Joined: 12/2/2011
Posts: 129
Location: Moscow, Russia
A TAS in a game category rather than a platform? o_O Does that mean I'll be able to get HL1 on here when the tools are done?
Anyway, cool run! OoB in Doom? D:
Joined: 12/2/2011
Posts: 129
Location: Moscow, Russia
In the Source games' speedrun community the runs done with host_timescale are called TSA-s (timescale-assisted) rather than TASes, especially now that advanced TAS tools are available for GoldSource and very basic scripting tools are available for Source which show how much the difference is between human-slowmotion and fully scripted runs.
Joined: 12/2/2011
Posts: 129
Location: Moscow, Russia
Agreed, the difference between blinking at 60 FPS and deblinked is very apparent and the former seems better to me after watching many TASes with blinking.
Joined: 12/2/2011
Posts: 129
Location: Moscow, Russia
I believe YouTube is VSynced by default (not to mention windowed applications are VSynced by Windows no matter what, and fullscreen flash players are usually windowed borderless), but for <something> + VSync I was referring to the video player's option to enable or disable VSync.
G-Sync is NVidia's resent technology that is supposed to refresh the monitor as soon as a frame is ready and not on a fixed timestamp, and it seems to work well in games, so perhaps it's just the video player not playing well with it?
Joined: 12/2/2011
Posts: 129
Location: Moscow, Russia
Additionally to YouTube, I thought it's a good opportunity to check how G-Sync behaves. So here is what I've got (Samus blinking test; MPC-HC x64 from K-Lite Codec Pack Mega):
Lossless footage:
120 Hz + VSync is almost perfect (there were a couple of "late" frames, but due to 120 Hz they are almost invisible);
120 Hz no VSync is a bit worse due to there being more "late" frames;
60 Hz + VSync is worse (the "late" frames are really visible and there were a bunch of them;
60 Hz no VSync is almost perfect, there were only two "late" frames, but this is still worse than 120 Hz + VSync due to the "late" frames being much more visible;
G-Sync actually turned out to look the worst of everything (including YouTube) for some reason, I didn't expect that.
So the flow goes like this: 120 Hz + VSync > 60 Hz no VSync > 120 Hz no VSync > 60 Hz VSync > G-Sync.
YouTube:
120 Hz - many "late" frames;
60 Hz - nearly perfect.
Once again, G-Sync for some reason turned out to be the worst of all although it should've given a perfect footage(?). And 120 Hz + VSync on the lossless footage turned out to be the best of all for me.
Joined: 12/2/2011
Posts: 129
Location: Moscow, Russia
By the way, we're working on tools that will allow for complete TASability of the game, including the RNG manipuation and consistency and one TAS input file for the whole run including potential saveloads and changelevels!
Joined: 12/2/2011
Posts: 129
Location: Moscow, Russia
Spikestuff wrote:
What was missing in your post when you made it, was mentioning that there are a few mistakes within the submission text. Instead it was just straight up pointing some of them out.
Joined: 12/2/2011
Posts: 129
Location: Moscow, Russia
Spikestuff wrote:
YaLTeR, have you ever had a look at what Got4n's Rayman 3 submission text was like?
No? Well shut it. Grammar Nazi.
Also, you're not doing your job of being a Grammar Nazi.
I have spotted more mistakes that what you have shown.
Edit: Fixed a little bit of the submission text. I know there are some errors that are still in there but I did a bit.
Well, I didn't read the whole text yet for once, and I'm not a native speaker either so I'm not sure about certain parts. And I don't have any way of knowing if you're gonna spot some mistake so I just pointed them out.
Joined: 12/2/2011
Posts: 129
Location: Moscow, Russia
Epic TAS. Yes vote, definitely.
Not all surfaces are solid. Yet, not all solid surfaces are visible.
Not all visible surfaces are solid. - is better IMO.
By gliding against an angled wall, you can avoid losing height during the glide, allowing to reach places usually to high or too far for Spyro.
too high
Turning while doing a stationnary jump often results in Spyro gaining more height from the jump. Although this technique works with all characters that can jump, it is unly used with Spyro and Bentley.
Joined: 12/2/2011
Posts: 129
Location: Moscow, Russia
TASing games with complex acceleration on Hourglass isn't really worth the effort until some kind of lua and whatnot is added for calculating the angles for perfect acceleration.
Joined: 12/2/2011
Posts: 129
Location: Moscow, Russia
Angel wrote:
уважаемые завсегдатаи, подскажите пожалуйста в какой топик форума мне обратиться с просьбой "переиграть" готовый TAS и перезалить его на ютуб в HD или выше?
просто медиаинфо оригинала
Width : 320 pixels
Height : 240 pixels
и смотреть на это без слёз имея монитор FullHD просто нельзя.
заранее благодарю откликнувшихся!
В настройках плеера метод расширения/сжатия/изменения размера/как у тебя там ставь Nearest Neighbor. Пример для Media Player Classic (для того, что из K-Lite Codec Pack'а, по крайней мере): View->Options->Playback->Output, там находишь Resizer справа снизу и ставишь Nearest Neighbor. Только потом обратно не забудь вернуть.
Randomness isn't under control yet and there's no good way of doing proper savestates as well for now.
A way of exporting scripts executable by unmodified game is to be done.
Joined: 12/2/2011
Posts: 129
Location: Moscow, Russia
I think that Source Engine TAS's can be made the same way I make HL1 ones at the moment, by writing a script and executing it every time. The problems still stay, though. I thought of solving them, but I want to bring HL1 to a complete TASable state before (savestates and randomness are the two main things left there).
Joined: 12/2/2011
Posts: 129
Location: Moscow, Russia
MESHUGGAH wrote:
An easy way to understand this is that the maximum speed Gordon can go is probably ~edit4: much more than2000units/speed, while the "running" speed is around ~400unit/speed (don't trust me, I played HL ages ago).
Maximum straight running velocity is 320 - default value of sv_maxspeed, and the player can go up to 2000 ups (units per second) - default value of sv_maxvelocity - in each direction (x / y), that makes 2828 ups the highest velocity if travelling diagonally at an angle of 45 degrees.
MESHUGGAH wrote:
A special trick that required going back to the previous map and re-entering the current one - hence the stop.
MESHUGGAH wrote:
around ~4:30, in theory, wallsliding gives a very slight boost while strafing there, but it's very hard to pull it off in real time and probably a TAS could make better strafes than falling on wall.
Perfect wallstrafing is a little bit slower than perfect air/groundstrafing.
MESHUGGAH wrote:
I know it's a TAS once someone able to pull objects while boosting off it infinitely, same way as in ~4:44.
That was done with a +use spam script.
MESHUGGAH wrote:
At ~7:48, a wall bug would be better (90 degree facing towrads a wall, forward and duck). Next scene why not jump on scientist and barney head?
This is a cutscene (until the tentacles die and the black brush covering the hole disappears), so speed is not very important in this part.
MESHUGGAH wrote:
edit3: "jumpbugs: 14"? I admin I didn't watched it fully, but was there actually a jumpbug (jumping down to a platform while ducking,release duck and jump when 2~4 units above the floor)? I saw mostly edgebugs (falling no a slope wall to avoid taking damage and saving horizontal speed) and "slow wallbugs" (jumping between 2 walls closing in 45 degree)
Joined: 12/2/2011
Posts: 129
Location: Moscow, Russia
Weatherton wrote:
Note: The 20:41 Half-Life Speedrun is, in fact "tool-assisted". It uses multiple frame-perfect scripts. Granted, large portions of the run are apparently played at real-time (for very short segments), but that just means they didn't have as many tools at their disposal as we've come to expect. You really can't compare this run to a run without scripts.
That's why there is a separate category in Half-Life speedrunning - Scriptless.