-
Notifications
You must be signed in to change notification settings - Fork 0
Post-process videos #128
Comments
Here's the link to view the tracking spreadsheet: https://docs.google.com/spreadsheets/d/1ovifZ1tLCy3m6fJGmabhOkpZ385ICk5EdrKDHcOD0Yw/edit?usp=sharing I've finished all the templating stuff, will look at my schedule later today and fill out the due dates accordingly. Also all the text in the intro (sans logo) uses the cool build-in typewriter effect that the conference title uses. Neat! |
Thanks for sharing your tracking sheet.
I'm not sure what this means, I thought you and @garrying had locked down the splash/intro stuff, could you post if you are changing? |
The splash / intro is nearly identical, the talk title and name just build in the same way as the conference title does instead of fading in. The templating bit is me making our designs work in Premiere so I can just copy paste text into a box and not have to re-render everything every time. |
@dcwalk For Dr. Jennifer Wemigwans' talk would you like the territorial acknowledgement left in the video or would you like me to start the talk when she begins speaking? |
If you think it flows to leave the land acknowledgement/intro in I think that would be preferrable. Your plan sounds good, maybe you can propose your approach to her by email before publishing as unlisted and sharing? I'm not sure about the degree of concern/sensitivity here |
It works fine with it in, just wanted to make sure you were on-board with being included in the video since it's a little more than the usual introduction or ending bits. It flows well! I think Jennifer just wanted to review the video before we released it to make sure she didn't do a bad job speaking. Unlisted videos can only be seen to those who have the link and we just won't share the link with anyone else (same thing as all my intro previews which can't be seen on my channel). If she likes it we will schedule the release to become public like all the other videos and if she doesn't it gets deleted off the channel. I'll make sure this is noted in my email to her so it's clear that the YouTube link we're sending isn't publicly accessible. That aside, all talks should be finished today! I can start uploading them and setting up the descriptions and tags on YouTube as well if you'd like, they'll all be private until a release schedule is decided upon. |
Yes I understand that, my comment about waiting before uploading it is trying to be sensitive to the perception of checking in before moving forward. |
Would you like me to share the video with Jennifer using another method? I can probably give her a Dropbox or Google Drive link if you think that's better than YouTube for this. |
This could be a question you ask her. I don't have a preference. |
Editing is finished for all talks without blockers! Jennifer's talk is finished editing and I've emailed her (and CCed the org email), the one talk with a blocker is Zach's (Zack's?) lightning talk because I don't definitively know how to spell his first name and unfortunately he didn't have his conference lanyard on during the talk. Once I get some sort of answer there I can plop in the titles and render. Other than that, thumbnails need to be generated and descriptions need to be put into text files and then we're done! MUCH nicer process compared to last year and overall I'm quite happy :D |
I think @Shrinks99 has valid points about scheduled releases on YouTube. One the other hand, we are almost ready to make a big announcement to have everything ready on every platform, including our 600 Twitter + 300 mail subscribers, compared to our < 100 on YouTube. That said, scheduled releases could let YouTube better promote our content to attract more subscribers. My preference is to release everything on IPFS, Internet Archive, and our partner ISOC, but schedule YouTube releases. In the newsletter announcement, tell people everything is on those platforms, and YouTube releases are scheduled over next couple weeks, and those releases have in their description link backs to our non-Google archives. This way we are fully utilizing YouTube's network effect, while giving a bias to the platforms we prefer. |
Next steps:
*By Nov 15, we need YouTube links of all talks available on website as in 2018, but it seems we cannot keep them as unlisted and then schedule for them to become publicly listed. If that's the case we'd have to release entire archive together on Nov 15. If there is a good way around this then we can schedule to release on YouTube one per day. **@garry and I will also sort out how to link the livestream archive m3u8 such that bookmark links from before are preserved, and @darkdrgn2k and I probably need to move chunks to a more permanent place than our server |
Huh, so this works a little differently than I thought it would. If you set the video to Alternatively, the YouTube links could just be put up on the website in one batch after they are all public if we don't want to manually make one video public every day which is a pretty dumb way of doing it but that would also work. As for the spreadsheet, I can give you all access to it and you can mark down in the progress column what the status is? |
I am definitely not going to manually unlist one video per day :D and we have set a hard deadline to get the 2019 site archived under the 2019.ournetworks.ca so it seems not worth it to do scheduled listing. We may as well publish the entire archive on Nov 15 so we have all the links for website. Yes please give me access to spreadsheet! |
Videos have been uploaded to the Alexandria server with the same naming conventions as the 2018 videos. Descriptions will be uploaded (taken from the website or created by me for the lightning talks) soon! |
Hank on chat:
Thanks @Shrinks99, I will follow the below checklist to review the archive, and add notes to a new column I created called Reviewer Notes so I don't mix up your production notes. Here is the spreadsheet link. For each video:
For the archive:
When all these verifications are completed, we'll be publishing the entire archive to:
Updates:
|
@benhylau -- we should work from the data file for sessions for all descriptions, presenter names, etc..., as it builds the recorded talks page. Can the txt files be updated to mirror what is here: https://github.com/ournetworks/ournetworks.ca/blob/master/_data/sessions.yml? As you are the one doing that work I'm not sure what I'm supposed to verify. I've reviewed the production spreadsheet prior and confirmed that no talks were included that shouldn't be. It doesn't seem like I need to review again |
The two mostly match, but not 100% as they are intended to be pasted into YouTube descriptions, etc. with project links and context information. For example... Our_Networks_2019_Channel_Channel_by_Biyi_Wen.txt contains
which I added. Our_Networks_2019_DHT_BFT_OMG_Know_Your_Building_Blocks_For_Decentralized_Applications_2019_Edition_by_Kirill_Pimenov.txt contains
which is in the original GitHub submission. Other submissions have links that I extracted to put into the description, some giving important context such as in Our_Networks_2019_Building_A_Modern_Memex_by_Andrew_Louis.txt. Our_Networks_2019_The_Impotent_Medium_by_Farhad_Bahram.txt has more information about the video, whereas in sessions.yml it just says
Then there are items like lightning talks, where it isn't a 1:1 mapping, as talks are broken up into individual video files, whereas on sessions.yml it is a single record. @dcwalk is it important that they match 100%? I'd prefer descriptions in the archive to provide best context of the video on their own. Matching sessions.yml means you'd like the Keynote text to be included in the filenames and video title page right? cc. @Shrinks99 this probably means we need to re-render both keynotes and Channel Channel?
Regarding 6 and 7, I have checked that but want a second person to verify. If you feel the production spreadsheet is correct then we're fine, as I have verified that it matches the archive. |
We didn't add the full descriptions to the youtube videos last year, only
the shortened ones that I believe matched the data file. E.g.:
https://www.youtube.com/watch?v=gDtO0qRAuTA (I'd suggest we follow that
practice this year too. That would shorten many of the descriptions I
already see)
Matching sessions.yml means you'd like the *Keynote* text to be included
in the filenames and video title page right?
No those don't have to be included, that is a workaround for the program.
Can you PR back in the changes that you see are improvements that you
mention in yours. I'm not saying the data file is perfect, Farhad's for
example was based on the late addition / lack of info we received, but I
think the descriptions should be in sync.
…On Tue, Nov 12, 2019 at 11:44 AM Benedict Lau ***@***.***> wrote:
The two mostly match, but not 100% as they are intended to be pasted into
YouTube descriptions, etc. with project links and context information. For
example...
*Our_Networks_2019_Channel_Channel_by_Biyi_Wen.txt* contains
Video presentation by Biyi Wen, accompanying the exhibited artwork:
Channel channel, a network for VPN access.
which I added.
*Our_Networks_2019_DHT_BFT_OMG_Know_Your_Building_Blocks_For_Decentralized_Applications_2019_Edition_by_Kirill_Pimenov.txt*
contains
It is going to be a revised and extended variant of my talk on Web3 Summit
in Berlin in 2018: https://www.youtube.com/watch?v=J2lnjFqi-No
which is in the original GitHub submission. Other submissions have links
that I extracted to put into the description, some giving important context
such as in *Our_Networks_2019_Building_A_Modern_Memex_by_Andrew_Louis.txt*
.
*Our_Networks_2019_The_Impotent_Medium_by_Farhad_Bahram.txt* has more
information about the video, whereas in sessions.yml it just says
Video presentation.
Then there are items like lightning talks, where it isn't a 1:1 mapping,
as talks are broken up into individual video files, whereas on sessions.yml
it is a single record.
@dcwalk <https://github.com/dcwalk> is it important that they match 100%?
I'd prefer descriptions in the archive to provide best context of the video
on their own.
Matching sessions.yml means you'd like the *Keynote* text to be included
in the filenames and video title page right? cc. @Shrinks99
<https://github.com/Shrinks99> this probably means we need to re-render
both keynotes and Channel Channel?
title: "Keynote: Dr. Jennifer Wemigwans, A Digital Bundle Protecting and
Promoting Indigenous Knowledge Online"
title: "Keynote panel: May First, technology and revolution"
Regarding 6 and 7, I have checked that but want a second person to verify.
If you feel the production spreadsheet is correct then we're fine, as I
have verified that it matches the archive.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#128?email_source=notifications&email_token=ABDZWYQDHAO7JZTBYO3ENELQTLMPZA5CNFSM4I2Q2UZKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOED24TDY#issuecomment-552978831>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABDZWYRZ5BY34ATXJMYQJ7DQTLMPZANCNFSM4I2Q2UZA>
.
|
Then there are items like lightning talks, where it isn't a 1:1 mapping,
as talks are broken up into individual video files, whereas on sessions.yml
it is a single record.
We'll have to add these to the datafile to have the talks added to the
website
…On Tue, Nov 12, 2019 at 2:51 PM Dawn Walker ***@***.***> wrote:
We didn't add the full descriptions to the youtube videos last year, only
the shortened ones that I believe matched the data file. E.g.:
https://www.youtube.com/watch?v=gDtO0qRAuTA (I'd suggest we follow that
practice this year too. That would shorten many of the descriptions I
already see)
> Matching sessions.yml means you'd like the *Keynote* text to be
included in the filenames and video title page right?
No those don't have to be included, that is a workaround for the program.
Can you PR back in the changes that you see are improvements that you
mention in yours. I'm not saying the data file is perfect, Farhad's for
example was based on the late addition / lack of info we received, but I
think the descriptions should be in sync.
On Tue, Nov 12, 2019 at 11:44 AM Benedict Lau ***@***.***>
wrote:
> The two mostly match, but not 100% as they are intended to be pasted into
> YouTube descriptions, etc. with project links and context information. For
> example...
>
> *Our_Networks_2019_Channel_Channel_by_Biyi_Wen.txt* contains
>
> Video presentation by Biyi Wen, accompanying the exhibited artwork:
> Channel channel, a network for VPN access.
>
> which I added.
>
>
> *Our_Networks_2019_DHT_BFT_OMG_Know_Your_Building_Blocks_For_Decentralized_Applications_2019_Edition_by_Kirill_Pimenov.txt*
> contains
>
> It is going to be a revised and extended variant of my talk on Web3
> Summit in Berlin in 2018: https://www.youtube.com/watch?v=J2lnjFqi-No
>
> which is in the original GitHub submission. Other submissions have links
> that I extracted to put into the description, some giving important context
> such as in
> *Our_Networks_2019_Building_A_Modern_Memex_by_Andrew_Louis.txt*.
>
> *Our_Networks_2019_The_Impotent_Medium_by_Farhad_Bahram.txt* has more
> information about the video, whereas in sessions.yml it just says
>
> Video presentation.
>
> Then there are items like lightning talks, where it isn't a 1:1 mapping,
> as talks are broken up into individual video files, whereas on sessions.yml
> it is a single record.
>
> @dcwalk <https://github.com/dcwalk> is it important that they match
> 100%? I'd prefer descriptions in the archive to provide best context of the
> video on their own.
>
> Matching sessions.yml means you'd like the *Keynote* text to be included
> in the filenames and video title page right? cc. @Shrinks99
> <https://github.com/Shrinks99> this probably means we need to re-render
> both keynotes and Channel Channel?
>
> title: "Keynote: Dr. Jennifer Wemigwans, A Digital Bundle Protecting and
> Promoting Indigenous Knowledge Online"
> title: "Keynote panel: May First, technology and revolution"
>
> Regarding 6 and 7, I have checked that but want a second person to
> verify. If you feel the production spreadsheet is correct then we're fine,
> as I have verified that it matches the archive.
>
> —
> You are receiving this because you were mentioned.
> Reply to this email directly, view it on GitHub
> <#128?email_source=notifications&email_token=ABDZWYQDHAO7JZTBYO3ENELQTLMPZA5CNFSM4I2Q2UZKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOED24TDY#issuecomment-552978831>,
> or unsubscribe
> <https://github.com/notifications/unsubscribe-auth/ABDZWYRZ5BY34ATXJMYQJ7DQTLMPZANCNFSM4I2Q2UZA>
> .
>
|
I changed the keynote title name because it didn't follow the format used in the other videos. I would propose that we follow the format used for everything else to keep things consistent. For the panel I felt that it's important to signify that it's a panel in the title and I suppose the filename should be updated to match. This is sort of subjective and up for debate though so I'll defer to what you want on both cases. I would highly suggest that we don't put "Dr. Jennifer Wemigwans" in the video title however or else we'll be left with two instances of it in the title and that seems foolish. In any case, it seems that we are in agreement :P |
I have reviewed and approved all videos after @Shrinks99 re-published the incomplete files. I think we're good to go and have updated #128 (comment) to reflect that 🎉 @Shrinks99 can you start the upload to YouTube, using descriptions from here. I will wait for a final decision re: Jennifer's talk then publish to IPFS pin on Alexandria and Infura.io, and publish to IA. |
Fixed mismatched filenames and confirmed they now match across directories. Useful commands for the verification:
Changed file owner and permissions to be same as 2018 on alexandria. The archive to be published has the following sha1sum:
@Shrinks99 can you confirm these are the same on your local? This is excluding Jennifer's talk in all folders. |
I will be able to confirm these changes and start uploading on November 23rd. |
All of the talks have been uploaded to YouTube! They are currently all up as unlisted on our channel. YouTube has changed their system and it seems that we can now schedule videos to publish while they remain unlisted, this would allow us to release one video per day on YouTube while keeping the video URLs which can be accessed through the website or the conference playlist. I'll get on thumbnails ASAP and hopefully we can push this thing out the door soon! |
IPFS pinning recorded videosAfter @Shrinks99 used sftp to transfer all files to Alexandria, and @benhylau did file name and content verifications, we started using Alexandria as IPFS origin, with the following procedures. alexandria.tomesh.netUsed following command in
File hashes:
ipfs.ioIn a
At one point had to
But eventually
Gateway: https://ipfs.io/ipfs/QmaWc83FvLkyvw8p7Q45prW7HTXAaLgk8jxmkdciokaE1K/ ipfs.infura.ioUse Infura API pin_add:
The call will timeout after a while, but the pinning seems to be happening in the background. Make subsequent calls later to check on progress:
Gateway: https://ipfs.infura.io/ipfs/QmaWc83FvLkyvw8p7Q45prW7HTXAaLgk8jxmkdciokaE1K/ Infura pinning never completed, progress seems to get stuck around 3xxx and returns inconsistent results on repeated API calls :( |
Internet Archive uploadArchive description:
|
IPFS pinning livestream archiveI assume live.mesh.world already has all the livestream chunks on the m3u8 files in ournetworks/ournetworks.ca/pull/154/ pinned to its IPFS node. So I generated a list of the CID hashes using:
Attached here: ipfs-cids.txt ipfs.infura.ioThere are 1449 CIDs. Then in a
The log file gets mostly
So I assume the After one iteration over the list, 142 of the 1449 CIDs have been pinned onto Infura:
After 3 iterations we have 542 of the 1449 CIDs pinned. Another couple iterations and having the CIDs pinned on ipfs.io seems to help and finally all 1449 CIDs are pinned 🎉 ipfs.ioConnect live.mesh.world (origin of CIDs) to IPFS Cluster for faster resolutions:
In a
Check on pin status using:
You should find the CID is either pinned or queued for pinning:
All 1449 CIDs are pinned when I checked back after a couple hours 🎉 |
This was done and the videos are added! Closing this :) |
A tracking issue for the work of processing the videos. Pulling on previous notes from August:
0.5 hour * 22 talks + 15.5 hour * 1.5 padding < 40 hour
So, we are talking ~40 hours, with the expectation that @dcwalk will take care of publishing planning (separate issue)
Timeline
Begin: September 30
End:
Todos
The text was updated successfully, but these errors were encountered: