Doom 2 in under 10 minutes [TAS]

[baja blast rd.]

On 4/19/2022 at 2:14 PM, Dimon12321 said:

Will we see the first time in the history when Cacowards mentions a TAS demo? As far as I remember, even Zero-Master's Nightmare run wasn't mentioned there

 

The Cacowards will be getting rid of a lot of the smaller side features, to encourage people in the community to take the initiative and do thorough, cool essays on stuff they like instead of just passively simping for shout-outs. 

 

(I'm kidding but act like I'm not -- because if this is one of the best TASes of all time, and the actual speedrunning community doesn't give it a high-quality breakdown, that would be a shame.) 

[RonnieJamesDiner]

My ass was puckering so hard watching you lower into that pit of Shotgunners with 13% hp at the 3:35 mark. But then, the whole 1% hp situation going on at the 7:01 mark basically made me stop breathing for a second, haha, and again watching you nearly killing yourself for that rocket jump on The Living End (so close to the finish line!!). This is so much fun to watch, and unbelievably impressive.

 

Adding on to what rd mentioned above, a breakdown of this run would be really nice to have. As impressive as it is (that can't be overstated), there's a whole slew of things going on here that I simply don't understand, and it would be fun to learn!  

[Redneckerz]

Some zombiemen in the video are asking what the hell just ran past by.

 

But goodness, this is 9.42 minutes of awesome. Some sections have so many turns that eyes can't catch up even at 60 fps.

 

A lot of speedrunning trickery involved, and some i can't immediately recognize (and yet it still happens ofcourse). I echo that this deserves a breakdown.

 

Quality mapsets are rightfully reverred for their skill, but this TAS is of equally exceptional quality. When 19 hours is poured into making it, i wholeheartely believe that, and it shows.

 

Bravo @almostmatt1 for delivering a TAS Magnum Opus. Exceptional craftsmanship from beginning to end. Straight into the TAS hall of fame with other famous TAS artists.

 

I deliberately italicized that, because i 100% feel that's what this is. TAS Artistry.

[HavoX]

You bet this was worth the 9:42 minutes of my life.

[almostmatt1]

Thanks again for the very kind words, everyone! Honestly I've been a bit taken aback by the response to this demo, I'm very grateful that so many people have taken the time to watch it and leave comments.

 

I agree that a breakdown would be good, so I'm writing a tasvideos.org submission text that at this point looks like it'll be between 12,000 & 18,000 words describing all of the tricks and breaking down the technical and routing considerations throughout the demo. It's probably 3/4 done, and I'll post it here when it's finished.

 

14 hours ago, Redneckerz said:

Quality mapsets are rightfully reverred for their skill, but this TAS is of equally exceptional quality. When 19 hours is poured into making it, i wholeheartely believe that, and it shows.

 

Thank you Redneckerz! But I worry that I failed to be clear enough in my original post so I thought I should clarify, 19 hours is just how long I spent on it on the final day of working on it, I knew I was close so I decided to just devote one day to hammering it until it was done, but in total this took hundreds of hours over more than a year. For example, the last 6 seconds of map 2 took 4 days due to RNG problems with monster movement. I had map 20 completed, but went back and spent another 9 hours on it to save 7 frames, and that involved already having large sections of frames that I could directly copy over from my already completed map 20. I had map 25 completed up until the arch vile jump, but scrapped it and spent several days rebuilding because there was a single wallrun earlier in the map I could improve that saved 7 frames. Maps 29 and 30 on the final day took 15 hours, with the map 29 jump alone taking 6. Those numbers have stuck in my head, but every part of the demo was that much of a horrendously inefficient use of my time an optimisation effort.

[Redneckerz]

16 hours ago, almostmatt1 said:

Thank you Redneckerz! But I worry that I failed to be clear enough in my original post so I thought I should clarify, 19 hours is just how long I spent on it on the final day of working on it, I knew I was close so I decided to just devote one day to hammering it until it was done, but in total this took hundreds of hours over more than a year. For example, the last 6 seconds of map 2 took 4 days due to RNG problems with monster movement. I had map 20 completed, but went back and spent another 9 hours on it to save 7 frames, and that involved already having large sections of frames that I could directly copy over from my already completed map 20. I had map 25 completed up until the arch vile jump, but scrapped it and spent several days rebuilding because there was a single wallrun earlier in the map I could improve that saved 7 frames. Maps 29 and 30 on the final day took 15 hours, with the map 29 jump alone taking 6. Those numbers have stuck in my head, but every part of the demo was that much of a horrendously inefficient use of my time an optimisation effort.

All this tells me that this is only even more impressive than before. 100s of hours over more than a year. I was going to say if it wouldn't be easier to have an algorithm calculate the best method of running knowing all tricks, but the whole kick of built demo's is to find the best outcome on your own.

 

If this isn't even mentioned at the Caco's, ill cry wolf.

[Pirx]

you seem to run... fast? is doomguy going strafe50 each frame, or how does this work

[almostmatt1]

10 hours ago, Pirx said:

you seem to run... fast? is doomguy going strafe50 each frame, or how does this work

SR50 movement was used as much as possible throughout the demo, but I also took advantage of things like wallrunning, thingrunning and plenty of momentum preservation tricks very often throughout the demo which help a great deal in covering extra distance and keeping me speedy.

[almostmatt1]

https://tasvideos.org/7458S

 

The tasvideos.org write-up for this demo is done. It's about 12,000 words, and covers a lot of basic fundamental stuff as well as individual map-by-map comments about a bunch of tricks. If anyone at all feels as though I haven't explained certain things well enough or gone into enough detail, please let me know and feel free to ask me any follow-up questions at all. 

[Redneckerz]

@almostmatt1 before ill make a later comment on the entire production process, my mind went to this:

 

Quote

Tools used were XDRE 2.22a, my custom build of XDRE 2.22, based on PrBoom+ 2.5.1.5

What features does XDRE 2.22a add over 2.22 aside I implemented an XDRE feature that reasonably accurately generates a good brute force range for you, but with or without this feature a good manual setup prior to brute forcing is required?

[almostmatt1]

It's still a work in progress and there are more things I'd like to add and change, but for now it has a lot of visual differences and quality of life improvements. Information boxes for player movement, linedef tracking, sector tracking, thing tracking and other general info are now in more organised boxes that can be individually enabled or disabled if you'd rather have a less cluttered environment. The list of frames is considerably bigger so you can see many more inputs at once. The header editing window now has boxes you can tick and untick where relevant, whereas before you needed to type the number 1 or 0. Complevels can be selected with a dropdown box instead of typed, and you're warned if you choose a nonstandard one. There are some help windows on various screens. It has macro keys for MF50/MB50/SL50/SR50 inputs, since they're by far the most common. Plus a handful of other small things. The brute force range generator has been by far the most significant change for me though, it has automated a task I used to have to spend a lot of time working out and typing commands for manually.

 

I'd love to share it eventually but it's not yet in a state where I'm happy enough with it to do so. It has mostly been a good project for me to work on while I've learned how to program, so I'm not in much of a rush to get it to what I consider to be a releasable state. In particular I want it to at least have MBF21 support, the ability to brute force linedef skips, and Min/Max brute force conditions before it leaves my computer. By that time, Kraflab will have very likely implemented good enough building features to DSDA-Doom to just about satisfy everyone's needs anyway. :) 

 

Spoiler

Also I'm a moron and can't work out Github, so I have to learn that first too!

 

[Redneckerz]

Really accustomed to your own liking and i can tell it would be useful for many.

 

Ill get back assessing the entire breakdown, but for now, if anyone is coming in here and not being impressed on what's on display, just imagine this video, but set to Doom speedrunners. Because i am left thinking why so many of the radical left participate in speedrunning.

 

 

[MathsDevil]

There is some witchcraft involved here and I can't understand what's going on most of the time. Maybe Earth shouldn't let Doomguy back after he destroys the Icon of Sin, after all. Seriously --- well done! 

[Wagi]

Even with all the void glides involved, that MAP29 rocket jump might very well be the most insane part of the TAS.

[Searcher]

An outstanding achievement!  Great job!

[HavoX]

Well, it took one of the publishers a goddamn month, but it's finally been published.

[Dimon12321]

6 hours ago, HavoX said:

Well, it took one of the publishers a goddamn month, but it's finally been published.

The site itself is overloaded because the key stuff has to do multiple things at once: review movies, perform researches in various fields, program emulators. One of the things that delayed the publication was the implementation of wipe screen effect in PrBoom's -viddump feature for a more authentic DOS-like view.