Difficulty the same under various ports?

[Pirx]

As I was testing Plutonia 2 with Gzdoom and Jdoom I noticed that somehow the maps appeared more difficult in Gzdoom. I played map 1 a few times using both ports on UV.

For example, in Gzdoom I got shot pretty badly every time i rushed the 4 former humans at the start with the chainsaw, while in Jdoom they died most of times without causing too much trouble.

Same for the first caco at the right door, it managed to bite me twice at a time as I was sawing it. In jdoom it did what they mostly do: turned, tried to get away and died.

The pain state seemed to be more reliable in jdoom, preventing monsters from such actions.

On the other hand, i'm quite sure the developers didn't change such fundamental parts of doom's gameplay like pain state or damage tables, so if it wasn't just my bad luck or a statistical aberration (i didn't test hundreds of times for a more precise average), then this difference was possibly caused by different collision checking.

Perhaps the programmers can shed some light here. I'm also interested in your personal experience with maps that seemed to vary in difficulty when played with different ports. I'm excluding jumping as a factor here, obviously. I didn't use it in jdoom since plutonia2 disables it in gzdoom.

Visibility might also be a culprit. For example plutonia appears less brownish in jdoom, at least on my LCD, making it easier to see through those vines hanging in every door frame.

Granted, I rarely get to play these days, so my doom-fu is quite weak now.


Secondly, while we're at it: what port offers you the most crisp and accurate mouse control in your opinion? don't turn this into a silly port war please.

[EarthQuake]

While I play neither GZDoom nor jDoom, I've experienced the same thing with other ports. I think this mostly has to do with the randomization routines, where some ports show more truly randomized results, instead of pulling numbers out of the same set of generated numbers (where some numbers might not be present, and where some might be present multiple times), producing a fixed set of outcomes.

I don't have any experience in programming with the Doom source, so I can't shed any solid light on the matter, but I'm just stating that I tend to notice this as well. For me though, source ports that stick closer to vanilla behavior seem a bit harder than those like ZDoom. I'm sure physic changes and graphical enhancements play a part in it too.

I'm interested in what the developers have to say as well. Until now, I never really brought this up because I thought it was something that I was imagining. :)

[kristus]

Well. There's two sets of randomisation in Legacy for instance. the regular vanilla random that isn't really random. It's pseudo random so it works with demos. Then there's true randomization that actually is random every time. Maybe that's what one of the ports are using and the other is not.

[EarthQuake]

Eh, all source ports use a pseudorandom number generator. It's just the variability of the output that makes them different. For example, vanilla uses a table of static numbers that doesn't change from game to game. Other ports probably get around the limitations of this by filling the table with different values each game. If this is the case, my guess is that a seed is used to produced the same table when a demo is recorded, so randomized events occur properly during playback.

Afaik, a truly random number generator isn't possible, unless you base the system off of something like microwave background radiation. Or something. :P

[DaniJ]

In jDoom we stick pretty close to vanilla DOOM play mechanics in most respects, especially when there might be an impact on gameplay. For example, many of the "quirks" of the original DOOM collision detection are either left "broken" or have been re-implemented in such a way that their anomalies can be retained/reproduced without changing the game balance (much). Sadly, the majority of these decisions are not sufficiently documented at present (at code-level, or otherwise).

Many of these decisions were made very early on in development and required high-level design choices in order to make (like embracing a client/server arch and dumping the DOOM demo stuff entirely).

[esselfortium]

I think it's a combination of different collision checking for players and for hitscan. I don't know about Doomsday/JDoom, but ZDoom fixes Doom's buggy blockmap-checking code that would cause melee and hitscan attacks to sometimes not affect something that they should, so as a result both players and monsters are somewhat more prone to taking damage from them.

[HackNeyed]

Yeah Torr, I use PrBoom-Plus and I find Boom's Doom Compatibility level more difficult then pure Doom compatibility level settings... mostly when dealing with hit-scanners.

[Nuxius]

Torr Samaho said:

Secondly, while we're at it: what port offers you the most crisp and accurate mouse control in your opinion? don't turn this into a silly port war please.

The best is definitely GZDoom. The worst would have to be Edge, followed closely by Doomsday (don't know about the 1.9.0 betas though).

[Khorus]

heh Doomsday is my favourite for mouse control. The latest Eternity is very good too.

[Graf Zahl]

esselfortium said:

I think it's a combination of different collision checking for players and for hitscan. I don't know about Doomsday/JDoom, but ZDoom fixes Doom's buggy blockmap-checking code that would cause melee and hitscan attacks to sometimes not affect something that they should, so as a result both players and monsters are somewhat more prone to taking damage from them.

That is a factor in hitscan attacks but not in monsters' melee attacks. Those only do a simple distance check that doesn't rely on the blockmap or other resources.

If those and general movement behavior are different it's more likely the random number generator which is causing these differences.

[Maes]

The vanilla Doom RNG is basically a table with 256 8-bit positive integer values spread around, which are used cyclically.

Yeah, some values are not present, and the probability of certain sequences to appear is very low or even impossible in actual gameplay (e.g. firing the shotgun uses no less than 14 sequential values for damage and spred, and there's no chance of obtaining the maximum damage since there are no 7 alternating values leading to maximum damage per pellet anywhere on the table).

Source ports on the other hand use higher-quality RNGs, while some retain the original table for demo/bug compatibility. No idea if some of those RNG are used in a broken way, but it can be entirely possible to consistently experience e.g. shotgunners doing 45 HP worth of raw damage, or getting 20+ damage from each imp fireball, making them only marginally less dangerous than barons or revenants.

[myk]

Torr Samaho said:
Secondly, while we're at it: what port offers you the most crisp and accurate mouse control in your opinion? don't turn this into a silly port war please.

I don't use too many different source ports, but on my computer ZDoom engines seem smoothest and most accurate. The DOS executables are about the same, though. SDL based engines, on the other hand, tend to be a bit more inaccurate or loose. That said, while recording in Doom (or Doom compatibility or something similar) the mouse is notably less accurate, unless the longtics hack is used.

[Pirx]

Graf Zahl said:

If those and general movement behavior are different it's more likely the random number generator which is causing these differences.

that's exactly how things looked to me. sometimes the mentioned 4 former humans at the start just walked around without firing, or did minimal damage, and on other occasions they started shooting as soon as i opened the door. especially the shotgun sergeants were annoying. iirc the shotguns have a random spread, so this might be the cause.

myk said:

I don't use too many different source ports, but on my computer ZDoom engines seem smoothest and most accurate. The DOS executables are about the same, though. SDL based engines, on the other hand, tend to be a bit more inaccurate or loose. That said, while recording in Doom (or Doom compatibility or something similar) the mouse is notably less accurate, unless the longtics hack is used.

that's my impression too, even with 3 different mice.

[ReFracture]

I've always felt that ZDoom was easier to play, especially on tougher maps like Hell Revealed 2's.

Torr Samaho said:

Secondly, while we're at it: what port offers you the most crisp and accurate mouse control in your opinion? don't turn this into a silly port war please.

ZDoom and/or GZDoom, for sure. I prefer PrBoom-Plus most of the time, but I find it's mouse control to be fine after getting used to it (I used to hate it.)