[anonymous]: Can you talk about your streaming setup

[v0lkan]

Can you talk about your streaming setup?

[v0lkan]

I’ve been asked this question a lot, so I think it’s time for me to elaborate on this a bit.

I stream on ZeroToHeroDev Twitch channel at least three times a few, often more.

And every time I figure something out, I modify my setup to provide a better viewing experience for people.

Aside

By the way, if you are curious about my overall developer setup, I have written it here. Here, I’ll only talk about the relevant parts of that setup to streaming.

Hardware

First things first, the hardware:

• Green screen: I use a retractable green screen that I hang on the wall, and I’m pretty satisfied with it so far.
• Microphone: I still use an love my AudioTechnica AT2020 USB Cardioid Microphone. I have plans to use a Shure SM7B to compare and contrast the differences, and I suspect Shure will provide a warmer voice because of the XLR and the amplifier that I’ll use too, but at the same time, it comes with a higher price point. And AT-2020 is wonderful, especially if you apply some real-time dynamic compression to the stream, which I’ll get to it soon.
• Key lights: Streaming is partially about looking good, and a good lighting setup helps a lot in that regard. I use Neewer dimmable lights. I have two of them pointing slightly above my eye level from the edges of the screen to my face. Using them not only shows my face clearer, but also helps me fix shadows, color bleed, and other greenscreen-related lighting problems.
• Computer: Apple iMac (Retina 5K, 27-inch, 2017), 3.5 GHz Quad-Core Intel Core i5, 32 GB 2400 MHz DDR4, Graphics: Radeon Pro 575 4 GB*)
  • I use it the display at its highest display resolution (3200⨯1800), I’ll get to the “why”s of it down below.
• Secondary monitor: ASUS VS238H-P 23’’, it’s good enough to monitor things, open an ebook, or a browser window on the side, or monitor the output stream for bit rate, latency, and other stream quality problems, and remediate them as soon as they happen—streaming is a very CPU-intensive process, so if, for example, a rouge backup program decides to go haywire in the middle of your stream, that can reduce your bitrate, cause a lot of frame dropping, and choppy streaming experience. If you don’t monitor your stream quality in realtime, you might not even realize if there’s a problem with it.
• I use my iPhone to both detect how the stream looks and plays on mobile devices, and also use it as a moderation tool to block spammers and scammers, which happen to hover around from time to time.
• Headphones: I have a pair of AudioTechnica ATH-M50X studio monitor headphones. They are great for… well, anything: doing post-production audio fine-tuning, listening to music, it’s a pretty decent set of headphones.
• Audio level control know: I use a VolBox inline audio control attenuator to manually adjust the background music and other audio that I hear on the headphones.

Audio Setup

Since I’m streaming on a Mac, it’s not as straightforward to stream the computer audio. I’ve tweeted about it, too, but the gist is:

• Install IShowU Audio Capture.
• Open Mac’s “Audio Midi Setup” and define a “Multi-Output Device” as shown here:

• Make sure your output is set to the multi-output device you just configured.

And you should be off to the races.

Capturing Desktop

Aside

OBS on Mac does not capture the mouse pointer when you do select the RDP session and capture it only. So to show my mouse pointer too, I had to do a full screen capture and scale up the capture to only show the 1080p portion (on the top left) and use a full Display Capture instead of a Window Capture so that it shows my mouse clicks too.

Also when you do a window capture, there is a lag between the video and your voice, so even without capturing your mouse clicks, it’s sill not the best experience.

As far as I know, this problem does not exist on the Windows version of the software.

I use OBS Studio to broadcast my desktop.

My screen is fairly large, I select a 1080p portion of it and only broadcast that part, which gives me some extra space to put things like the chat window, and the music player out of the view, just showing what I want to present.

I do my development on a Azure Windows VM; I connect to it through RDP and fit the RDP window into the 1080p section that I outlined above.

Let me show it with a picture to clarify:

The part you see is the portion of my screen that is captured and streamed by OBS, whereas the “light grey / dark grey” hatched part is the rest of the screen that is not visible. I had to experiment with sizes, and manually move the Display capture around, and scale it up, until the top left portion perfectly fitted the 1080p region that I wanted to stream.

Here’s my entire screen for comparison:

Stream Capture Settings

Here are various OBS Studio settings that I use.

Especially making the audio work with the capturing device took a while to figure out.

I’m putting screenshots here for reference.

Stream Settings

Output

Audio

Video

Other Advanced Settings

Also on my Mic/Aux capturing device I have defined a compressor and a noise filter which greatly improved the output sound quality, though they consume a fair amount of CPU too, so use them at your own risk.

Other Stream-Related Apps

• I use chatterino as a lightweight chat client to send messages, copy/paste links, etc to the chat audience.
• I use miro as a whiteboard, and also to keep track of what we’ve covered in the former sessions.
• I use remarkable for describing certain concepts with drawings and sketches on a paper-like environment.

And that’s all the extra tooling I use, I guess. I’ll add more if I remember any.

Why Are You Using a Virtual Machine to Stream Your Work?

For several reasons:

• Because I have a low-latency, high-bandwidth connection, so, I can :)
• Using a virtual work environment for my side projects helps me to get them done whereever I am: I can work on my desktop, exit my RDP session, take my laptop, go to the living room, re-open RDP, and my workspace will be exactly how I left it, all my connections still alive, all my editor windows and browser tabs open etc… I would not need to context switch, or reconfigure anything.
• Likewise, I can go to a friend’s house, borrow their laptop, and still get stuff done if I have my connection details. I’m not tied to a specific machine to get work done.
• That also means, if my mahcine is broken, my hard disk malfunctions, or it gets stolen or anything… I’ll give zero flops because my work is already safe in the cloud.
• Also, I don’t have to back up my work, because my hard drives in the cloud are reduntandly backed up in three separete regions and Azure promises 100% durability for my data — that means, nothing will happen to my data unless I explicitly remove it.
• There are many more that I don’t remember, but the moral of the story is… “this is the future” and, I’m and early-adopter who likes living in the future :)

But Why Not Linux? Why Windows for a VM?

For several technical reasons again:

RDP, as a protocol, works much better than VLC. I can share files between my guest and host OSs; I can use both of my monitors if I need; the video quality is much better, I can stream audio of the connected machine too, and the frame rate that I get off of it, especially when the machine is remote, is way better.

So overall, RDP is a better protocol, so it more-or-less constraints me to use a windows anyway.

The only alternative protocol that can reliably compete with RDP is PCoIP, and the only “as a service” solution that gives that reliably is Amazon Workspaces.

Now, Amazon Workspaces is a little more expensive than my Azure setup; well it’s not “that” big of a difference for me to not try it out. And when I use an Amazon Linux Desktop, thru PCoIP, using Amazon workspaces I can say that the overall experience is comparable to RDP and even better at certain areas.

But, aside from that, I was bitten by Amazon’s proprietary DNS implementation and a [GoLang patch] for Go 1.11 and above that expects the DNS provider to follow industry standards, and hence not honoring Amazon’s custom DNS rules.

But long story short, I was not able to connect to my MongoDB Atlas cluster unless I edited my /etc/resolv.conf. And when I edited my /etc/resolv.conf I opened a different can of worms (as in, if I get disconnected, I’d need to reboot the machine to reconnect; or when I need to execute commands as root, my virtual machine was not recognizing my root user, etc.)

So if I was going to use GoLang and MongoDB Atlas together and debug things locally, then an Amazon Linux Workspace was not an option.

I also tried Amazon Windows Workspaces, but I would not recommend them as a remote workspace, as it had its own set of issues, the biggest of which is the Desktop that Amazon advertises as a “Windows 10” machine is modified windows server; it is not a Windows 10 box, so it’s really hard to install WSL on it. And without WSL, windows is a “no go” for me anyways.

Which leaves the only option as an Azure remote VM for the time being.