[mythtv-users] New deinterlacer for perfect image quality when using an interlaced display, mode that matches the source

[Paul Gardiner]

Tom Dexter wrote:
> On Mon, Apr 6, 2009 at 12:47 AM, Paul Gardiner <lists at glidos.net> wrote:
>> Jean-Yves Avenard wrote:
>>> Hi
>>>
>>> 2009/4/6 Paul Gardiner <lists at glidos.net>:
>>>> Thinking about this a bit more, an easy slip that would cause this
>>>> would be to hardwire the "field" flag - present on trunk, but not
>>>> on fixes - to a fixed value. Instead "dirty" can be used in place
>>>> of the field flag.
>>> Does Mark version work for you ?
>>> Or should I revert to your previous patch ?
>> It works for me, but only after two build fixes. There's the
>> ConstFilterInfo => FilterInfo one, but also FieldorderDeint
>> needs changing to accoung for filters taking only two arguments
>> on fixes, rather than the three on trunk. Here's the version
>> I'm using:
>>
>> +static int FieldorderDeint (VideoFilter * f, VideoFrame * frame)
>> +{
>> +    ThisFilter *filter = (ThisFilter *) f;
>> +    TF_VARS;
>> +
>> +    AllocFilter(filter, frame->width, frame->height);
>> +
>> +    int dirty = 1;
>> +    if (filter->last_framenr != frame->frameNumber)
>> +    {
>> +        if (filter->last_framenr != (frame->frameNumber - 1))
>> +        {
>> +            memset(filter->got_frames, 0, sizeof(filter->got_frames));
>> +        }
>> +        store_ref(filter, frame->buf,  frame->offsets,
>> +                  frame->pitches, frame->width, frame->height);
>> +        dirty = 0;
>> +    }
>> +
>> +    filter_func(
>> +        filter, frame->buf, frame->offsets, frame->pitches,
>> +        frame->width, frame->height, dirty, frame->top_field_first,
>> +        dirty);
>> +
>> +    filter->last_framenr = frame->frameNumber;
>> +
>> +    return 0;
>> +}
>>
>> Note "dirty" appearing twice in the arguments to filter_func.
>>
>> Cheers,
>>        Paul.
>>
> 
> 
> Wow...I'm confused here.  Yesterday I was running for a while using
> the mythtv-0.21-field-order.6.patch, which seems to have everything
> you're referring to, and I swear the motion just didn't look right at
> all.  I went back to the original patch.
> 
> I can't believe I was just seeing things.  Maybe I'll try again.

I'm sure you were seeing a real problem. But be something quite subtle.
Not sure how we can track this down.

P.