That's a reasonable suggestion, but how would you rationally do that? As an experienced player, it's easy to know when good and bad times to offclass are, but it's still essentially a judgement call based on the fine details of the given situation. How would you turn that into practical rules for offclassing that FaceIt could implement?

The only possible way forward I see with that would be a simple timelimit on how much each player can offclass, but there are some problems inherent with that. What if your team's stuck defending last half the game and offclasses are the better choice for defense? Would the timelimit simply kill you if you reached it and you were still playing the offclass? Your engineer or heavy suddenly having a heart attack would be a pretty shitty way to lose a push IMO. Also, a timelimit doesn't really prevent people offclassing at stupid times (i.e. Pyro to mid).

I believe that it would be much simpler (and more effective!) to just teach new players when appropriate times to offclass are. For example, if you offclass in a game with new players, when you switch classes you can just take 5 seconds to explain why so that they understand (i.e. "I'm going Sniper to pick the med since we wiped on mid and they have full Uber ad."). If you do this well, new players will start to restrict themselves from offclassing at bad times because they know when and why it's acceptable.

Over time, more players queuing up at a time should make games even more balanced than they were before, once elo gets sorted out properly. More players to choose from should mean that the matchmaking system works better, so I think the increase in player count is for the best. Also, as others have said, if you want a level of play that's a little higher, come play mixchamp or pugchamp.

Also, I think that people should remember that all these new players are the future of comp TF2. Every time you're toxic to someone who simply doesn't know how to play 6s properly yet, you're actually hurting the chances of survival for the community as a whole by limiting the chances that that player will continue to play and actually get good at the game.

We were all absolutely garbage at 6s at one point. I believe that the more you help new players through constructive criticism, the more likely they are to improve and stick with the game, sustaining the community as a whole.

Here, lemme get down off my soapbox.

#3 would be useful as long as the image quality is retained. I've always found startmovie recording far too tedious, so I'm happy there's someone with some ideas to make it faster.

More options tend to be better so you can cover people's personal preferences, but if you include options that are slower (e.g. PNG output), you should note how they will impact rendering speed/quality.

You could just record the same clip twice, one with viewmodels on and one with them off, and cut from the clip with viewmodels on to the one with them off at the moment that the view transitions from normal to "scoped in". (and simply do the same thing in reverse when they scope out again). That way you can create the illusion of continuity even though they are two separate clips.