Easier way to watch demos

[Grazza]

I haven't come across a version of the demo-pack with those names (that's why they are unsupported) - it sounds like someone renamed them.

Easy enough to incorporate. Just add "|(avold[0-3]\d-uv)" into demo_pattern31 (so that instead of "(av24tele))\.lmp" there is "(av24tele)|(avold[0-3]\d-uv))\.lmp" ), and it should work fine.

Any idea at all where you might have downloaded this from? Perhaps there are other renamed demos there too.

[emailking]

deleted

[Searcher]

I have a version like that demo set for the old av mapset. My zip is called avlmps.zip I got it in 2002.

Many of them run on the new mapset. And all of them run on the old mapset.

The naming convention above using the term "old" is 10 characters, so someone must have renamed them. On my system I renamed the new wad to avce.wad to keep them apart.

I can't recall where I got my old demos but they use the avXX-uv.lmp names.

I would have most likely gotten them from the old av page:
http://www.doom2.net/~av/

Or perhaps from DSDA. Not sure if this will help anyone.

I still have the zip if anyone needs it, just email me.

[emailking]

deleted

[Grazza]

BTW, the way I have set up the pattern-matching for the two releases of Alien Vendetta is as follows:

The wad named av.wad should be the current (compet-n) version. The wad called avold.wad (created by myk, IIRC) should be the one of that name that contains those maps that were different in the initial release. In the autoloading for old av demos, av.wad is loaded, followed by avold.wad.

This means that there is no need to rename anything (demos or wads). It also works fine if you have renamed the initial version to avold.wad though (the fact that av.wad is loaded first does no harm to anything, even though it is not needed in this case).

The pattern-matching also works for the files at the DSF FTP site, in case you have those (some of them were renamed for inclusion in the demo-pack as released).

http://www.doomworld.com/sda/av.htm

[Grazza]

Another update. This now includes patterns for almost all the demos included in with wads in the idgames archive. I say "almost all" because although I went through them systematically, this was with a slightly outdated copy of the archive. Also, some of the demos must have been recorded on different versions of the maps, or with an altered exe - I didn't include them when they were obviously hopeless cases.

This brings the number of patterns to 1417. Given that this is quite a leap, I decided to bump the version number to 9. BTW, the current test version of Prboom-plus has a couple of new features that are relevant to this. Firstly, bex files are now supported (one of the new patterns includes one), and the info given if not all files are found is more user-friendly. This means that if you have a lmp file lying around and have no idea which wads it requires, you may be able to find out quite easily. Click on the lmp: if it is recognized but the wads it needs are not found, the program will tell you their names. Many thanks to Andrey for these features.

Incidentally, I found a surprising number of 1.666 demos that wouldn't play back. Any idea if any difference in the engine might cause this? (Note that this can't have anything to do with differences in the iwad, as these are pwad demos.) I haven't managed to find much info on changes between 1.666 and 1.9 that might affect demo compatibility. I saw one vague reference to sound bugs being fixed, but nothing specific. Did sound propagate differently in some cases? Something to do with sound targets? It's possible that recording errors were to blame in all these cases (using wrong versions, etc.), but before I do time-consuming testing, it would be useful to hear if there are any "known issues".

[Kristian Ronge]

I can't believe I didn't post this before (I thought I had, but looking over the posts, I was obviously mistaken):

Graham, thank you very, very much for putting in so much hard work to make it easier for all of us to do something we love to do. :-)

[Ryback]

Yes, excellent work Grazza on all this.

I do have a feature request, although maybe it would be better off put to Andrey - would it be possible to mirror the structure of the /idgames archive, so that if you've got a local copy then Prboom can go to the appropriate zip file and extract the wad itself, rather than piling uncompressed wads into a separate directory. Because then this would be *perfect*...

[Grazza]

I wouldn't start to use the word "perfect" until/unless Andrey adds this feature - without that, quite a number of demos can't be made to play back without additional parameters being added by hand.

With respect to your idea, that might be nice, but maybe a lot of work. And many of the wads supported by these patterns are not in the archive, and there are also some naming clashes. So it might be fiddly in several respects. Additionally, many wad's names don't match their zip's file name, so there would need to be that additional info provided too.

BTW, while working on this I noticed some interesting files for those keen on "lmp archaeology" (the first four are all from 1993):

http://cd.textfiles.com/nogames3/006a/level3.zip
http://cd.textfiles.com/ugameware/DOOM2/DOOMTOUR.ZIP
http://cd.textfiles.com/ugameware/DOOM2/DOOMDEM1.ZIP
http://cd.textfiles.com/ugameware/DOOM2/DOOMDEMO.ZIP
http://cd.textfiles.com/ugameware/DOOM3/LMPVOL1.ZIP
http://cd.textfiles.com/ugameware/DOOM3/LMPVOL2.ZIP

[Searcher]

Those old demos are certainly an interesting find. Very cool.

[Grazza]

Minor update, only just worth mentioning. Much of the stuff at http://cd.textfiles.com (info) is now supported, with a new total of 1482 patterns (edit: amended and reuploaded, as I had forgotten something; now 1494 patterns).

A fair few of the new demos supported are for older versions of Doom.exe (1.2 in particular) and don't play back fully. Even in cases where they do, there is no stats screen, as 1.2 demos end at that point. I decided to include them anyway though, as there may be future improvements in compatibility, and also in case anyone wants to use these patterns for some sort of launcher that selects between exes, or whatever. So don't be too surprised if you load up a demo from the first half of 1994 and it goes haywire at some point.

Thanks to Opulent for giving me a collection of cd.textfiles.com content (amongst other stuff) in Prague - I had only downloaded a subset of this myself.

[emailking]

deleted

[Grazza]

Thanks for letting me know. Explanation: two demos with the same name (the other one is here), but on different versions of this map. No ideal solution.

But I daresay it makes sense to give priority to the compet-n one, especially as the earlier one also appears (renamed) in the AV demo-pack. I'll remove it from the pattern for avold.wad.

I wouldn't expect there to be many problems like this with AV demos. It will only have a chance to occur when the map changed between versions and the compet-n time on the new version matches one of the old demos (and that old one had regular compet-n-style naming).

[emailking]

deleted

[Grazza]

Another minor update, with a small increase to 1547 patterns, and a fair bit of growth in some of the existing ones.

Mostly just miscellaneous stuff from various places added, and some things improved. I've gone through the lmps section at the idgames archive a little more systematically than I had previously, finding quite a number DM demos that I had missed (there are now 215 DM patterns - I wouldn't have imagined there were anything like that many DM wads that people had recorded on), and adding support for irregularly-named demos in DHT exams.

[emailking]

deleted

[emailking]

deleted

[emailking]

deleted

[Kristian Ronge]

Those demos are two of Grazza's stroller demos made for the stroller pack. They were named et13s010.lmp and et14s018.lmp in the stroller pack, and they had the same names in Grazza's original strl_gb.zip. Maybe you (or someone else) renamed them?

[Grazza]

No, they are the names I gave them initially. I changed the file-names in a later version of strl_gb.zip, which is why these file-names don't appear in the version of that which is online.

OK, I'll check for any demos ending "strl" on my hard-disk and add support for them, just in case (like with these) other people have copies of them. But if you just watch the ones in the stroller pack (which is fully supported), you'll be seeing the same demos (or improved versions of them).

Edit: there were just two others that weren't supported - those on STRAIN (both also in the stroller pack, renamed). All the other demos were on wads that had sufficiently "greedy" patterns. I have deliberately kept patterns starting with things like "et" and "st" non-greedy, in view of the high probability of conflicts.

As for tricks in those Eternal stroller demos, 14 uses a switch trick first spotted by Sam Woodman. 13 is just a transitional map, designed that way so the player starts the first new map of Eternal's 2nd release with just a pistol. No tricks involved here - I just chose the quickest of the three options.

[Kristian Ronge]

Sorry, my mistake (as usual!).

[myk]

Grazza said:
BTW, while working on this I noticed some interesting files for those keen on "lmp archaeology" (the first four are all from 1993):

http://cd.textfiles.com/nogames3/006a/level3.zip
http://cd.textfiles.com/ugameware/DOOM2/DOOMTOUR.ZIP
http://cd.textfiles.com/ugameware/DOOM2/DOOMDEM1.ZIP
http://cd.textfiles.com/ugameware/DOOM2/DOOMDEMO.ZIP
http://cd.textfiles.com/ugameware/DOOM3/LMPVOL1.ZIP
http://cd.textfiles.com/ugameware/DOOM3/LMPVOL2.ZIP

Pretty cool. Well, watching them was dull at times, as they are like a random '94 map is to a Scythe2 level, when compared to most of today's demo uploads. But indeed interesting nonetheless from a more archeological perspective.

The first one desynched at the beginning for me on v1.0, for some reason (the E1M3 demo), and the second (DOOMTOUR) kind of sucks because the guy was complaining that you started with only a Pistol. So he played most levels on HMP. The third set is on a crappy skill level too, but DOOMDEMO bears the merit of being a full set of UV recordings made in '93.

If you compare these guys' movement with Romero's (from the DEMOx lumps), they pretty much suck, even the couple of mouse users that have an awkward sensitivity or a lack of practice. I guess we can say John and his id buddies were "Doomgods" by early playing standards, although the guys did of course have quite a few months headstart in getting acquainted with the game.

Also, in one of the v1.2 demos, NAT1.LMP, the guy is suprised that in E1M8 it appears that a Demon is still alive somewhere and he can't spot it. For two reasons; one, that it's an ExM8, where you can always hear monsters to a degree (as we discussed with fraggle for Chocolate Doom), and also that early versions of Doom had bad clipping sometimes allowing monsters to see you through walls. In this case a Demon in the end room awoke while the Player was atop the star. The E1M3 demo also has that bug; at the start a monster sees the Player before he opens the door. I recall many versions of ZDoom, at leadt till v1.22, had this bug (because ZDoom borrowed Heretic's clipping, which carried the bug from Doom v1.2).

In the other similar demo, NAT2.LMP, a Demon seemingly blows up a Barrel in the first area. And indeed it does. Demons in early versions had an attack that was similar to the Fist or Chainsaw. You can even see the blood splats when one hits the Player. If you were to move away from an attacking Demon it would continue its attack possibly hitting something else, in this case a Barrel, as it turned towards the Player's location. In the E2M7 demo, one of the Demons at the start hits the other in a similar fashion as the Player moves away and the second runs into the attack, and that Demon then starts attacking the offending one. I had often wondered whether reports of Demons or Spectres fighting each other could be true, and all this confirms it (I personally stated playing with v1.666, so I hadn't had any real experience with v1.2, so I was skeptical about this).

One part that amused me was a recording of E2M8 where the Player stood at one end of one of those cracks between a spike of the building and a marble block, waiting for the Cyberdemon to hurt itself with spalsh damage. For a second I wondered whether maybe in early versions it did affect the bosses, but there was no indication of this (the Cyberdemon didn't ever grunt from the many explosions he was exposed to).

Also, another curiosity, the demo showing a monster suicide (a Baron of hell clawing itself) was recorded by the guy who made UAC_Dead. I was like "hey, look, a demo by a 'DOOM celebrity'".

[Grazza]

I'm pleased to hear you found these of some value, and thanks for the detailed info you have provided - things to look out for when watching these and other 1.2 demos.

Regarding emulation of 1.2 (e.g. in prb+), I was aware of a difference that meant monsters could prematurely awaken (and that the start of e1m3 was particularly affected), but wasn't sure what the basic cause was. I had looked for relevant differences between the Heretic source code and the Doom source code, but hadn't found what appeared to be a smoking gun. Maybe your info about it being the clipping code will help in that respect.

The different demon attack sounds like a difficult one to emulate, since there will be no clues in the Heretic source on that one ... unless it turns out that the golem or sacreclaw's attack is an exact copy of the 1.2 demon's.

[Grazza]

Small update. Now 1564 patterns, and some improvements in the existing ones, especially for iwads. In particular, most DANG demos with irregular naming are now supported (there were some that weren't picked up by the defaults, most notably a couple of contests).

[emailking]

deleted

[Grazza]

I recall this happening at some points in the past, but right now I don't have this problem, and can't find any way to trigger it.

A few questions: are you normally running the demos from inside zips? (If so, what unzipping program?) Are you using the Fast Exit option? Are you pressing <End> to quicken the end of demos (rather than waiting for it to end naturally). Or using <Quit> for the same purpose?

Perhaps if you post your cfg file, I can see if there is anything you are using that might trigger something in Windows to be doing this. Otherwise, it just looks like some random Windows issue.

A rather radical solution, if this is really bugging you and there is no other solution apparent, would be to make a huge zip file of a mass of demos and run them from inside that, bypassing Windows Exporer entirely.

[emailking]

deleted

[Grazza]

In your cfg, find the line that says:

gl_colorbuffer_bits          16

Change this to

gl_colorbuffer_bits          32

This is sufficient to solve the problem for me (which I did observe when using your unmodified cfg). If that doesn't work, you can try 24. I'd guess that what matters is not so much the value itself, but whether it matches the value for your desktop settings. Edit: In case this doesn't solve the problem for you, then the next thing I would suggest is setting the resolution to match that of your desktop.

emailking said:

I apologize for the ignorance, but I don't know what the fast exit option is, although I have seen that listed. I didn't know there was an <End> method to exiting a demo. I don't press Q though. I hit escape and go through all the stuff.

The "Fast Exit" option can be enabled via the menus (Option - General; 4th page) or via the cfg (give "misc_fastexit" the value 1). This reduces the delay upon exit to a bare minimum.

The reason <End> is a quick way to quit a demo once the final stats screen has been reached, is that it causes the program to fast forward (without rendering) to the next stats screen. If it fails to find one, then it just exits once it reaches the end of the demo.

I have <Quit> bound to my Y key. This is so that when it asks the "Are You Sure?" question, this can be cleared with a second tap on the Y key. So basically a double tap on the Y key exits the program immediately.

[emailking]

deleted

[Grazza]

I don't know exactly why it has those effects, but it seems natural that the more in tune the program's settings are with those used by your OS/display, the fewer problems there should be.

In general, it makes a lot of sense to have a look through all the settings in the cfg (and perhaps those for your display/video card) just to see what there is that you can adjust (cautiously!), and if doing so is of some benefit.