Stay away from full-saturation colors, pure whites, pure blacks. They light oddly due to the mathematics of RGB color space. Choose bright reds only when something is hugely important, and bright yellows only when you want to draw the eye. Red is a danger color that humans have a hard time ignoring, and yellow peaks the most cones, biologically speaking, in the human eye, and is therefore bright and attention-grabbing. Blues recede to the background, so don’t make something very important blue or grey-blue.

As you complete your color blocking, think about the elements of design. Although there’s no real surface quality yet, you could say that some parts of the model are more detailed than others due to the size or shape of the colored areas. If there are lots of small shapes, you could interpret that as detail. Are there areas of lower contrast and detail where the eye can “rest”, allowing the viewer to focus on the more important aspects? If not, you’ve gone overboard. It’s important to have a balance. Is the detail focused in the areas that the eye is drawn to? Don’t spend a lot of time putting detail everywhere. If it’s not on an area of your model where viewers don’t look very much, you can simply suggest detail, or leave it out altogether. You have to be really critical. Just because you like a detail doesn’t mean it’s important. Fancy buckles on boots? Probably not that important. Simplify or skip them altogether, unless the boots are a big plot point. Facial tattoo? Probably pretty important, since the viewer will look at it for a long time and it probably says a lot about your character.

When the color blocking is finalized, your piece should read just about as well as it will when texturing is finished.

As you move on to the painterly detail, don’t use a brush that’s too small, and don’t use the brush at full opacity. A medium-small round brush at 40-70% opacity will give you a good starting spot. Overlap strokes to get more opacity, instead of switching colors, or varying opacity. This’ll be frustrating at first. Keep with it. Only blend as a very last step! You don’t want things too soft, overall, or you’ll lose the painterly style you’re looking for. Don’t switch up the size of your brush a lot, and stay away from small brushes altogether. If you need to use one, do it as a very last step, with a very light touch. Check out conceptart.org for tutorials and examples.

Focus first on the dark areas. This shouldn’t take you too long. Lots of detail in the dark areas won’t get you a lot of mileage — people aren’t looking there. Spend most of your time on the mid-tones. The highlights should take you, on average, five percent of the time you spent on the mid-tones, mostly because a little highlighting goes a long way. You don’t want too much of it, and it should be in a very few spots. Try not to stray too far from your original color blocking. A color that looks quite dark can function well enough as a highlight, if the background is darker. Keep your color block as a reference layer, and turn it on every now and again to verify that you’re sticking with your plan.

Use the brushstrokes to complement the model. Stroke along edges and around curves. If you’ve got a cylinder, you’d want to stroke around the circumference, or diagonally in a spiral, but not from top to bottom. The directions of the brushstrokes are going to help your viewer interpret the surface. Take a look at old-fashioned engravings or etchings for an idea of directionality (but not style, obviously). Now, these are suggestions, not rules, but follow them for a bit to see how that works out for you. You may have cases where you want to do the opposite, but bear in mind that stroking along the flattest direction of a curved surface will make the surface appear flatter, whereas stroking along the tightest curve will emphasize the shape.

Reference is important, too. If you know what style you like, you’ll have a better time replicating it. Find an image that you like, and zoom waaaaay in until you just see the brushstrokes, kind of abstract. Then see how you can replicate that. I would bet the edges are softer than you’d think, that the contrast is much lower than you would think, and that there is actually a lot less detail than you suspected.

Don’t be afraid to put colored filters on top of your image (I imagine you’re using Photoshop). Check out the “photographic” filters. You can use them to unify a texture if you feel you’ve pushed the colors too far apart during the detailing phase. You can reduce or enhance contrast as well — and if you’re aren’t familiar with them, learn how to use masks. That way, you can stack up a bunch of color, hue, or levels filters and mask them to specific areas of your texture as you balance the colors during the final step.

And all throughout this process, keep checking your work in-game! Don’t get too focused on what you see in your paint package. Remember it’s just a component of the final image. Don’t lose sight of the end result.

You’ve also got to understand the shader and renderer you’re using. Does it use specular highlights? Normal maps? Projected shadows? These are going to impact your choices. If you’ve got specular highlights in the shader, for goodness sake, don’t paint them into your texture! Let the shader do its work. Likewise with shadows. If you’ve got strong projected shadows, don’t muddy up the final result by painting strong directional shadows into your texture! You can end up with a really confusing result, with some shadows that react to light and some that don’t. Eep. If you’re making normal maps, be sure you take them into consideration when you’re painting your texture. They may impact your stroke direction and edge highlighting.

Hopefully that helps you out! Good luck.

This is a transcript from an e-mail that was posted as an image in the TF2 Emporium modding thread, at Facepunch.com.

I’m writing this blog post to serve as a summary of the story that’s happened between me and this guy during the past two months.

Despite me having done everything that he had asked at the beginning, giving him two “chances” after he had dug himself deep into a hole, even after I made an apology that went much further than it should have been given the consequences… he still refused to let this blow over.

I’d like to repeat it for yet another time: I don’t condone harassment or attacks. I only want our story to be out there — what me and Rebb had to go through — you can draw your own conclusions from it. But let it be known: I’m not a doormat, and I have no intentions to be one.

Actions have consequences.

Here’s a quick summary of the whole story.

On January 16th 2012, Rebbacus and I got fed up of how Jonas “Mecha the Slag” Kaerlev, the owner of SLAG Gaming, was treating us in regards to our input to his network. Among many things, I think what took the cake was his complete lack of concern for his players getting what they bought by donating. So we decided to leave, and we also asked for our content to be pulled off his servers.

Mecha the Slag promised that he would do it, if that’s “what we wanted”. That promise would then be repeated another two times. To this day, it still hasn’t been honored.

But then, he said it would be after I explained my actions in a public statement on his forums, and Facepunch. Keep that in mind, now — Mecha the Slag is the one who made this go public. *HE* started this by not removing *OUR* content when we simply asked for it.

I first posted my public statement on his forums. The thing is… nearly everyone was in support of me and Rebbacus. Posts were deleted, then the thread was closed and trashed, and my account banned (and eventually permabanned).

So I posted it on Facepunch.com… and well, it went pretty much the same way. But it got pretty funny, because Mecha wasn’t a moderator there. Read the thread yourselves if you want a laugh.

After Mecha the Slag pretended he didn’t know who Rebbacus was (despite him selling a donation-only skin that she had made on his server store!), tried to get an inflammatory reaction by putting a video (which used my face) he knew I wasn’t comfortable with back up, and many other things you can see in the Facepunch thread… until he “went to sleep”…

Eventually, I got IP-banned from his forums. Previously, only my account was banned, which means I could log out to still look at posts. By using a proxy, I spotted this thread, in which Mecha said that he would look into an idea that would… well… involve re-using our content. Which he had just promised, for a third time, that he would wipe off his network. Right after IP-banning me. I’m sure you can agree that this is pretty shady.

I knew he was never going to do it, so I sent a DMCA takedown request. The DMCA is a law that makes stuff like this much more manageable. However, Mecha the Slag lied under the law, under the penalty of perjury, to claim intellectual property / copyright over my work and the work of Rebbacus. You can read more about it here, or in the 22mn long YouTube video further down this post. This is why the Parkour Fortress content we made still isn’t off his servers — but at this point, we don’t care about that point anymore.

Fast forward to February 3rd. I receive about ten DMCA e-mails from Tumblr.com… like this one:

So Mecha the Slag is claiming that I’m violating his copyright with Tumblr posts… like this one, or that one. This is just text. This is a support message from someone. Yet Jonas “Mecha the Slag” Kaerlev claims that it violates his copyright. Another low-blow… not surprising.

Quite interestingly, the day before, Mecha the Slag got a part about blackmailing removed from his user article over at Mario Fan Games Galaxy. Blackmailing, yes. Best part: he admitted to it.

Following this, I made a 22mn long exposé on his actions and the whole story up to February 6th. Mecha the Slag filed a privacy complaint on YouTube to get it, like my other videos, taken down (which is very hypocritical, as you’ll see in a few seconds). It’s outdated on several points, so there isn’t really any interest in keeping it up any longer anyway.

He then responded to it. His video was however deleted by YouTube for violation of their terms of service. It’s not like anything valuable was lost; one argument that had been proven wrong already, the rest being personal attacks which went as far as saying my parents were “going through a divorce”, and that it “explained my anger problems”. He later re-uploaded this video on Metacafe, which took the video down for the same reasons as YouTube.

That’s Jonas’s style: no arguments? Attack the other person’s family. Classy.

I replied to it in this video.

Now, here’s where the fun begins, and the summary of the situation ends.

On February 17th, I received a Google Alert for a new result on my name.

Kind of creepy. An empty page titled “Maxime Lebled portfolio”, with my address, my phone numbers… the WHOIS registrant information was sadly protected, and contacting the host (SiteGround) yielded no replies. The domain was bought on February 7th… the day Mecha the Slag uploaded his video reply. Despite this, I didn’t think he was behind the website at first. After all, he had his back against the wall, an absolutely enormous, to say the least, history of low-blows and debunked lies…

So yeah. At first, I thought “404″, the owner of two TF2 servers named “UNF Gaming”, had done it. Why? One, because the domain name company was in the same country as him, Canada, and he had, the day before, posted this announcement. 404 had a grudge against me and Rebbacus, ever since we refused to collaborate with him after he got himself instantly permabanned from Facepunch in a single post.

The day after, MaxOfS2D.com turned into a full-fledged page with photoshopped screenshots and forum posts as evidence that I’m a crazy person or something along those lines.

But then, I eventually discovered proof that it was Mecha's doing.

Here’s the video on this exact subject, and a MP4 download of it. The following text explains things a little bit better, so it’s what I would recommend looking at :)

Evidence #1: EXIF metadata

EXIF metadata is information stored in pictures. Stuff like exposure value, camera model, etc. but some software can leave other footprints, like e-mail address, GPS data…

In this case, here’s the thing that tipped me off in the first place, and incited me to look further. Let’s look at the YouTube screenshot of me “going mental in YouTube”:

1) the metadata says this image was saved at 10.44 p.m.
2) the gmail notification shows this comment was made at 10.13 p.m.
3) the timestamp on the screenshot itself says 25 minutes ago. that leaves us with 10.38 p.m., add a few minutes for photoshopping a few sentences out.

Therefore, it is someone in the same timezone as me.

And here’s the juicy bits.

In all of the JPG images at MaxOfS2D.com, there is a string, a footprint. A reference to some deskjet printer. Not all EXIF metadata viewers will display the string, I used this Firefox addon. You can go check by yourself; the footprint isn’t going away unless all images on both websites are cleaned of their metadata, and re-uploaded.

If you don’t feel like checking yourself, visual proof of this footprint existing is available in the video at the very top of this post, around the 8mn30 mark.

%20Deskjet%201000%20J110%20series8BIM%04%3B%01%B2%10%01%12p

All the JPG images at MechaTheSlag.net and MaxOfS2D.com have this footprint.

(furthermore, they’re also all saved using 80% quality, using Photoshop CS5 on Windows)

This is an undeniable link, especially considering how scrambled the footprint is. Even in the worst of cases — assuming Mecha the Slag isn’t the author — it would still mean he has made all the images hosted on there, and that he knows the author… unlike what he says here.

Evidence #2: video encoding settings

There are two video files, one on MechaTheSlag.net, and the other only on MaxOfS2D.com. The first is Mecha’s reply to my 22mn long exposé, since it got taken down by YouTube and Metacafe, and the second is an excerpt of a video by 69starmix96 about me.

• http://mechatheslag.net/public/maxofs2d.mp4 [MEDIAINFO]
• http://www.maxofs2d.com/maxofstd.mp4 [MEDIAINFO]

Both files were encoded using the same settings. 59.94 fps, Main 3.2 profile, 1600kbps VBR video, 128kbps audio… things that are too similar — maybe a coincidence, but if it’s just one, it’s starting to be a lot of them.

Evidence #3: whereabouts of Mecha’s IP

Mecha the Slag’s IP address accessed a Tumblr post that was screenshotted and uploaded onto MaxOfS2D.com, only four minutes before it was saved.

Where did I get the IPs from?

1) Facepunch.com’s logs, thanks to Garry and another moderator, Daimao.

2) The SLAG Gaming forums! I linked to this screenshot further above — this was when I still had access to the Snowman event account, since it had admin rights.

3) And finally, Tumblr. Not directly Tumblr, however, but logs from the servers on which files embedded on my Tumblr are hosted. In this case, audio.maxofs2d.net and img.maxofs2d.net. Here are the two entries:

87.57.14.106 - - [01/Mar/2012:17:57:56 -0600] "GET /jukebox/xspf/xspf_player.swf?playlist_url=http://audio.maxofs2d.net/jukebox/xspf/playlist.xspf&volume_btn=off&bg_color=EEEEEE&button_color=6088AA
&txt_color=424242&blend=1&album=off&bottom_controls=off&autoload=1
&autoplay=0&shuffle=1&start_track=1 HTTP/1.1" 304 - "http://tumblr.maxofs2d.net/" "Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/535.11 (KHTML, like Gecko) Chrome/17.0.963.56 Safari/535.11"

87.57.14.106 - - [01/Mar/2012:17:57:57 -0600] "GET /website/homeportal_background_wildlife_white.jpg HTTP/1.1" 304 - "http://tumblr.maxofs2d.net/" "Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/535.11 (KHTML, like Gecko) Chrome/17.0.963.56 Safari/535.11"

Keep in mind that my server is in the US. 17:57:57 (5:58 p.m) shifted to my timezone, gives 58 minutes past midnight. That’s only 3mn31 before the screenshot of my Tumblr was saved…

87.57.14.106 and 87.57.14.55, despite being dynamic IPs…

… both match to an ISP upnode that is only 8 kilometers away from where Mecha the Slag lives.

Note that, while MechaTheSlag.net was registered in a city that is at the extreme west of Denmark, Esbjerg (the Facepunch registration IP matches to there!), he does indeed live as a student in Aalborg nowadays. (and before you say I’m violating his privacy by revealing those two cities, keep in mind that 1) they’re pretty big, 2) I’m not posting his street address, and 3) they’re here to support the fact of dynamic IPs matching)

Conclusion

So there you go. Solid links that prove, at the very least — even if you were to be giving him the benefit of the doubt — a connection with whoever made MaxOfS2D.com. But when you take into account Mecha the Slag’s past of low-blows and lying…

Yeah, this is childish at best. But not completely unexpected from you, Mecha.

Every time you replied to any of my points, you’d construct a straw man out of one or two of my sentences, then destroy it with a fallacy. You can’t confront the evidence head on (I still haven’t seen you justify the Tumblr DMCAs), because you know you’re in the wrong. And what was your reply to this? Bringing the video down, saying that the EXIF metadata is based off “magical printer moondust that connects everything together”, and that I “didn’t give the source of the IPs”.

That’s pathetic. I am, however, very flattered that you would spend money on a domain name and WHOIS privacy just in an attempt to spite me. Anyone with half a brain cell wouldn’t take your website seriously anyway — especially with a counter that claims that I visited over 25k times.

I do also hope that you realize none of this would have happened if you had kept your promise to remove our content in the first place; a promise that you made not only once, not only twice, but three times. You’ve been digging your own hole ever since, and while the metaphors on the subject are not in short supply, there’s one in particular I find most appropriate: on January 16th, you were doing it with a shovel, and today, it’s with an industrial-grade grill.

You know, I haven’t had the chance to actually attack you, the person behind this, yet — after all, all I’ve done until this very sentence was merely exposing facts — so I’ll make up for that now.

You present all the symptoms of anti-social personality disorder. “A pervasive pattern of disregard for and violation of the rights of others occurring since age 15″, “callous unconcern for the feelings of others”, “lack of remorse, as indicated by being indifferent to or rationalizing having hurt, mistreated, or stolen from another“, and last but not least, “deception, as indicated by repeatedly lying, use of aliases, or conning others for personal profit or pleasure”.

(That’s spot-on, don’t you think? Thanks, Wikipedia!)

Please get a grip on the real world, and realize that your actions have consequences.

Sincerely,
~ that dude you made a slanderous website on

P.S.: another one for the road?

• Mecha filing a DMCA takedown request over evidence hosted on this server
• Mecha wordfiltering my nickname on his servers
• My reply to a chatlog from Mecha the Slag
• Mecha permabanning one of his players for sending me evidence
• A direct quote from Mecha the Slag himself: “Respect is for dicks”
• And last but not least: Mecha made a Facepunch.com account just to rate my posts as dumb

This could probably work with any other music players, if they have a plugin that allows you to output the currently playing song to a XML file. Winamp has one! Though it’s not freeware:

Anyway, this is simpler than it seems: I use three title objects in XSplit, each one configured to read the title, artist, and album XML tags in the file created by the media player plugin. Of course, you can leave out any of these if you wish.

The spacing between them has to be adjusted down to the pixel, it’s pretty tricky. The background is a 72 pixels tall slight gradient made in Photoshop.

And there you go!

Valve pretty much broke Source 2009 enough so the SDK tools only works with the Team Fortress 2 binaries. On other GameConfigs, you’ll get a black screen in HLMV & Faceposer (along with not being able to browse models), and Hammer will not display viewports correctly. But here’s a fix that should work (until they accidentally break that too). You’ll have to make yourself a new SourceMod in which GameInfo.txt will need the following keyvalues:

FileSystem
{
    SteamAppId 440
    ToolsAppId 211

    SearchPaths
      {
        Game |gameinfo_path|.
        Game ep2
        Game episodic
        Game hl2
        Game C:GAMECONTENT
      }
}

C:GAMECONTENT (or whatever folder name you want) will need to have whichever game’s models & materials you need, since switching the SteamAppID to 440 results in TF2 content being mounted first.

THIS LINE WILL NEED TO BE AT THE BOTTOM OF THE STACK.

On a more positive note, this brings two (maybe more) notable improvements to HLMV; there’s a more detailed counter, and having normal mapping off doesn’t glitch phong & rimlight.

First off, a status update on EarthBound Fortress 2.

Remember when I said that the first chapter would be there by late July…? Yeah. Me too. But hey! Even though I’m increasingly starting to dislike this one, compared to the much more interesting stuff there will be in later episodes, there are several more art assets being crunched down by people far more talented than I am. How awesome, you ask? Well, I’m shamefully gonna admit the number one reason that made me go back to working solely on EBF2 is that I really want to use what I’ve got in store for later.

For instance, Rebb recreated one of her old mods from scratch (hint: it’s related to her awesome drawing right above), and tell you what, it’s beyond amazing. I’m hoping to maybe, eventually, make something with him outside of EBF2, but he’s definitely coming there in the second chapter. Hideous (creator of the Western Wear hat for the Engineer in TF2) is creating sweet map props which are also insanely optimized.

On the topic of optimization, since the Onett map is starting to progressively destroy the Source engine with its size, I’ve reduced the displacements more or less everywhere (resulting in the VMF losing more than a third of its size), and added occluders in the buildings. It helped a bit. I should give a try at putting fade distances into some of the vegetation, which most likely accounts for half of the map’s polygon count, which is nearing a million faces.

Anyway, let’s try, once again, to quantify progress. (the following stats are sponsored by the dozens of “will you ever finish EBF2″ mails & formspring questions!)

• Nearly all of the voice files have lipsync data
• Onett is nearing a sufficient level of detail
• Ness has 3m04s of animations (I’d estimate that’s a good half)
• Paula has 1m17s of animations (only lacking around ten more seconds)
• 90 seconds are already completely done (i.e. rendered, edited, with reconstructed sound)

I’m gonna follow stick to the post’s sub-title and not promise anything: just that, if it continues at this pace, maybe you could expect it in three weeks. The thing I’m worrying about the most is the battle (yes, I finally made a machinima with an ACTUAL CONFLICT in it, at last) — I’ve made a few animations for the characters already (like Ness drawing his bat, dodges, etc.), but I don’t quite know yet how I’ll be handling the camera and the odometers.

Concerning the camera: if we want something interesting in the battle, it’s gotta keep moving. Which means I need camera sequences more or less everywhere. Except most of the battle animations are built with modularity (i.e. they’re gonna be re-used several times) in mind. Which means I’d have to build generic camera animations too. I could probably stick to Catmull with a long enough trigger time, but where’s the fun in that? Haha.

Concerning the odometers: oh, yes, they are done and everything. Except I have no idea of how to make them roll like EarthBound does. I’m not sure if it’s actually even a concept worth translating into the movie, since it’s just something used for gameplay. Does it really make sense after all (I can think of only one way, bleeding out, and honestly that does NOT fit the MOTHER series)? But yeah. Oh, by the way, this is how they look like.

The battle “interface” is probably going to be the only thing which keeps the pixelated style. And since it does, it needed more detail… I think changing the background colours was one of the best ideas I’ve had in a while, really.

Last thing: there’s something you’ll probably notice when watching chapter one: the quality of animations is going to be very variable. A picture is worth a thousand words, so an animated GIF must be worth quite a lot more…

Experience truly is the best of teachers.

I believe that sums up what I had to tell about EBF2.

Let’s take a look at the videos I’ve uploaded lately.

125,000 views under a month. It turns out the video was featured in the community news section of TeamFortress.com, which promptly resulted in me having a stroke, and joining the club of the cool cats who got their stuff featured by Valve!

This is probably the closest I’ve been to a “meet the” in terms of quality so far, to be fair. The Scout flashing the cheesy smile would probably be the funniest, too. However, I’m not really happy with the post-processing on this one — I should have picked a different sky. It’s like flat, bright, saturated blue everywhere in the background, yuck. Miss Pauling could have had more interesting animations, but as the guys over at KritzKast said, “standing there and staring is pretty much all she needed to convey”. Finally, I forgot to add motion blur on one shot. Too bad!

My dad saw this video and started questioning my sexuality. It was pretty funny… kind of.

The idea struck when I uploaded a TF2 replay using the same music. I made a run cycle from scratch using CATMotion (the presets are pretty awful and don’t seem to really fit my rigs much), and then I did the animation in an incredibly dirty way which involved, as far as the Heavy was concerned, three absolute layers, seven local and one world adjustment layers. Which is why I won’t release its CLP source. But anyway, I was forced to use FK instead of locally-constrained IK targets, and it shows… especially when the Medic comes into contact with the Heavy. Other than that, pretty satisfied. It could have been more interesting, though…

What I learned while making this one:

• CATMotion is really awesome!

• The errors induced by the lack of precision in the SMD format become noticeable after an offset of more than 2,000 units (the Medic stands roughly 2,300 units away from his local space origin in the video). If you look closely at the him in the “opening arms” shot, you’ll see the motion is sligtly jittery. Same for the camera, when it’s around there.

• Continuation of the above: I’m pretty sure StudioMDL attempts to optimize position data when compiling. You usually can’t notice it, but that’s an effect which kicks in at very small offsets. you can see it two times in EBF2: Psychokinesis during the closeup, 11 seconds in (you should notice two small hitches). I believe Paula is only around 100 units away from the origin there. I’m not completely sure, but I believe it messes up with “baked” position noise (i.e. the one existing in the exported animation, as opposed to one you could layer above the sequence in Faceposer or the QC) if it’s too subtle (the threshold probably increases as the offset does).

• I don’t know how I didn’t think of this yet, but I can simply disable the pelvis using a weightlist in the QC, and in Faceposer, the model won’t move off the view.

• You know how the camera interpolation usually results in stray frames you need to delete between shots? It threw my timing off, so I raised the speed of the music a little bit. For some reason, with the pitch raised by a quarter of a half-tone, it actually sounded very cool. I’ll probably use that trick more.

Animated in a handful of hours, this was an attempt at creating facial animation more detailed than it usually is in my videos. I wanted to do that when I saw that the intro I quickly threw together for the Heavy & Medic video above was actually not that bad in terms of faces.

What I learned while making this one:

• Blending several expression poses actually works well to create detailed facial animation.

• I really need to fork off the tf_hwm library for each class. It was made with the Scout in mind, and many phonemes don’t fit other classes (which you can see in the video which allowed me to truly try out the system and its fallbacks on non-HWM models). That, and if I make one phoneme library for each class, I’ll be able to use the tongues on the Heavy and the Medic (ever noticed that they’re the only two to have modelled tongues…? WHAT WAS VALVE PLANNING WITH THAT)

• Corollary of the above: it results in a lot of clutter. A total of around 18 facial poses were used here, which makes me think I should go and fork off all of them in a separate expression file. tf_hwm_face used to only contain distinct parts of the face (eyebrow / lip poses, blink, etc.)

• Although it doesn’t show up in Faceposer, the ray-traced eye rendering may result in eyes going past their limits, i.e. they’ll roll inside the head if $EyeballRadius is too high. That was fixable on the Heavy, but I know the female scout is at the absolute minimum she can go on that one…

And there you are, at the bottom of my wall of text. I should make these more often.

— How to use CLP files

Load up a CAT rig (preferably one that follows my preset). The secret lies beneath the feet of the character, in that arrow-shaped object, the CATParent.

From there, you can go to the Motion panel, then to the Clip Manager rollout. That folder button allows you to browse to a CLP file easily. You shouldn’t change anything from the default import settings — just leave “Scale Data” ticked.

What’s really neat about this is that you are importing the original keyframes and assigned controllers! However, some may not behave as intended due to their procedural nature; this includes noise generators. Therefore, they are left out of the sources.

Another problem is that clips are, as far as I’ve seen, interpreted in the CATParent’s local space, which royally messes up some of the source files, as IK targets are still stored in world space. Those will be fixed later.

My rigs also don’t have CAT objects for the additional bones (for props, clothes and such), so they won’t be there. If you’re gonna import a CLP on the FemScout, you’ll have to deal with having to re-animate her dress!

So yes, it indeed isn’t a flawless system. But I believe it is more practical than going through standard motion capture data formats! Here’s a GIF with the same pose applied on six different rigs; the only thing I did there was to shift the pelvis down in most cases to allow the feet to connect with their IK targets. And it’s very easy to do (even more so with an adjustment layer).

We had already found out a special way to customize HWM models, but it turns out we can customize faces relatively easily by simply applying a flex which is permanently applied (unless I’m mistaken, it’s merely set as the default reference VTA frame). I used it to fix a tiny bitty oddity on Paula’s upper lip — but since the flexes concerning that part of the face had already taken care of it, using them pretty much applies the fix twice, which looks a bit odd but much less than how it was before.

The point being, this is pretty darn cool — and I hope I’ll be able to edit the HL2 citizen faces this way! I actually have all their enhanced source files by Bloocobalt lying around, but it’s pretty messy because absolutely everything is bodygroups and the UVs are a complete waste of space. At least, we’ll get something better than always the same reskinned mesh, heh?

Onett, which is in the making, is quite skewed in its layout (the main reason being the towns in the game have this weird 30 degrees perspective), but there’s one thing I’m rather proud of; it fits the world. That is, the whole world. Before we look at how I transcribed it into Source, let’s look at the actual Earthbound maps; and first, the promotional world map that was, if I recall correctly, printed in Famitsu around 1993:

It’s pretty to see that the whole planet is based off our real world. That said, there are a few inaccuracies that appear if we stitch all of the actual in-game sprite maps together:

Some argue that a few areas, such as Saturn Valley, are overlapped by others, but it’s all up to personal interpretation— which is a good thing, considering it gives me more freedom, haha! Anyway, Onett is a sea-side town on the west coast, alright? Let’s make it so. (please note that Onett’s map is a work-in-progress and is compiled with what you could call a “draft” quality right now)

Twoson is, logically speaking, at the south. Right. There’s a long road between it and Onett which isn’t mapped out (yet?). The EBF2 representation of Twoson is surrounded by mountains, so they’re in its 2D skybox (at the north) and to the south of Onett, in its 3D skybox.

Well, it’s probably waaaay out of scale, but once again, that could depend on whether what we see from Onett is the actual mountain (it’s probably not; I’m guessing we’d need to be farther away from the ocean to get mountains that high). We’re gonna guesstimate the actual distance from Onett to Twoson to be around 20 kilometers, then!

Then there’s the looooong road to Happy Happy Village (i.e. Peaceful Rest Valley) from Twoson. It’s got the same skybox than Twoson, except edited to make it look like sunset. It sort of makes sense to me to call a certain location a “valley” when it’s in the middle of mountains! The actual length of that road would make sense to be around 6/7 kilometers (I think it used to be mentioned in the script before all the crappy lines got cut out). Around one kilometer is actually mapped out across both road maps.

The problem here is where the sun sets & rises. It’s pretty much inconsistent— but good lighting matters to me much more than accuracy in that regard.

So yes, I think you could say I’m relatively consistent with Earthbound’s world, and I’m trying to be consistent with all the details I need to create myself— there were no such things as proper heights or skyboxes in the SNES game! The world didn’t really feel huge back in 1994, even though it supposedly covers the whole west to east coast of America (Fourside is pretty much New York), the south of Great Britain (Winters), the east of Spain (Summers) and Northern Africa (Deep Darkness, Tenda Village). That’s something I want to change, not only because it’s pretty cool, but also due to the scenaristical possibilities. What if the Dusty Dunes desert was 100, 200 kilometers long, instead of that petty cakewalk?

Meeting the requirements of such a scale is rather hard, because Source engine maps can really be only one kilometer by one kilometer (+ 8km on all sides with a 3D skybox. TF_Summers has quite a lot of sea that way). But it’s nothing impossible! And of course, it’s an interesting challenge…

Maybe one day, I’ll get someone to convert those maps into actual TF2 CP/PL/KOTH maps.

• 27 voice files out of ~50 are lipsynced. That’s around 40% of the audio length (~05mn)
• 21 sequences out of ~90 are animated. That’s not really a lot.
• Two maps are complete, another one needs a few props, and there’s another map to start.

On the upside, a ton of sequences are going to be motion captured, which is likely to speed up a big portion of the episode. I’ve also had the luck to welcome a new guy on the team, who’s hard at work making props for Onett’s houses right now! Speaking of which… I’m rather satisfied with my not-so-respectful adaptation. Then again, considering most of Earthbound is using a weird perspective, that’d be tough to recreate exactly as-is in Source!

Let’s have a run-down of what happened during the past month:

• Over 7,000 subscribers
• Over 2,000,000 global views on my channel
• 200,000 views on all of the EBF2-related videos (fillers included, tech demos excluded)

Oh, and it’s been two weeks, and Valve still hasn’t fixed the phong exponent bug. I’d be surprised if it’s still here by the time I have all of the assets ready, but you never know. It was bypassed in the two recent videos by using a fixed exponent (which is why Paula’s gloves do not look like how they really are).

So two days ago, I suddenly thought about batch compiling all of my animation .QCs. It’s very simple and works well, so I thought I’d share it here. All you’ve got to do is create a .bat file, and put as much as you’d like of these lines:

"%sourcesdk%binorangeboxbinstudiomdl.exe" C:blahfoldercompile.qc

What’s pretty nice is that even if one of them fails to compile, it won’t stop the others — and you can add :: to comment them out of the process.

Definitely a huge time saver.