Hi-Res support for NeoLemmix? Bring back SuperLemmini?

[WillLem]

Really interested to see what the general consensus is on this. If demand is high, I'll be up for getting involved in a re-vamp wherever possible.

Let me know your thoughts!

[namida]

If you want an actively maintained engine that also has high res graphics, take a look at Lix.

As the only currently active developer of NeoLemmix, high res graphics are something I can agree would be neat, but extremely, extremely low priority. Of course, NeoLemmix is open source, so you're welcome to submit code for this yourself - I'll integrate it if it can be done as a purely visual change with no impact on physics, sure. (Anything that introduces two different sets of physics is a big "no" from me and will remain so as long as I'm in charge of NeoLemmix; while just having a single set of physics but changing it from the current set would need significant community agreement, but I am open to it - though I think it's very unlikely that the needed community support would be there.)

I'd also ask whether it's really worth just supporting the WinLemm resolution, which is only double that of DOS and still very low by modern standards. The only advantage over a higher one is that graphics already exist for the official styles - but not the 50+ custom ones.

[ccexplore]

Can someone remind me whether SuperLemmini had its own level editor at any point?  Seems absurd to me that NeoLemmix was the only level editor for it.

The blame lies squarely on the maker of SuperLemmini for leaving it in such a state where somehow it needs some other game now as its level editor.  SuperLemmini has been around for a while, seems like it has gotten every chance there is to be something as actively used as the other games here.  There's probably a lot of history I glossed over, but perhaps how SuperLemmini fared over time compared to the other games here, also speaks a little to the general preferences of this community in terms of what they value most?

[namida]

Can someone remind me whether SuperLemmini had its own level editor at any point?  Seems absurd to me that NeoLemmix was the only level editor for it.

Neither Lemmini nor SuperLemmini have ever had their own dedicated level editor. The old Lemmix editor, and early versions of the old NeoLemmix Editor (the one based on the Lemmix editor), were able to edit regular Lemmini levels. LJLPM's jLevelBuilder, which I believe is designed for his Android clone, also has support for Lemmini levels. The old NeoLemmix Editor specifically had support for most SuperLemmini features, including a couple that NeoLemmix didn't support, and it retained this SuperLemmini support for a while after dropping support for Lemmix and Lemmini levels - it still retained, right up to the last version released, support for loading SuperLemmini levels; and new-formats NeoLemmix itself is capable of loading SuperLemmini levels, assuming a suitable translation table for the style exists. (The new-formats level editor only supports new-formats NeoLemmix levels (NXLV), although it can ask NeoLemmix to convert any level it supports to NXLV.)

SuperLemmini is fairly (but not completely) backwards compatible with Lemmini, so creating levels with a Lemmini editor and porting them over is feasible, as long as you don't want to use things like vertical scrolling.

In the case of Lemmini itself, the lack of an editor was more justified: The author developed it primarily as a way to play the official game on non-Windows / newer Windows systems, not as an engine for custom content.


I think though, a big reason SuperLemmini fell out of favor was due to the shift in perception in the community of how Lemmings clones should play. SuperLemmini's creator chose not to go along with the fine-control, "puzzle, not execution" approach that NeoLemmix (and before it, regular Lemmix) and Lix had been taking for some time - even untimed bombers was something the creator was very reluctant to add. The community on the other hand had developed a preference towards this style, and thus, an engine that didn't follow it didn't have much hope of overthrowing the two that do.

Maybe in the future, this preference will change again. After working on a custom Lemmings 3D pack a while ago, I've started to really see the advantage to tile-based rather than pixel-based mechanics. Depending how far the "avoid execution difficulty" mentality goes, maybe in the future a tile-based engine might overthrow both NeoLemmix and Lix. I do, however, suspect it's unlikely that we'd go backwards, towards higher execution difficulty.

[ccexplore]

To echo what a few others have already said here and elsewhere:  doesn't Lix pretty much fit the bill completely?  You got the high resolution graphics and all the 8 classic skills are there.  Other than the change of timed bombers to instant bombers and the removal of release rate control, and the use of all new graphics, sound and music to avoid copyright issues, isn't Lix basically what SuperLemmini would be if it kept up with NeoLemmix?  What is missing there for you to still prefer SuperLemmini?

[namida]

There are some object types supported by SuperLemmini but not by Lix - one-way arrows and directional force fields, to be precise. It is possible to replicate these setups by carefully constructed terrain, though.

There's also a few "these mostly exist to be unfair" features such as fake or invisible terrain in SuperLemmini, but I'm hoping that's not what appeals to anyone about it. It also has support for backgrounds made up of arbitrary terrain pieces rather than a single preset image (NeoLemmix) or none at all (Lix), although this is virtually a moot point given that no editor has ever supported this. Another difference would be the lack of support for variable release rate and time limits in Lix.

One possible other concern people might have is to do with the graphic bit depth, rather than the resolution, in NeoLemmix. NL uses the DOS versions of the official graphic sets, which you're well aware of the palette limitations of. However, this is not an inherent restriction of NL itself anymore - recent versions of NeoLemmix even support alpha blending, let alone just 24-bit color, which in this regard even puts it ahead of Lix (which supports 24-bit color, but transparency is an all-or-nothing deal as far as I'm aware). And yes, there are specific algorithms in play to determine the alpha-blended physics map that don't rely on any specific graphic library's implementation. You just have to dive into the custom graphic sets to actually see this - plenty that use 24-bit color, but as far as I know, only the three LPVI sets use alpha blending (and even then, only for objects, as it wasn't properly supported for terrain at the time LPVI was made).

[WillLem]

If you want an actively maintained engine that also has high res graphics, take a look at Lix.

As the only currently active developer of NeoLemmix, high res graphics are something I can agree would be neat, but extremely, extremely low priority. Of course, NeoLemmix is open source, so you're welcome to submit code for this yourself - I'll integrate it if it can be done as a purely visual change with no impact on physics, sure. (Anything that introduces two different sets of physics is a big "no" from me and will remain so as long as I'm in charge of NeoLemmix; while just having a single set of physics but changing it from the current set would need significant community agreement, but I am open to it - though I think it's very unlikely that the needed community support would be there.)

I'd also ask whether it's really worth just supporting the WinLemm resolution, which is only double that of DOS and still very low by modern standards. The only advantage over a higher one is that graphics already exist for the official styles - but not the 50+ custom ones.

Hi namida, thanks for your reply and for engaging with this topic, it's really great to be talking about this with the NL developer!

I have tried Lix but find that it just... isn't Lemmings! The gameplay is the same, more or less, but I'm a very visually-oriented person (hence the obsession with Hi-Res graphics!) and the look & feel of Lix is not my cup of tea. I also think that wayyy higher resolution, such as in the Sony versions, is going a bit far - WinLemm is a nice balance between slightly prettier graphics and old school look/feel.

Thanks for offering to let me have a go at coding. I have to say though, my coding skills are extremely limited (I don't have any!), but I am good enough with cartoon graphics to be able to help with getting a Hi-Res set together for NeoLemmix.

As a start, and to see if you're interested in this offer, I have attached red & green versions of the WinLemm animation sets for walker, faller, climber, floater, bomber and splatter. These would be for the 'highlighted' and 'skill assigned' Lemmings in NL (which is a super idea, btw - as are the forward & reverse step-frame buttons. Very nice!).

For some reason, I can't seem to find the animation sets for blocker, builder, basher, miner or digger anywhere in my WinLemm directory... if you can point me towards these, I'm happy to give these the same treatment.

Furthermore, if you like I am more than happy to create WinLemm-style graphics sets for all of the custom skills included in NeoLemmix. If you send me over the animation frames for each one, I'll scale them up to 24-bit and add the extra colours/pixels.

Of course, I'm assuming here that the WinLemm graphics follow the same physics as DOS/NL... if I'm wrong about this, please do let me know.

Best,

-WillLem

[WillLem]

SuperLemmini has been around for a while, seems like it has gotten every chance there is to be something as actively used as the other games here.  There's probably a lot of history I glossed over, but perhaps how SuperLemmini fared over time compared to the other games here, also speaks a little to the general preferences of this community in terms of what they value most?

Fair enough. It would be nice to have the additional option though, for those of us who prefer the chunkier look of WinLemm/SuperLemmini

[WillLem]

I think though, a big reason SuperLemmini fell out of favor was due to the shift in perception in the community of how Lemmings clones should play. SuperLemmini's creator chose not to go along with the fine-control, "puzzle, not execution" approach that NeoLemmix (and before it, regular Lemmix) and Lix had been taking for some time - even untimed bombers was something the creator was very reluctant to add. The community on the other hand had developed a preference towards this style, and thus, an engine that didn't follow it didn't have much hope of overthrowing the two that do.

Maybe in the future, this preference will change again. After working on a custom Lemmings 3D pack a while ago, I've started to really see the advantage to tile-based rather than pixel-based mechanics. Depending how far the "avoid execution difficulty" mentality goes, maybe in the future a tile-based engine might overthrow both NeoLemmix and Lix. I do, however, suspect it's unlikely that we'd go backwards, towards higher execution difficulty.

I've had a play with NeoLemmix over the past few days whilst I've been active on here. I have to say, I'm loving the emphasis on puzzle-solving rather than execution. It's a welcome development that I've been missing out on whilst I've been mainly playing the Windows & Lemmini versions. Again, I prefer those purely because of the graphics and the pretty GUI (as mentioned above, I'm a visual kinda guy). However, the gameplay of NL is far better! I can see why people tend towards this version, and why SuperLemmini fell by the wayside.

Maybe we could have the best of both world by integrating 24-bit graphics as an additional option. That way, users who prefer the DOS graphics ("we fear change!") can stick with that, whilst those of us who appreciate the chunkier look of WinLemm/SL can choose to enjoy NeoLemmix HD!

(I say HD tongue-in-cheek, of course! )

Thoughts?

[Dullstar]

I would like to note that I don't consider a fear of change to be my reasoning for preferring the low-res graphics to the high-res graphics (sorry, you're the one who said "we fear change!").

Some modernized ports of various games have decent graphical face-lifts, but WinLemm's, in my opinion, aren't very good from an aesthetic point of view. They're certainly higher res, and therefore more "technically impressive" but that doesn't mean they actually look better due to the actual quality of the art direction. I find the terrain of the DOS versions more believable because the shading is such that everything looks like it's actually made of the material it's trying to look like, but the WinLemm graphics are too shiny (especially on the dirt set), which makes everything appear much more metallic than it should. Higher resolution, lower quality.

[ccexplore]

That way, users who prefer the DOS graphics ("we fear change!")

I know you're half-kidding here, but let's be crystal clear that if anyone prefers the DOS graphics it probably has nothing to do with "fear of change".  NeoLemmix already changed in many, many ways from original Lemmings.  Just like most of the people who bought NES Classic Edition are probably not also avoiding all the newest generations of Nintendo consoles.

Given that Lix started off from inception with higher resolution graphics and people hadn't been complaining about that, I'd say most people here are actually pretty neutral when it comes to graphics resolution.  That said, I think the record here with NeoLemmix at least, also looks pretty clear that higher resolution graphics support has never been amongst the more frequently requested feature; we had been seeing far more requests (well, so far anyway) for things like new skills for example.

Lix consciously avoids borrowing any graphics from the original games to steer clear of any copyright issues.  So the game in the official repository can never include the WinLemm graphics for example.  But there's nothing preventing you or anyone else (well, other than time and effort) to port the WinLemm styles over to Lix, and provide the files for other people to put into their copy of Lix.  Lix doesn't currently support a level customizing the lix animations, but there's also nothing preventing you from changing the graphics in your own copy of the game to use WinLemm's (though you'll have to make your own version of animations for the lix skills that are not the classic 8, and some of the lix skills have deviated a little from lemmings so it might not look quite right to directly use WinLemm's animations in Lix without some tweaks), in other words modding them globally to look like lemmings than lixes.  AFAIK most if not all graphics in Lix are in PNG files that you can mod yourself without changing any of the game's programming.

[namida]

Okay, so for starters - we wouldn't need recolored sprites. NeoLemmix recolors them via code, not via alternate spritesheets.

What we would need graphically, is (a) high res Xmas lemming sprites, and (b) high res original lemming sprites for 8 of the 10 new skills in NeoLemmix (the Walker and Cloner don't need sprites).

However, as mentioned, this is going to be very low priority on the coding side, and likely, any creation of them at this stage will be a wasted effort. If anything, it would make more sense for the code to be there first, perhaps using quick and dirty upscaled sprites as placeholders during development, and swapping in proper ones later.

A more immediately useful effort might be to recreate the graphic sets in higher quality. Same resolution, but full 24 bit color (alpha blending would need to be avoided to prevent physics breaks on existing levels). NeoLemmix already supports this, as you'll see if you look at graphic sets like those from Lemmings Plus IV onwards.

With that being said, some people may be averse to such a change, so it would need to exist as a mod rather than a change to NL's standard copies of these sets.

[Simon]

I suppose you want hi-res physics where 1 pixel equals 1 physics unit. Then be careful with near-invisible terrain relics that impact physics. Nasty source of annoyance in Clones; the clones would constantly hit their head and turn.

Also, hi-res physics with 1 pixel = 1 physics unit ("chunk") will produce really narrow vertical gaps where lemmings will fall through, but that are easy to overlook.


The alternative is to define a chunk to encompass several neighboring graphical pixels, but that is a constant source of pathologies. We must invent a rule that governs exactly when a chunk becomes solid.

The obvious rules, e.g., every 2x1 chunk (two pixels next to each other) becomes solid if at least 1 pixel is opaque, produces weird behavior dependent on horizontal offset. The right half of this example is a mere shift of the left half's terrain, but the shift is by an odd number of pixels. Since a 2x1 chunk is defined as an even-coordinate pixel together with its right-side odd-coordinate neighbor, and a chunk is defined solid if at least one of its two pixels is opaque, shifting terrain by an odd amount affects the physics of the terrain.

I suppose that that there are similar pathological examples for the Winlemm rule: 3 out of 4 makes a 2x2 chunk solid.


Maybe there are fancy new and exciting rules that I haven't considered. E.g., maybe chunks shouldn't be reliant on global coordinates at all, unlike the 2x1 chunks of Lix. A level should ideally play exactly the same after you shift everything by 1 pixel to the right.

-- Simon

[namida]

Given the complexity involved in a physics change, I'd think any change to high-res in NL would need to take one of these two approaches:

1. Physics stay as close as possible to current. The "is pixel solid?" check, instead of checking a single pixel, checks a 2x2 area (which may start at an odd offset). This would still need significant changes, as we could likely no longer pre-determine each individual pixel's solidity in cases of alpha blending, but would instead have to do so on-the-fly during runtime due to the different ways the pixel could be included in a 2x2 chunk. As such, even this approach would be pretty complicated.

2. Physics don't change at all, and the higher-resolution is a purely visual element. Terrain pieces, objects, etc can only be placed on 2 pixel boundaries, and a low-resolution physics map is generated for them, probably again using some 2x2 area rule (but based on the terrain piece's standalone graphic, not the overall level physics map).

[Simon]

Or keep lo-res, and optionally pipe the output through a pixel art scaler. They're a massive improvement over blurred bilinear scaling.

But does anybody really want lo-res physics with hi-res rendering? I still believe that WillLem wants hi-res physics.

-- Simon