Sreenshots vs Youtube pimps

[LoveFest]

Why are we still doing screenshots? A video of your work is far more persuasive. We put so much time and sweat in our maps and mods, I think a video is way more compelling.

[GreatEmerald]

To be honest, creating good quality videos is a lot harder than screenshots. A 1080p screenshot is one command, upload and a paste away, while in order to create a video you need to use capture software, have a whole lot of disk space (preferably a whole hard drive just for the videos), quite a bit of time editing and more for rendering.

But yea, for big mods it makes sense to create videos. I'm planning to do a video of all Unreal II-related mods in one go and another video showing the progress on Arcomage.

[ChaosWolf]

I can't even record videos on mine! The game crashes when I try to play the demos back.

[meowcat]

@LoveFest: How do you edit videos (what software etc.)? I've used the demo version of FRAPs (I think) to record gameplay, but trying to upload the resulting <gigantic> .AVI file to YouTube takes forever.

[arbee]

Youtube videos are far too teasy when I search for them, like bragging about unreleased stuff.

[GreatEmerald]

Having been doing videos on YouTube for the past 2 years, I have to say that the best option is to use a program that allows you to choose the encoder, like ZDSoft Game Recorder. The thing is, Windows is horrible for this. There are no normal free programs for capturing 3D accelerated graphics. Fraps is a fine choice, it uses a fast and near-lossless compression method to capture video, but the downside of it is that it's zero configuration and it's not free. And no configuration is really, really bad if you have complex setup like I have (two soundcards, one records voice and the other records sound!)... There are others, though, like GameClaw, WeGame etc. But I haven't followed their development that much. Should try them out once again to see if there are any improvements.

After you have the resulting lightly compressed files (use yuv if you can choose the compression method while capturing), you need to process them to a usable size. You can do this in editing software such as Microsoft Expression Encoder (which once again doesn't give you many options) or Camtasia Studio. I'm not sure if you can use standalone FFmpeg, but if you can, then it will be hands down the best thing to use. If you have options again, this is what YouTube accepts:
WebM format videos. They are optimal right now, since they can be used directly without much processing and are really well compressed. If there is no WebM preset, the codecs are VP8 for video and libvorbis (OGG Vorbis) for audio. When choosing a bitrate, the higher the better - it doesn't affect the size that much at all.
MPEG-4 ASP. Also known as XviD. This format is a lot more cumbersome since it's not optimised for small files, but you can still get a fair size/quality ratio. The codecs are MPEG-4 vX (X = a number) for video and AAC or MP3 for audio.
Even though they claimed that H.264 support was there, I've had bad experience with it - it takes long to process and the resulting videos have extremely degraded quality. However, you can still use H.264 when capturing if you have a powerful PC and low disk space.

Now if you were to record on Linux, things are a lot more bright there. You can use tools like FFmpeg directly to record and process all your files. It can capture the output of X Window System (x11grab), but unfortunately doesn't have accelerated OpenGL support. If you need that, then GLC is your friend. It works great when capturing accelerated graphics, but unfortunately it doesn't work on 64-bit Wine. You can still capture UT2004 just fine, though, since it's a native application.

Again, use ffvhuff to capture things if you can afford the disk space or one of the other fast format if you can't. After that, compress to WebM or MPEG-4. For WebM, FFmpeg already knows the codecs to use when you use the .webm extension, you just need to set your quality. For MPEG-4, you need to do something like this:

ffmpeg -i MyHuffFile.avi -vcodec mpeg4 -qscale 8 -qcomp 0.8 -threads 4 -acodec libfaac -y MyMPEG4File.mp4

Naturally, each of those programs on Linux are controlled from the command line. While it might sound uncomfortable, the console allows you to do things that you normally can't do with a GUI. For instance, this is how I solve the problem of having this strange sound setup:

/usr/bin/ffmpeg -f x11grab -xerror -r 25 -s 800x600 -i :0.0+4,24 -vcodec mpeg4 -qscale 8 -qcomp 0.8 -threads 4 -an -y Video.mp4 | sox -t alsa -v 0.3 hw:0,0 -t alsa -v 0.45 hw:1,0 -m Sound.mp3

This makes it so FFmpeg records video and SoX records the sound, starting the exact same moment so I wouldn't get a desync. You can also pipe SoX input directly, but it makes FFmpeg check for desync which is terribly slow, so I just use them separately.

Of course, you can also find GUIs for those, but don't expect to be able to use pipes with them (they're still good for reference, like choosing the window to grab). Although you could make a GUI that allows that by executing everything through the command line, just that people rarely think about such things.

[Piglet]

The best way I've seen it done is to put the output from one PC into a video-capture card on a second. You get fully synched sound and graphics at full screen quality in real time.