[FFmpeg-devel] GSoC 2017

Umair Khan omerjerk at gmail.com
Tue Jan 31 11:24:18 EET 2017


On Mon, Jan 30, 2017 at 11:44 PM, Thilo Borgmann <thilo.borgmann at mail.de> wrote:
> Am 30.01.17 um 18:41 schrieb Umair Khan:
>> On Wed, Jan 25, 2017 at 5:45 PM, Thilo Borgmann <thilo.borgmann at mail.de> wrote:
>>> Am 25.01.17 um 06:14 schrieb Umair Khan:
>>>> On Wed, Jan 25, 2017 at 7:45 AM, Michael Niedermayer <michaelni at gmx.at> wrote:
>>>>>
>>>>> On Mon, Jan 23, 2017 at 03:39:09PM +0100, Michael Niedermayer wrote:
>>>>>> Hi all
>>>>>>
>>>>>> GSoC 2017 mentor org registration has opened a few days ago
>>>>>>
>>>>>> Our current ideas page lists just a single mentor and single sugestion
>>>>>> http://trac.ffmpeg.org/wiki/SponsoringPrograms/GSoC/2017
>>>>>>
>>>>>> About the Suggested project
>>>>>> It is a hard design task which few students would be
>>>>>> capable to do. I dont know if someone already knows a student who is
>>>>>> qualified for this and who would be available in FFmpeg GSoC but if not
>>>>>> I think finding a qualified student will be hard or require some luck
>>>>>> and success with a less than qualified student unlikely.
>>>>>> At least thats what my gut feeling says based on past years
>>>>>>
>>>>>> About the number of ideas
>>>>>> one is not enough, and i think its a waste of time if we apply with
>>>>>> such short list, i doubt we would be accepted with one idea and one
>>>>>> mentor
>>>>>>
>>>>>> About the number of mentors
>>>>>> one is not enough
>>>>>
>>>>> ok so our page now contains
>>>>> 4 project ideas and 3 mentors
>>>>> big thanks to the people volunteering and adding them
>>>>>
>>>>> Can we get a few more ?
>>>>> Ideal would be numbers similar to previous years
>>>>> We wont have a student for each idea we list most likely.
>>>>> If we list just 4 we might have just 1 or 2 students, having a
>>>>> wider range of ideas to select from would increase the number of
>>>>> contributions and number of potential new developers joining.
>>>>
>>>> One idea would be to finish rest of the ALS codec tasks.
>>>> What's remaining is :-
>>>
>>>> 1. Implement RLS LMS mode in the decoder.
>>>
>>> See 3.
>>>
>>>
>>>> 2. Get the encoder merged into the main repository.
>>>
>>> This was part of last years (yours) project... I admit that this has not yet happened is due to my lack of time for working on it with you (which will most likely change soon :) )
>>> However I would not like to have this iterated but being finished by the start of the next GSoC round (which basically means by yesterday...).
>>>
>>>
>>>> 3. Encoder featureset still lacks support for floating point sample
>>>> data encoding.
>>>
>>> 1. & 3. could be part of a "complete feature set for ALS"-project. Both are too little for separated projects and too much for a qualification task. But I like the idea of a comprehensive feature set project. You came up with this idea first so if you're up to mentor it, go ahead and add it to the wiki. Otherwise I'll do it during the weekend.
>>
>> The floating point decoding test fails on qemu arm. May be we can put
>> fixing this issue as the qualification task, and rest all as the main
>> project.
>
> Just added a project to the wiki.
> https://trac.ffmpeg.org/wiki/SponsoringPrograms/GSoC/2017#MPEG-4ALScodecimprovements
>
> I included fate coverage. Feel free to add/modify it.
> Would you prefer to be mentor?
> would you like to be backup mentor?

I'll add myself as the backup mentor.

- Umair


More information about the ffmpeg-devel mailing list