• Topic Archived
You're browsing the GameFAQs Message Boards as a guest. Sign Up for free (or Log In if you already have an account) to be able to post messages, change how messages are displayed, and view media in posts.
  1. Boards
  2. Sword of Moonlight: King's Field Making Tool
  3. 2016 Sword of Moonlight status report
arcnet 6 years ago#1
2016 is around the corner, so I am writing another year in review for SOM here, since it's the only corporate website I know of with a Sword of Moonlight forum.

Everything about SOM that is relevant today is at www.swordofmoonlight.net and affiliated wiki en.swordofmoonlight.org. And my still barren Patreon is at www.patreon.com/swordofmoonlight.


Past years this has been a multi-post job. Partly because of the limited character count. I will likely spread this out over days if that happens.


This year has been the biggest ever for Sword of Moonlight. It's maybe hard to tell because it isn't quite fit for the general public yet. And the last few months the forum has received very little in the way of hits. I don't know what the hits indicate myself, probably mostly web-crawlers that do not give a damn about SOM, but you never know. In any event, if honest to goodness people are following its development, they are too shy to make their presence known to me.


I will paint with broad strokes because the details of the year are mostly lost to me...
Remember, this is Gamefaqs:
http://www.gamefaqs.com/boards/661226-pokemon-black-version-2/64668593
arcnet 6 years ago#2
When 2014 ended I was two or three months into a project to make an all new tool for SOM out of SOM_MAIN, which up to then had been a very simple four button jumping off point for starting a session with SOM.

That project continued until the end of March 2015. So it was a marathon effort. I seem to do about one of these per year, and as it happens I am two months into one, which will probably tail off around March 2016. It is a new initiative to focus on the 3D modeling side of SOM media development. I will speak of it again at the end of this series of posts.


The project that began in 2014 is a script editor. I am sure I wrote about it in last years year end/in review. It took so much time largely because it had to be a fully formed tool right out of the box, and it's a big complicated tool, comparable only to the 3D level designer, SOM_MAP.


After March as my memory serves I worked on things related to controls, and clipping. By clipping I mean how the player character is prevented from moving places in the game world, and how they do not fall through the floor for example, which is just a way of saying preventing from moving below the floor!

I extended the clipper's level of sophistication greatly, and identified several serious bugs along the way. Clipping also has to do with climbing up, onto and through things as well. As well as jumping onto and off of things. Now SOM plays more like Tomb Raider. You can go just about anywhere you can imagine in a way that feels natural, without going into too great detail.

As for controls. Jumping was finally formalized, and doing so necessitated adding a squat-walking mode, that is somewhat reminiscent of Echo Night. To jump the Action button must be let go, along with the movement controls, at the same time. This initiates a jump. When crouching you do not move appreciably, so it's safe to jump from a crouch (which is not a crouch per se as just bending your knees to prepare to jump) without worrying about getting a running start and missing your jump. Beside that running jumps feel very reliable, because if you are going to freeze on your jump, nothing is easier than letting go of the controls. It sounds complicated but it comes very naturally. What's more complicated is getting used to transitioning to a walk after running so you do not jump when you stop running. Or you can just stop moving and then let go of the Action button, either way does the trick.

The squat-walking play comes in when you are crouched and exit from the crouch without standing and without jumping, then instead you walk away from the crouch squatting. You can stay squatted this way until you stand up or run. Standing up is like doing a standing jump. Once inside a crawlspace you cannot stand up, and so will squat no matter what. You can run and jump in the crawlspace, but are not allowed to run at full speed, which now requires holding down all three buttons when not in a crawlspace to keep running that is (a game with a concept of stamina would want to penalize for this kind of sustained sprinting.)

Furthermore there is an auto-squatting functionality when running. If there is something ahead the player character will begin squatting in anticipation in order to pass beneath the obstacle. Eventually it will be possible to go places that require getting even closer to the ground.
Remember, this is Gamefaqs:
http://www.gamefaqs.com/boards/661226-pokemon-black-version-2/64668593
arcnet 6 years ago#3
After working on clipping/controls (which involved vastly more points of interest than I described just now) began what started as a brief project for relaxation, before snowballing into a major undertaking.


This leg of the journey began by addressing the shadows that appear beneath NPCs (incl. monsters) which originally were like flat discs. The discs created visual glitches on uneven terrain, like if they touch anything that isn't a completely flat floor at the same level, or if they float through the air or stick out over a terrace, basically anything became a problem for these kinds of shadows.

So, I set out to give the characteristic shadows volume, so they blend naturally into the environment. I'd decided the absolute correct way was to continue using image based shadows. SOM uses only a single image, that is a soft round disc. It's symmetrical in all directions, so it doesn't even turn when the NPCs turn. We can expand upon this some day if we want, to make the shadows turn...

But the shadows need a sense of volume, or height (up and down through the ground) and they need to disappear when they are in thin air. So I set about doing the necessary preparations for this extension to SOM. But I soon realized that the backdrops were a related problem to the shadows, and I wanted to work on them first. I started by trying to optimize the use of the "z-buffer" by SOM. But I ran into problems with the transparent parts of the 3D scene. So like the job before that focused on the clipper, this job meant digging into SOM's renderer. Which is tasked with "drawing" the things that the clipper is supposed to keep you from passing through!


It so happened that the backdrops (classically sky domes) were even more related to these shadows than I knew. Because they could both benefit from, and required, what is called a deferred-rendering technique. So in effect this task became sky/shadow upgrade, plus any basic optimizations to do with the renderer, which as I recall was chiefly limited to lighting concerns.


The trouble with the backdrops is they are like a snow-globe that slices through the 3D world as it appears in the distance. This effect is unwelcome, so long ago one of the first extensions I worked on was having the backdrops blend smoothing into the background. But the extension was not perfect. It represented an improvement, but came with its own glitches that made it unlikely that the extension could be turned on by default, out of the box. [The glitches were reminiscent of the glitches when NPCs fade in/out, and so they were visually consistent and charming, but glitches nonetheless.] But what the new sky/shadow upgrade meant was perfecting this blending, so that it COULD be on by default, and so it became obviously very tantalizing, and the focus of a lot of work from thereon out (the operation was a complete success,)


But it was during this point in the year (which I can't place months wise, but probably mid-year) that a number of unexpected developments were occurring. An artist had half-completed the first form of a project called Moratheia. And so I was looking at it on my workstation of the time (which died this year after a 7 or 8 year run.) but could not play it at higher than standard definition resolution...

So I began looking at ways to improve the performance of my workstation. I started by disabling multisampling anti-aliasing, which I'd grown accustomed to. This had the single largest noticeable impact. However I was unhappy with the image without anti-aliasing. So as a side project I began exploring ways to improve the picture without resorting to AA. I developed numerous techniques, that are possible because I advocate using dithering, which is uncommon in games today, but it produces a superior image, just inferior screenshots and videos, since the images do not thumbnail well.
Remember, this is Gamefaqs:
http://www.gamefaqs.com/boards/661226-pokemon-black-version-2/64668593
arcnet 6 years ago#4
Right off the bat I reasoned that a simple change would get results comparable to the base-line MSAA I was using. And it did. I developed a couple more techniques to be combined together, which made me want to go even further... solving cases that even MSAA could not.

I should add that the techniques I developed all come at absolutely no-cost in computational terms.

Long story short, pushed to the brink, one day I discovered/invented a technique that I don't believe any video game has ever used, and that I wouldn't be surprised if no one knows about, because I reason if anyone did it would be very popular... and I was able to refine the technique largely through trial-and-error until it now produces absolutely stunning aliasing free pure-3D images...

The crux of the technique has to do with anti-aliasing over very long runs of pixels... like when a line of pixels is nearly straight, but a little off... so you have a row or column of pixels that is very long, until it breaks and has to go one pixel up or down or left or right. That knot is very evident. So the idea is how to make them go away? Because if that can be done the picture immediately looks like a much higher resolution display. That's pretty much the nut I cracked. And I won't go on about the details here, but they're all over the SOM website. In the main blog on the front page etc.


I have tried to give this technique away, and no one has expressed real interest in it. I believe people are basically idiots to be blunt. I don't really have time to evangelize it, I have enough of a time evangelizing SOM to zero traction, so this little exciting discovery is just a part of SOM for right now. It has perfectly straight lines and looks like television, better than anything out there by leaps and bounds... in fact I was white as a ghost for more than a few months after this development, but I am beginning to get used to it.


It is especially impressive combined with Moratheia's artwork, since it's all original and somewhat natural, more reminiscent of Ico or King's Field IV than SOM's stock artwork. It produces images that are more realistic than any video game I have ever witnessed.


It was also during this period that I guess spending so much time scrutinizing the new non-aliased graphics I began to really notice distortions called by 3D projection. This is something all games have, and I was aware of it in SOM for a while, but I hadn't realized how severe SOM's was. It turned out SOM was using a very high "field of view" and so I set about lowering its zoom angle to something with less distortion that would look more like an accurate image. This small change has made a big difference. It brings the eye view much closer to the action, and you feel like you are really there, and not stuck in the PC's stomach like Krang from TMNT!

But also low zoom angle amplifies everything. Little subtle things become very obvious. And so throughout the last quarter of the year here and there I've been making tweaks, until I think just yesterday I uploaded the final patch for a new high-fidelity analog control system. And that is more or less the end of the year.


Except in the last two months I was planning to work on controls again, and possibly animations, since there is a small problem with SOM's animations running at 30fps while the game goes at 60fps (but this could just be because of the sample rates in the files.) For controls I have in mind the final system for using the attack buttons more like hands, and drawing the weapon out only when required (which is only seldom really, especially if a game is more story than killing focused.) Which is really required though to settle an ambiguity, between climbing onto things or picking up items or talking to NPCs. There needs to be separate buttons for that basically, and they can't swing weapons. The system in mind is pretty cool as you can draw with either hand, so it automatically covers left handed players.
Remember, this is Gamefaqs:
http://www.gamefaqs.com/boards/661226-pokemon-black-version-2/64668593
arcnet 6 years ago#5
1 character left that time...


But I could not work on that! Instead something came up, and SO I felt like I had to work on 3D modeling support instead. And the form this has taken is a very major project that you can read about on the website blog and forums. It has been ongoing for two months now and will probably run throughout the first quarter of next year, dominating it.

The reason being is that "pure-3D" technique places certain requirements on 3D models, that many of SOM's stock artwork does not satisfy. The map tiles especially require attention, whereas the other models are largely already in agreement with the new regime.


The artist who made/makes Moratheia is supposed to be renovating the map tiles, but they are taking a long time. At one point I tried to do this myself, and I discovered how inadequate Microsoft's X file format was for the job, which From Software had chosen as SOM's exchange format. It was not a bad choice, but in 2015 almost no software has anything to do with X files, and even though Microsoft provided a software library on their website, that is good, no one uses it. X has become a bastard format. So realizing this I decided something had to be done immediately. Not just because there had to be an other way, but also because SOM is becoming so graphically impressive, that it is high time to start taking 3D modeling very seriously. It had languished as "not a priority" for years, but suddenly it's coming into its own as a distinct priority.


So now I've almost completely put SOM down to work on model solutions. I won't write about this here because it could double the length of this thread if I did. I'll just say that SOM is switching from X to DAE files, and that DAE will not so much be its exchange format as just its new format. The proprietary formats that the game player uses will still likely exist, but they will be generated from the DAE files, which will be considered either temporary cache files, or pre-generated files for standalone games stored on read-only media.
Remember, this is Gamefaqs:
http://www.gamefaqs.com/boards/661226-pokemon-black-version-2/64668593
arcnet 6 years ago#6
Update: One last thing to add before the end of the year.

I think as of some unexpected recent developments, and patches, and noticing a major recent addition bug, as of like Xmas day, the basic controls feel like they've finally achieved perfection.

They move perfectly through the world, and I cannot think of anything to do to further innovate in terms of the basics. Which is actually incredibly complicated, it's more or less a kind of robot emulating every aspect of human locomotion, probably somewhere between 50 and 100 variables.

It's good news because it means it's possible to consider formalizing the entire system of movement and accompanying walking effects.
Remember, this is Gamefaqs:
http://www.gamefaqs.com/boards/661226-pokemon-black-version-2/64668593
  1. Boards
  2. Sword of Moonlight: King's Field Making Tool
  3. 2016 Sword of Moonlight status report
  • Topic Archived

GameFAQs Q&A