Technical Volunteers for Video and Livestreaming for IndiaFOSS

[Shree_Kumar]

I am looking for volunteers for an effort related to video for IndiaFOSS 2025. Last year’s conference had @Arya_Kiran and @mangesh doing livestreaming (documented here). This year, the conference will be recording devrooms as well - that’s 6 devrooms * 3 hrs = 18 hrs of content, roughly comparable to the main track itself.

Why this effort ?

First, we’re a “tech” community - so we should use our talent to solve challenges we face. We have to create working, reliable, well documented systems that we can use in the entire community (not just IndiaFOSS) and hopefully beyond. How creative can we get ?

Second, this relates to sustainability of the conference. Video and photography are significant costs. We should look to keep them to the minimum level possible, so that as our activities scale, those costs don’t baloon.

Last (but not the least), I feel by doing more of these things is how we actually become an active community. This is probably a useful part of our “education” efforts, where people with specialized knowledge can work on and share some of it in the process of doing useful things.

This is all part of things we can learn from other communities - e.g. FOSSDEM conference has been making their own “video box”. FOSSDEM runs ~70 devrooms ! The latest iteration of their solution is a custom hardware box. @fossdot got one and sent it to me. I will be testing it soon.

Here are some of the things you could be doing as a volunteer in this effort:

1. Video live-streaming
2. Evaluating video and various choices surrounding it
3. Audio, video mixing
4. Audio, video sync across multiple streams
5. Various video recording setups, equipment. Cost and quality tradeoffs. Which are good ones to rent…
6. Final editing to create high quality videos
7. Creating a robust system that is resilient to failures such as network connectivity, power failures, user errors or oversight.
8. Evaluating specific local hardware (low cost) for reliability
9. Interesting tech approaches to record content - e.g. a camera that follows speaker on the podium, or follows a mic in the audience.
10. Anything related to the web side of things - hosting/configuring software
11. Lots of testing
12. Documentation - very important. We’ll do things so that others can do more of the same.

We may or may not do everything above, and we may not do it all at once. The list is meant to be a sampling of possibilities. Most of the pieces already exist, so this is about us finding the mix that works well for local conditions. We may end up doing some software development too !

We need 2-4 volunteers. If you’re unsure, consider pitching in first. You will probably understand if you want to do this or not in the first kickoff meeting !

It helps to know some of these things, but you don’t have to know to be a volunteer. But, you’d have to be willing to spend at-least one full day a week towards this effort (plus 1 hr in a group meet), to do some technical activities. We have almost 2 months to go for IndiaFOSS, enough to get everything working well and tested !

To volunteer, drop in a message below. It would be good to know any prior experience with video/livestreaming, and which parts of this excite you the most. We’ll kick this off next weekend (26/27).

[Muneer_S]

Hey @Shree_Kumar, I would love to volunteer for this role. I’ve always found this sort of things very interesting. Never had the chance to work with these equipments, but im willing to spend time learning about these. Also ive done a lot of video recording and editing before(OBS Studio, Kdenlive, Capcut, and a million other video editing apps)

[mangesh]

Arya and I have published some form information about the configuration and how we did it. Since MumbaiFOSS 2024, mumbai folks have always been playing with different livestreaming setups. It’s a great learning experience and getting everything to work perfectly takes a lot of work and experience :p. Doing this is a very good practice and fun initiative, since the streaming setup will scale to perfection with time.

Arya’s detailed blog: Arya | Streaming our FOSS Meetups using just free software!
My talk in a Blr Meetup: https://mangeshm.xyz/talk/livestream.html (might not be very useful since it was a talk)
Repo: GitHub - fossunited/streaming

[hardik]

Hey @Shree_Kumar I have previously worked on video editing softwares like OBS, Nero, Zulu and although I have never did a production level streaming as above, I have a generic understanding and curiosity of how these things work. Hoping to contribute and learn as much I can with this opportunity.

[Mohd_Yasir_K_N]

Hi @Shree_Kumar, I would like to volunteer for this role. I have some experience managing media content for events, basic editing using davinci resolve and Shotcut. I’ve also used used OBS and vMix(used vMix to operate an LED wall at a previous event). I don’t have production experience with live streaming yet, I am open to learning and happy to contribute in any way I can.

[Shree_Kumar]

Thanks for volunteering, folks!

Heads up : I’ll setup an online meet for Sunday (July 27th) at 9 PM. I will be posting a link to the meet and tag you all in advance. (Will tag anyone who volunteers after this as well)

[Harsh_Patel]

Noted @Shree_Kumar , will be joining in as well, trying to help wherever I can.

[Shree_Kumar]

Here is the Jitsi online meeting link:

https://meet.jit.si/TechnicalVideoVolunteersJoinEnthusiastically

Meeting at 9 PM on Sunday, July 27th. See you there @Muneer_S @hardik @Mohd_Yasir_K_N @Harsh_Patel @Arya_Kiran

[Shree_Kumar]

Our first meeting ended just sometime ago. Attendees: @Muneer_S @hardik @Mohd_Yasir_K_N @Harsh_Patel .

Quick intros. We briefly discussed the IndiaFOSS and FOSSDEM approaches.
We had a long chat on overall what we need to do. Basically a robust system for video recording and capture for devrooms and main track.

For IF 2025 we need to livestream the main track. We also need to capture+livestream the devrooms. We want to create edited videos of the devrooms, post the conf.

We want to setup a robust system that can handle network, power failures, audio/video disconnect. We want to create something that can be easily operated by folks other than us, without much training. Ideally we’d want to offload most of the actual on-site work to other volunteers; we’d only need to occasionally oversee the task. We want to (ideally) source some hardware (e.g. PCs) from the community and use them. Other hardware may be rented as required (video camera, tripod, accessories, etc). These are the overall goals.

For the next one week - volunteers will be looking at basically their favourite parts of this, or everything in general. We’ll sync back at 9 PM every week on sunday. We’re co-ordinating discussions on a separate telegram group.

[Mohd_Yasir_K_N]

@Shree_Kumar

We’re co-ordinating discussions on a separate telegram group.

https://t.me/+t-ReYsK1eBxmZmVl

@hardik @Arya_Kiran

[Shree_Kumar]

I made some diagrams to clarify that the scope and partitioning of the work is clearer. The diagrams show vendor and our scope in different colors.

In IF2024, we livestreamed Audi 1 and Audi 2 (main track). In IF2025, we will do that. Plus, we will both record and livestream devrooms - 6 sessions which will happen in two different places: Audi 3 (4 devrooms, each 3 hrs), and a room in the first floor(2 devrooms, each 3 hrs). The setups differ slightly.

Main track in Audi 1 and Audi 2 will use 2 cameras - one for speaker, one for audience during QA. These also have audio mixing done by the vendor, they manage the stream going to the projector as well. From our POV, Audi 1 and Audi 2 are similar setups.

Devrooms will only have 1 camera, but there are differences in those setups. Devroom 1 (Audi 3) is larger, and thus needs dedicated audio. Vendor will do the audio mixing, and handle the loudspeaker as well. Video recording will be handled by a volunteer, who has to make sure that the camera follows the speaker.

Devroom 2 being a small room doesn’t need a loudspeaker, resulting in a much simplified audio workflow (may or may not be simpler for us though).

I will post more detailed notes on these soon, talking about what exactly the green boxes do, and what the selection criteria might be. The final setup we deploy may differ slightly compared to this, but will have almost all of these pieces.

[Souhrud_Reddy]

Hey there, I’ll my joining too!

I have used OBS, shotcut and Kdenlive casually for a little bit of Twitch streaming and basic video editing in the past

While I’m a noob with AV, Camera and tracking stuff, I would love to be a part of it and help out any way I can!

The speaker-following camera thing seems interesting

[Shree_Kumar]

I will describe the setups a bit more in detail here, with the simplest first.

Devroom 2 is in a small square room (35ish capacity). Speaker’s laptop is visible on a large screen TV. To capture this screen, we need to replicate the signal. That’s what the HDMI splitter does - it’s a commodity part. We’ll be dealing with 1080p resolution, so the absolute quality of this device will not matter so much.

The replicated signal goes to the large screen TV, and to a video capture device. The capture device converts from HDMI to USB 3.0 on the laptop. Similarly the video camera is also made available at the laptop. (Note : 2 USB ports gone).

Audio - will come in via the video camera - this becomes part of the USB 3.0 audio-video stream. We have QA at the end of the talk. When audience asks questions it doesn’t typically register in the video camera’s audio stream. There are two possible solutions : one that the speaker repeats the question before answering it… so that the question gets recorded. Other is to have an audience mic. Which then needs to be wireless, and this audio needs to be mixed with the speakers audio for recording.

The two video and audio streams are mixed into a video on the laptop. The laptop also sends the livestream to the cloud.

The video camera has local storage for the videos as well. This storage is limited and must be managed, using multiple storage cards, plus periodic transfer to the laptop.

An option here is to use a modified version of the FOSDEM video box. The Radxa X4 SBC inside the box may be powerful enough to mix the 2 video streams and stream it up.

A rental video camera must be chosen based on

1. Reasonable low light performance
2. Good autofocus with least volunteer assistance

Accessories would be multiple storage cards, power supply (ideally DC powered rather than battery). We’ll need a tripod also.

We also need a human volunteer to manage the setup. The tasks of the human volunteer are:

1. Ensure that the speaker is in the frame of the video camera… just in case the speaker moves a bit farther during the presentation
2. Ensure that all elements are working. Loose contact, pulled cables, loss of audio etc needs to be monitored continuously
3. Handle “scene changes”. In different sections of the talk, the livestream will show different elements. E.g. in the begining the title. During the presentation, the speaker will be rendered side by side against the talk slides in the livestream. During QA ideally only the speaker video, but audio mix between the mics. During the lunchbreak/transition something else, etc. This will be input by the volunteer manually. This scene selection must be easy to do and take effect immediately.

The choice of the video mixer running on the laptop. It should be simple to use, easy to change scenes intuitively. OBS is commonly used. FOSSDEM uses voctomix.

We also need to look at where the livestream video goes, how the users will access it, and how it will be referenced in the schedule. To be resilient to network disruptions will take some work. Good amount of testing will be required to ensure we have a rock solid system for the conf. If we can give the job of this volunteer to a student and ensure everything works with minimal supervision, then it’s a win.

Another fancy thing I’ve been thinking about is : what if the video camera can be automatically oriented towards the speaker using a motorized mount ? This involves video processing, maths and some control systems. Plus more testing. Good thing is that this activity is reasonably isolated from the rest. If we can get this working well, very useful - non-professionals aren’t typically good at continuous monitoring as they will likely get bored. Reliable automation can thus be very useful. But botched up automations can be painful. We’ll try this, but keep enough safeguards.

[Shree_Kumar]

We had a volunteer meeting yesterday (Aug 3, 9 PM). Attendees : @Mohd_Yasir_K_N , @Souhrud_Reddy and @Shree_Kumar .

@Mohd_Yasir_K_N updated us on what he’s been reading up related to DebConf livestreaming. Our setup might end up being similar. Looks like they stream 4K, and downscale to various resolutions.

@Souhrud_Reddy and me discussed about what we could do for the camera to automatically orient to follow the speaker. The minimal hardware test setup we could think of involved a mobile phone serving as a camera rotated manually, coupled to a laptop. We’re kind of thinking of this feature as at best something we will test independently at the conference (i.e. won’t put this into production). Souhrud has exams, etc, and this feature will need lot of testing. So best not to have unrealistic expectations.

I haven’t worked on the video box this week. Hopefully will find some time this week. The box I got is configured for production use. I’ll have to start with the software on the SBC inside the video box next.

Meanwhile, today @Suslime, me and other volunteers had visited the venue. Among other things, we met with the vendor who does the audio arrangements and the network provider.

Confirmed that the audio mixer will supply the mixed audio over a XLR connector/standard 3.5 mm jack. The vendor will deploy audio mixers in all three auditoriums, covering the main track and devroom 1.

Video streaming scheme is thus overall unchanged for the main track. Devroom will have the camera being operated by our volunteer. And devroom will entirely be owned by us.

Also confirmed that we’ll have wired internet access for all livestreams, meaning each laptop we use will get wired internet. (need to make sure every laptop has wired internet - many don’t so we may need an adapter or something)