The Contest (or, The Vanilla Question)

Before I can go through my thoughts on the entries of this contest, I have to talk about the difficulties this contest faced early on. When VLDC4 initially started, it mostly kept the same rules as VLDC3, with maybe some extra patch exceptions thrown in. The main changes went towards level length; in addition to the previous rule allowing only two exits, levels required a midway point (unless deemed short enough, I guess) and had to be under 75 screens long. For comparison, most SMW levels are about 20 screens, depending on sublevels, so you could still make really long levels, but it at least prevents them from getting too out of hand. Either way, these were pretty welcome changes to the contest.

However, five days later, the contest was rebooted completely, and even greater restrictions were placed. ExGFX was disallowed entirely, meaning no more mixing different tilesets. None of the fix patches were allowed either (aside from any within Lunar Magic itself), and custom music was once again off the table. So what happened? Well, since VLDC3 there were debates about what should be considered a vanilla level, especially concerning levels like Forever Factory. These debates continued into VLDC4 and intensified, probably playing some part in the change to the new rules. The real issue, and what FPzero stated was the reason for the changed rules, was that he was just getting bogged down by all the different requests and questions people had, whether it was different patches they wanted to use, or tiny graphical edits in YY-CHR they wanted to make. I can’t blame FPzero for wanting to start the contest fresh, as not only did it become too confusing to tell what changes were allowed and what weren’t, but it’s just tiring to read – let alone answer – all these questions that ultimately seemed to misunderstand the point of the contest.

Still, I feel this was a step backwards for VLDC. When VLDC4 first started, it was labeled as the “Vanilla Resources Contest”, and I think tileset merging fits that description. After all, everything used is still from the original game; it just requires an additional tool outside of Lunar Magic. Usually arguments against tile merging are related to aesthetics vs. design arguments, which I feel is more reliant on how the scoring is handled than whether or not tile merging is allowed (I will discuss the scoring for the VLDC4 and the previous contests later.) And, for what it’s worth, there was rarely any levels in the earlier VLDCs where I felt design took a backseat from appeal, and if anything VLDC4 had more egregious examples of that taking place, when tileset merging was disallowed. I think there’s this collective amnesia that tileset merging is a newer development of VLDC rather than something that existed from the start and was later removed; I hope this perception starts getting corrected, but I doubt it.

(I’m only mentioning this as an aside, as I don’t really have the technical knowledge to get into detail on things like Lunar Magic updates. The more eye-catching updates from the readme are expansions to Map16 and Direct Map16 Access, changes to the background and palette editors, a save-restore system, and a complete overhaul of ExAnimation. From the discussion threads, the most talked about feature new to LM was smkdan’s “VRAM modification patch”, which seems to expand the amount of graphical tiles used in a level. This is probably what allows VLDC4 to stand on its own in appearance compared to the previous VLDCs, but again, I couldn’t explain to you the differences at all. Unfortunately it’s a blind spot in my analyses.)

The other rule changes, on the other hand, are much more agreeable. YY-CHR edits should’ve just been disallowed completely, and I’m a little surprised entries got away with them back in VLDC3. Even though it was usually just for minor changes like fixing cutoff, it clearly crosses the line of what isn’t vanilla anymore. Also, I agree that custom patches, even just for fixes, shouldn’t be allowed; considering how many fixes Lunar Magic has already added over time, I think it’s fair for the authors to just work with what they got in that regard. I’m more accepting to custom music, even though it’s clearly not vanilla. I think the arguments defending it are pretty poor, though; people (and even the VLDC3 rules) argued that people can simply not judge the music, but it’s pretty obvious that listening to a level with custom music completely changes the experience. Honestly, I would just rather listen to different music than the paltry amount of original tracks, for variety’s sake.

Ultimately, I can’t blame FPzero for changing the rules so drastically; I would be frustrated too having to deal with so many people trying to find loopholes in the rules, and nearly always just for graphical purposes. Still, I can’t help but feel a little disappointed by this change, and that it would last for both VLDC5 and VLDC6 (not counting 6’s own weird spin on the rule). It’s a shame, because I think FPzero addressed this matter pretty well back in VLDC3:



Nevertheless, those were the rules. It was now up to the entrants to bring a good showing.

The Levels

In VLDC3, I mentioned how there were a lot of different ways certain levels failed, like how there were seven different broken patches that all seemed to break in different ways. Well, VLDC4 had zero broken patches submitted as final, something not seen since VLDC1; as it turns out, this is pretty emblematic of this contest also. After the sluggish, sprawling entries of VLDC2, and the scattered, chaotic entries of VLDC3, VLDC4 is a much more solid collection of entries. Despite all my grievances laid out earlier, I have to admit, this felt like a return to form for the contest.

Notably, there were more returning entrants for VLDC4 than there were for VLDCs 2 and 3 combined. Most of these came from the crowded VLDC3 field, though there were a few returnees from VLDCs 1 and 2. Even besides that, 61/90 of the authors joined SMWC while VLDC3 was still going on, a huge increase from previous contest and the first time that faction represented a majority. It seems like the designers are much more experienced overall this time around, and that checks out with the entries themselves; not that many levels feel terribly designed this time around. There are definitely some recurring issues, and we’ll get into those, but I think the combination of author experience and level length requirements made this contests’ levels much more solid to play, if not too outstanding.

As the perfect example for this, take Grinding Guides by Nimono, my favorite entry from this contest. Nimono is the only author that has entered every contest since the beginning, and has always had memorable, though flawed, castle levels. The levels tended to be pretty ambitious and experimental – Frost Cavern Castle was one of the earliest instances of a VLDC level using 1F0 tiles for a puzzle, and that was back at the very beginning. However, they tended to suffer from some design flaws which made the levels too tiresome to play, and were never able to break out of the middle-lower tiers. With Grinding Guides, however, Nimono took a simple concept, made great and varied obstacle designs with it, kept the difficulty from being overwhelming while still challenging, and also gave it a nice appearance. Maybe it’s not the most ambitious level, but I do think the years of experience the author had designing levels contributed to how tightly the level played.

Along that same line, Unlock the Key by x1372 is a huge improvement on the author’s previous levels. This level feels sort of like a sequel to their VLDC2 level, Curse of the Golden Mushroom; that level was very much of its time, a 7-exit gauntlet that only ended up being okay due to good pacing for each section. With Unlock the Key, the design is much more focused, with several different rooms requiring the player to use the key in the different way. The ending sequence is great at utilizing layer 2, and overall it’s a shining example of puzzle-platformer design in SMW. Sadly, the author left the site shortly after this contest, but they definitely ended their VLDC run on a high note.

Dottedgirl and CK Crash both deserve mentioning for hitting the 3-contest mark, even though the former didn’t appear in VLDC4, and the latter’s level was pretty average. However, the true honor belongs to WitherSkellie for consistently making top 10 levels three years in a row. Even though Ghosting Factory has a couple technical issues, and the castle-ghost combination isn’t anything new, it’s executed well and very fun to play. It’s just impressive how consistent this author has been for three contests, and I really want to explore their work further now.

As far as the other levels in this contest, most of them can be put into two categories. The more interesting of these groups are the levels that have creative ideas and tight design, but suffer from being too difficult and long to really enjoy. Levels like Sand Castle by neosaver or Skull Island by Aja are typical; the former has possibly the best beach sequence I’ve seen, while the later has a really unique gimmick with the time-limited caves. Unfortunately, each of their second halves extend the levels way too long and contain really difficult segments to where it starts becoming aggravating. Fort of Confusion by Artsy3… utilizes a mirrored level gimmick to a more developed degree than any other VLDC level, but by making the level so hard it doesn’t let the player get used to this kind of level without a lot of practice.

The most unfortunate case of this, in my opinion, is Vic Rattlehead’s Shining Moon Tourguide. It offers split paths with some great platforming and hands down the best P-Switch run section I’ve ever played, but there’s some technical problems that really hurt the flow of platforming; the elevator section, while a really cool idea, suffers greatly from slowdown early on, at least until the player hits a few P-Switches. It’s a shame, because this could’ve been a top three level easily if these were fixed, or if the difficulty was toned down just a little.

The other group is the many bog-standard levels made for the contest. Like with VLDC3, I had to make a whole category for the most average of these levels, though I did change the name to “Vanilla Woods” tier to fit the overabundance of forest levels. I don’t really have a problem with a certain theme being popular in a contest, but I feel a forest theme is a pretty bad choice; because the standard vanilla forest background requires forcing the camera to not scroll, else it looks screwy, there’s not a lot of vertical movement a player can make in these levels. (This becomes particularly painful in An Autumn Adventure by PowerStrike, which tries to have two vertically split paths without allowing the camera to scroll up.)

Case in point: Forested Field by E-Man (not to be confused with E-man38). Admittedly, it’s one of the better forest levels, with good execution and unrestrained platforming that prevents it from feeling stale. But I never would’ve expected this to be the first place winner of the contest; it doesn’t really push the boundaries of Vanilla SMW in any way, and the platforming isn’t THAT spectacular. I don’t want to turn this into an argument with the (half-inactive) judges, and it is a pretty good level anyways, but it’s definitely the most modest of the first place levels thus far; certainly it’s an 180 from something like Forever Factory. Speaking of which…

Aesthetic vs. Design

Seemingly intertwined with the arguments of what constitutes “vanilla” hacking is the vocal emphasis on putting level design over level appearance. While it’s mostly agreeable to say that how a level plays is more important than a level looks (though some have argued differently), there’s a perception that people, typically implied to be the judges, tend to like levels with flashier graphics more. Whether or not that’s a fair assumption, this kind of grumbling began at least as far back as VLDC3 with Forever Factory. For VLDC4, the focus was put on two different levels: Lunar Limbo, by Uhrix and Incognito and 105 Days by Ludus.

Lunar Limbo is maybe the best-remembered level of this contest, and for good reason. While these silhouetted or single-color levels have been done plenty of times in the early VLDCs, here it’s done remarkably well, with great atmosphere and graphical details; it wears its inspiration well on its sleeve. Design-wise, it’s not so solid; while the middle areas play fine (though the swimming section is a little rough), the opening segment has some pretty obtuse moments, and it’s hard for the player to tell where they’re supposed to go. Maybe that’s intentional, but it comes off clunky. The ending section, meanwhile, relies a bit too much on precise Podoboo jumps, and has the player go all the way back to the beginning of the section just to get a key stuck in the ceiling...just to go all the way back to the end of the level again. Also, the level doesn’t properly end; you go into some sort of transitional room and then the level loops again. It’s not really a flaw with the level, just something I found weird. Overall, it’s an enjoyable level, but I can understand some of the criticism regarding its aesthetic focus over its design; still, I’d recommend checking it out.

(BTW, there’s a few bits of trivia I should mention. One of the rule changes I haven’t brought up yet was the allowance of partners, two user per entry. Only six people (or three pairs) took that offer, so I didn’t think it was worth bringing up in detail. Coincidentally, one of the other partner levels used a similar single-color gimmick; Stuck in a Rainbow by Argumentable and limepie20 was much flashier and had much more complicated design, but suffered from some really annoying obstacles and enemy placement.

I should also mention that this was the contest where Uhrix designed the original VLDC trophies, with a little bit of help tweaking the trophy colors. They’re pretty nice, though admittedly they still kinda look like candy bars.)

For those familiar with Vanilla Rain from VLDC3, you probably have an idea of what 105 Days is like. Personally, though the level still has some serious design issues, it’s a marked improvement from Ludus’ previous level. While the platform layouts are on the cluttered side, the space station graphics are pretty incredible for the limitations of this contest. And even though it’s only in the level for a tiny bit at the end, I really like the bramble/vine section a lot; it would’ve been interesting to see a level just based on that. Unfortunately, the actual design of this level is a mixed bag. It opens with a ridiculously obtuse hidden door “puzzle”, and a lot of the obstacles either feel kind of janky or completely skippable. Also, it requires following its own savestate points, or else you’re going to replay a whole lot of the level. I think it’s this last point that really scored it low with the judges, even though I appreciate a lot of what it tries to do, and overall feel it’s a much more competent effort by this author (though I still didn’t get much of the story).

For all the attention those levels got, I think they both have merits beyond their aesthetics. Lunar Limbo, for all its design faults, is still a very playable level, and while I can’t really recommend 105 Days, I respect that it was trying to do a lot with the level, both aesthetically and with the design. If you really want a blatant example of aesthetics over design, there’s The Haunted House by Ninja X. While the level has pretty impressive 2.5D-style graphics (though I think it makes Mario and enemies look really out of place), the design itself is totally flat and uninspired, and the level ends on a pretty standard Big Boo boss fight. More than any other level I’ve played in a VLDC, this really feels like what people imagine with levels that are presentation over design.

You might be wondering if there were any levels that deliberately presented itself as a counterargument to this philosophy. As it turns out, there was one entry explicitly designed to be gameplay over graphics: Zeldara101’s Reznorland 1, taken from their hack in development at the time, Zeldara’s Glitch City. The main gimmick of the level is the aesthetics are purposefully glitched up, with cutoff and repeating tiles, the usually. This isn’t really used to benefit of the design in any way, and even seems to have a graphical focus of its own, like how the Fish Skeletons sort of look like they have jet boosters on them. What really kills the level for me, however, is that it’s ultimately just a blatant edit of Vanilla Fortress, with the only real difference being that there’s no water. This was the intended design, so it’s not like the author was being lazy or anything, but I really don’t see how this can be argued as “design over graphics” when most of the design is just taken from the original game. It’s unfortunate, as there hasn’t really been an example of this kind of level otherwise, and I’d be interested to see what I’d end up thinking about it.

Judges and Scores

For the most part in these analyses, I haven’t talked about the contest judges and scores at all. I’ve been trying to avoid this from being a breakdown of judges’ opinions or rank comparisons, as I want to respect that they have their own opinions and qualifications. Still, I have to comment on what a hectic judging process this contest went through. Before that, however, I want to give a breakdown of the scoring of these contests, since that was also one of the changes of VLDC4.

For the first three contests, levels were score out of 50 points, with 10 points earned from each category: Creativity, Difficulty, Appeal, Functionality, and Fun. Personally, I like that each category has only 10 points to work with; when you start getting higher than 20 points, it’s a lot harder to argue why, say, 42/60 design is worse than 43/60 design. Of those categories, Creativity, Appeal, and Fun are all pretty self-explanatory. The first two are pretty uncontroversial, and still exist in some form in modern VLDC, though people might scoff at Appeal taking up 20% of the score. I think Fun tends to get looked down on as a “subjective” category, but I really think it’s an important criteria to a level; even a level with tight, well-executed design could feel like a slog or just too frustrating to play through.

The other two categories, Difficulty and Functionality, are much more problematic. The idea of judging a level’s difficulty on its own is pretty questionable; it’s not like there’s a specific difficulty level an entry has to meet, so people are going to choose how easy or hard they want their level to be. Unsurprisingly, this level ends up just punishing harder levels, despite there being plenty of levels that are way too easy. As for Functionality, initially this sounds like a logical category, but based on the judge comments I’ve read, nobody really seems to know what to do with it. Usually, it ends up being a free 10/10 for levels that don’t break in any way – while most categories in VLDC3 averaged at 20/30, the average for functionality ended up at 23/30. Also, it’s surprising that Design in itself wasn’t its own criteria to be judged; I guess you could argue that Difficulty, Functionality and Fun all cover that, but those first two categories end up not working well on their own.

In VLDC4, the scoring system was updated to mixed results. Difficulty was removed, and Design was added, so at least those two issues were taken care of. However, Functionality was kept, and instead Creativity was merged into Appeal; this not only brought down the total points to 40, it raised the weight of Appeal even higher to 25% of the score, ironically making aesthetics a bigger focus. Supposedly, people viewed Creativity and Appeal as the same categories (or at least the former feeding into the latter), but I didn’t really see that as the case in the judge notes, and these categories weren’t judged that much more closer together than any of the others. The Functionality problem seemed to get even worse, too; Counterfeit scored all but 12 levels (out of 81) under 9/10 for Functionality, while tatanga made some weird decisions in the category too, like marking down levels for not having Dragon Coins. I can’t really blame the high score in Functionality on the judges, though, since most levels played pretty fine to me; it’s just a really bad category.

I guess now we should talk about the judges. It’s astounding how out of the four VLDCs I’ve played, only VLDC3 has had more than two judges, and two of them (VLDCs 2 and 4) have had huge issues with getting judge results. While VLDC4 is an improvement on 2 in this regard – at least 2 people judged all the levels instead of 3 people judging a partial amount of levels – it’s still a far cry from how well VLDC3 judging went. Because FPzero and S.N.N. were too busy to judge, and the former didn’t want to force judging duties on the hack team, FPzero decided on a judge application process that users could submit to; all people had to do was judge FPzero’s VLDC1 judge level, Flight of the Yoshi. While I have no idea how many people actually applied [EDIT: according to FPzero, it was 21 users], six people ended up getting selected: three as the main judges, and another three as backup. In the end, only one main judge (Counterfeit) and one backup judge (tatanga) were able to finish judging the levels, the latter apparently rushed to do so.

The two judge documents seem like polar opposites. Counterfeit went to great detail talking through her thoughts on each level, and while I didn’t read through all her descriptions (admittedly I don’t like it when judges write comments for each category instead of just the level overall) the selection I did read were very interesting and clearly described her opinions. Unfortunately, at least in regards to the contest, they’re all undermined by her scoring which ended up slanted very high; 75% of her entries are at least 28/40, or a 7/10. Tatanga, whose scores were both much lower on average and had a wider range, ended up influencing a lot of the rankings in the contest. However, their comments were much sparser, which led to a lot of backlash when it came time for the results. Considering they were pressured to judge all the levels in one week because four of the other judges had to drop out, I can’t really blame them. It’s an unfortunate situation overall.

Conclusion

At the end of the VLDC4 Results thread, S.N.N. left some scathing parting words for the contest:



In some ways, yes, VLDC4 was a hot mess; it was completely overhauled early on, and the judging process was both too long and fraught with issues. However, I don’t think that should overshadow the levels that came out of this event. Even with the false start, 88 levels came out of the contest, and while there’s not too many levels that stand out, overall it’s a pretty solid collection. I think it’s ripe for a compilation-style hack, much more-so than VLDCs 2 or 3. It finally feels like a step forward in quality, and I’m eager to see where it goes from here.