I have been playing videogames for over 20 years as one of my primary hobbies. During the last few years, I have watched other people play games on Youtube and Twitch.tv. Watching Youtube videos of people playing games has become my primary means of evaluating whether to get a new game or not. Trailer videos and reviews often paint an inaccurate picture of what the game will be like. Watching someone play the game is the most honest way to see what a game really is like playing. Some streamers also happen to be very entertaining to watch.
The experience
This weekend I experimented with streaming my own Diablo III gameplay on my Twitch TV Channel. This was my first time recording and broadcasting. It was an interesting experience.
Streaming a game is definitely an different way to play a game. Having the microphone on and your gameplay recorded feels a little bit like giving a presentation. For me, this meant that I felt very nervous, awkward and self-conscious at first. It made me lose my train of thought a couple of times, etc. After a while it got more comfortable, so the side-effects of being nervous went away a little.
Another thing to get used to is multi-tasking between the game and commentating on what you’re doing.
After streaming with zero watchers, I wanted to watch my recordings to see how I did. Twitch kept pausing the video to buffer every 30 seconds, despite my bandwidth being more than sufficient. I figured it was a limitation of Twitch. Luckily, Twitch makes it easy to export streams to my Youtube channel, which does not have issues with bandwidth.
Tech
For my setup, I started with Nvidia Shadowplay (which comes as part of the video card drivers) to record my gameplay, downscale the footage from my 1440p screen to a 720p video and stream it straight to Twitch TV. I recorded my voice-over with my gaming headset while playing the game.
Shadowplay was a very easy and low-barrier way to get started. The default settings are very conservative, so the result looked horrible to my very spoiled eyes. I had to increase the output resolution and bitrate to make it look a little bit better, but the Twitch stream output was nothing like watching Youtube videos in the same 720p resolution.
Looking for a better way to record and broadcast my video, I followed Twitch’s recommendation of Open Broadcaster Software (OBS). They have a good guide on how to configure it.
The visual quality went up, but I kept pushing too close against the maximum bitrate of 3500 kbit in order to make the stream look as good as possible. In fact, 3500 kbit is not enough to make the video look decent to me. Twitch starts getting buffering issues above 3000 kbit, but looked noticeably worse when I lowered my bitrate that far.
So: I was not happy with the quality of my stream on Twitch, yet my plan was to also get the footage on Youtube for later watching. Having zero subscribers to your (new) streaming channel means nobody sees it anyway. So I decided to skip Twitch and upload straight to Youtube. Uploading directly to Youtube allows me to record higher content at a (much) higher bitrate. Heck, I could upload a 1440p video in a crazy high bandwidth.
To determine what bitrate would be possible, what would be good and what would be a good baseline, I recorded a couple of short clips at different bitrates.
- 20mbit 1440p video looked fabulous, but my machine started making noticeably more noise by spinning up fans to cool itself. I’ve got quite a powerful setup, which I’ve tuned to make as little noise as needed even during reasonably high load. When it starts to make noise, I know I’m pushing against limits.
- 10mbit 1440p looked very good. It did not cause noisy fans.
- 8mbit 1440p still looked very good. In fact, I could not distinguish it from the 10mbit video. This is the bitrate I went with for recording my gameplay.
The Youtube video processing process is also interesting to look at. You upload the video, then they start processing it. Once this is done, the lowest quality version of the video (360p) is available. Invisibly in the background they keep processing increasingly higher quality versions (480p, 720p, 1080p and finally 1440p). Once the video list shows (HD) after the title’s name, the 720p version is available. This background process will take a couple of hours, depending on the length of the video.
I settled on uploading videos at night as unlisted, then making them public the next morning.
Gameplay
Oh, yes. I was recording playing a game.
I was playing Diablo III: Reaper of Souls, which has been out for over a year now. Last week Blizzard released the 2.1.2 patch in preparation for the Season being reset in early February.
At the launch of Season 1, I briefly tried it. At the time there were ways to exploit the game to get a maximum level character in two hours. For me this took the fun out of it. How can you compete on a ladder of sorts if you’re up against cheaters?
Blizzard hotfixed the ways to cheat out reasonably soon, but I had given up on it at that point. Now that a new season is coming up and it looks like people will actually be racing to reach the top, I’m interested in participating again.
So to see what my chances would be, I created a brand new hardcore character in Season 1. Hardcore characters stay dead once they get killed in game. You only live once. I like the added challenge of keeping a character alive.
The first night I played 6 hours and reached level 50 by playing the game in adventure mode. The second night I played about 3 hours to reach level 60. The third night I played another 3 or so hours to reach level 70 (the maximum) and reach my goal. Total time taken: 12.5 hours.
The key is to not push the difficulty level to extreme from the get-go, and to play in Adventure mode instead of the campaign. You have a more focused leveling experience and it’s nice not having to go through the low level parts of the campaign yet again. The gameplay is dumbed down quite a bit because this part serves as a tutorial of sorts. In adventure mode there is no tutorial, because you only unlock it after completing the campaign once.