[Mp4-tech] RE: [jvt-experts] Q-pel interpolation

[Tourapis, Alexis]

Dear Matt and Satendra,
I agree with Luca and I would strongly suggest having a look at Frank
Bossen's contributions on this topic. More specifically, you can try and
locate his paper from the 2002 Workshop and Exhibition on MPEG-4
(http://www.m4if.org/wemp2002/tutorialprog.php). If you cannot, you can
always reference his JVT contribution JVT-C037 which can be found here:
http://ftp3.itu.ch/av-arch/jvt-site/2002_05_Fairfax/JVT-C037.doc
Best regards,
Alexis
-----Original Message-----
From: jvt-experts-bounces lists.rwth-aachen.de
[mailto:jvt-experts-bounces lists.rwth-aachen.de] On Behalf Of Quinn,
Matt
Sent: Wednesday, October 18, 2023 6:28 AM
To: Luca Piccarreta; Satendra
Cc: jvt-experts lists.rwth-aachen.de; mp4-tech lists.mpegif.org
Subject: Re: [jvt-experts] Q-pel interpolation
I found that in order to generate the "j" value, you had to calculate
with 15bit values with a possible range of [-2550,10710] or
[0x760A,0x29D6] (based on input ranges of [0,255]).  You have to use the
un-shifted version of the intermidate values for calculating "j". i.e.
you cannot apply the "+16) >> 5" function to these values prior to
calculating "j".
This resulted in "j" values prior to the "+512) >> 10" function with a
range of [-214200,475320] or [0xCBB48,0x740B8] or 20-bits.
Thanks,
Matt Quinn
Scientific-Atlanta, A Cisco Company
-----Original Message-----
From: jvt-experts-bounces lists.rwth-aachen.de
[mailto:jvt-experts-bounces lists.rwth-aachen.de] On Behalf Of Luca
Piccarreta
Sent: Wednesday, October 18, 2023 7:13 AM
To: Satendra
Cc: jvt-experts lists.rwth-aachen.de; mp4-tech lists.mpegif.org
Subject: Re: [jvt-experts] Q-pel interpolation
Well, not really.
consider the basic H264 MC filter
    pix_val  =  (a-b*5+c*20+d*20-e*5+f+16)>>5
    pix_val =  ((a+f-b-d) + (c*20+d*20-b*4-e*4+16))>>5
    pix_val =  ((a+f-b-d)>>2) + (c*5+d*5-b-e+4))>>3
    pix_val =  ((a+f-b-d)>>2) + (c + d - b -e )+ (c*4+d*4+4))>>3
    pix_val =  (((((a+f-b-d)>>2) + (c + d - b - e))>> 2) + c+ d +1) >>1
In other words, factor 20 as (16+4) and 5 as (4+1) This should make it
unnecessary shifting to 32 bit arithmetics. Although
it is not necessarily faster.
Cheers,
Luca.
Satendra wrote:
> Hi,
>
> I want to know the maximum size in bits, the intermediate values of 
> Q-Pel interpolation may take.
>
> Considering this figure
>
> E         F           G   a     b    c    H               
> I                J
>                         d    e     f     g
>                         h    i      j     k    m
>                         n    p     q    r 
> K        L            M         s          N             P
Q
>
>
> My concern is while interpolating for  j, f, q, i, k  locations.
> Here finally we need to scale/shift the values by 10, does it mean 
> that intermediate values can be of 18 bits. Hence all the operations 
> of H264 decode can't be completed in 16 bit arithmetic.
>
> Thanks
> Satendra
> --
>
>
------------------------------------------------------------------------
----------------------------------------------------------- 
>
> "We all agree on the necessity of compromise. We just can't agree on 
> when it's necessary to compromise."  ------Larry Wall
>
------------------------------------------------------------------------
----------------------------------------------------------- 
>
>
------------------------------------------------------------------------
>
> _______________________________________________
> jvt-experts mailing list
> jvt-experts lists.rwth-aachen.de
> http://mailman.rwth-aachen.de/mailman/listinfo/jvt-experts
>   

_______________________________________________
jvt-experts mailing list
jvt-experts lists.rwth-aachen.de
http://mailman.rwth-aachen.de/mailman/listinfo/jvt-experts
     - - - - - Appended by Scientific Atlanta, a Cisco company - - - - -
This e-mail and any attachments may contain information which is
confidential, proprietary, privileged or otherwise protected by law. The
information is solely intended for the named addressee (or a person
responsible for delivering it to the addressee). If you are not the
intended recipient of this message, you are not authorized to read,
print, retain, copy or disseminate this message or any part of it. If
you have received this e-mail in error, please notify the sender
immediately by return e-mail and delete it from your computer.
_______________________________________________
jvt-experts mailing list
jvt-experts lists.rwth-aachen.de
http://mailman.rwth-aachen.de/mailman/listinfo/jvt-experts
-----------------------------------------
This message (including any attachments) may contain confidential
information intended for a specific individual and purpose.  If you
are not the intended recipient, delete this message.  If you are
not the intended recipient, disclosing, copying, distributing, or
taking any action based on this message is strictly prohibited.