imx-media-fim.c) on the ipu{1,2}_ic_prp{enc,vf} capture nodes was on
your plate to fix as currently the controls are only available on the
ipu{1,2}_csi{0,1} capture nodes. Do you have any patches for that or
can you explain what needs to be done?

I have a user that wishes to use FIM to deal with a lossy analog
signal coming out of sync but they also wish to use the VDIC to
de-interlace.

Best regards,

Tim

Re: IMX frame interval monitor (FIM) available on ipu{1,2}_ic_prp{enc,vf} capture nodes?

[Steve Longerbeam]

Hi Tim,

I gave this a shot a while ago, but I recall struggling with some 
locking problems. I haven't been able to return to it. I've updated the 
imx/add-fim-to-prpencvf branch to latest media-tree master branch 
(5.11.0-rc1), in my github fork [1]. The patches against master are:

0001-media-imx.rst-update-doc-to-reflect-FIM-changes.patch
0002-media-imx-prpencvf-attach-FIM.patch
0003-WIP.patch

The last patch was my first attempts to debug the locking problems.

I don't have access to imx6 hardware any longer, so I can't progress any 
further on it at this point.

Steve

[1] git@github.com:slongerbeam/mediatree.git


On 12/31/20 8:48 AM, Tim Harvey wrote:
> Steve,
>
> Some time ago you mentioned that adding FIM capability
> (imx-media-fim.c) on the ipu{1,2}_ic_prp{enc,vf} capture nodes was on
> your plate to fix as currently the controls are only available on the
> ipu{1,2}_csi{0,1} capture nodes. Do you have any patches for that or
> can you explain what needs to be done?
>
> I have a user that wishes to use FIM to deal with a lossy analog
> signal coming out of sync but they also wish to use the VDIC to
> de-interlace.
>
> Best regards,
>
> Tim

Re: IMX frame interval monitor (FIM) available on ipu{1,2}_ic_prp{enc,vf} capture nodes?

[Tim Harvey]

On Sat, Jan 2, 2021 at 9:55 AM Steve Longerbeam <slongerbeam@gmail.com> wrote:
>
> Hi Tim,
>
> I gave this a shot a while ago, but I recall struggling with some
> locking problems. I haven't been able to return to it. I've updated the
> imx/add-fim-to-prpencvf branch to latest media-tree master branch
> (5.11.0-rc1), in my github fork [1]. The patches against master are:
>
> 0001-media-imx.rst-update-doc-to-reflect-FIM-changes.patch
> 0002-media-imx-prpencvf-attach-FIM.patch
> 0003-WIP.patch
>
> The last patch was my first attempts to debug the locking problems.
>
> I don't have access to imx6 hardware any longer, so I can't progress any
> further on it at this point.
>
> Steve
>
> [1] git@github.com:slongerbeam/mediatree.git

Steve,

Thanks for the patches!

I have found via experimentation that if you are using the VDIC to
de-interlace and you loose/regain the signal to the adv7180 it will
roll endlesslessly. However if you are doing raw capture the video
will sync up within a few seconds (240 lines * 16.6ms-per-frame =
3.8secs max I suspect). Thus the simplest solution for users who
suffer from this may just be to capture raw and upscale.

Best regards,

Tim

Re: IMX frame interval monitor (FIM) available on ipu{1,2}_ic_prp{enc,vf} capture nodes?

[Steve Longerbeam]

Hi Tim,

On 1/5/21 3:57 PM, Tim Harvey wrote:
> On Sat, Jan 2, 2021 at 9:55 AM Steve Longerbeam <slongerbeam@gmail.com> wrote:
>> Hi Tim,
>>
>> I gave this a shot a while ago, but I recall struggling with some
>> locking problems. I haven't been able to return to it. I've updated the
>> imx/add-fim-to-prpencvf branch to latest media-tree master branch
>> (5.11.0-rc1), in my github fork [1]. The patches against master are:
>>
>> 0001-media-imx.rst-update-doc-to-reflect-FIM-changes.patch
>> 0002-media-imx-prpencvf-attach-FIM.patch
>> 0003-WIP.patch
>>
>> The last patch was my first attempts to debug the locking problems.
>>
>> I don't have access to imx6 hardware any longer, so I can't progress any
>> further on it at this point.
>>
>> Steve
>>
>> [1] git@github.com:slongerbeam/mediatree.git
> Steve,
>
> Thanks for the patches!
>
> I have found via experimentation that if you are using the VDIC to
> de-interlace and you loose/regain the signal to the adv7180 it will
> roll endlesslessly. However if you are doing raw capture the video
> will sync up within a few seconds (240 lines * 16.6ms-per-frame =
> 3.8secs max I suspect). Thus the simplest solution for users who
> suffer from this may just be to capture raw and upscale.

Yes, but for our users, a 3.8 second delay is unacceptable, for example 
in a rear-view-camera application, it's not acceptable to see rolling 
video for that long. Hence the FIM which will restore sync very quickly 
on loss/regain of the analog signal to the adv718x.

Steve

> Best regards,
>
> Tim