Denrei's Blog

Building the Pub0
thedenrei posted in Uncategorized on July 31st, 2006

Because there will be no normal maps in this version of the pub, I’ve gone through the textures I want to use and pulled extra detail from the normal map. Usually this meant copying the R and G channels of the normal map and blending them in to the color map in various ways. Because these textures will be thrown away when we get normal maps working, I wasn’t concerned about spending too much time here. In order to get things together quickly, I used maya’s “convert to file texture” thing. It bakes lighting in the scene and shader information into the UV space of an object. Here is Phara with her quick new color maps.

Its not my best work, but I’m keeping in mind all this texturing will have to be redone soon. Hopefully Tony will reappear soon so we can get moving on making things use the scripting system… I’ve got a few “game design” ideas I’d like to see implemented unless they take too long to do. I spent Sunday night in the library messing with Garage Band to see if I could get some quick ambient music which would be appropriate for the setting.

Also, I’ve been a refugee for the past few days. There was a really heavy rain storm here before the weekend, and my apartment was damaged to the point that the City of Madison demanded we vacate. The red cross gave us rooms in one of the dormitories around here, and I’ve been spending as much time as I can in the UPL grinding on the pub and listening to Dave Matthew’s Band while I wait for them to allow me back into my apartment. Hopefully tomorrow.

EDIT:

Not tomorrow, but 10 DAYS from now. I guess they have to ship a new transformer from overseas…? This means I’ll have moved in to my new apartment before they allow me into my current one.

Secret Games are best made using TextureMaker0
thedenrei posted in Uncategorized on July 18th, 2006

I’ve been doing a lot of texture work lately, using exclusively photoshop and my digital camera. Its a little frustrating to use photoshop for these tileable textures… you really have to go out of your way to make the layer styles (I so recently came to love) tile.

In a moment of weakness, I googled for a low cost tileable texture maker and found TextureMaker. I downloaded the demo and was presented with an unfamiliar and off putting set of icons. I mashed a bunch of them, and something cool came out. It was like, worn shingles. And it was perfectly tiled! It saves me the “putsing to get it perfectly tiled” phase, and cost no more than a trip to your local game store. Still, you can’t get polished results using TextureMaker alone, so I doubt that I’ll ever make a texture which doesn’t pass through photoshop at some stage.

The original idea for the current milestone was to distribute the current version of Outland. It doesn’t seem like such a good idea now, because what we have is very infrastructural. So we decided to put togeather a small, alternative environment to test what we’ve got now. In order to not get sidetracked and waste time, we put a 24 hour cap on the project. So we’ve got until midnight on the 19th to create everything and polish it, then another 24 hours to make it avalible for you to download and enjoy.

Art School0
thedenrei posted in Uncategorized on July 13th, 2006

I finished my summer sculpture class today. This second project went very well (I’ve mentioned before that the first one stunk). While it might have been frustrating at the time, I do feel like I’ve learned quite a bit from this class and the other fine art classes I’ve taken so far. “Digital artists” like myself often think of fine art as a waste of time, that we are above or at least aside from it. And I still think this. But in being forced to buckle down and work in common artist terms, and be critiqued by their standards – I’ve sort of expanded my appreciation for what they are trying to do – and where they might have strengths I lack.

Everything I work with digitally is very complex. Not referencing the technology of it, or the math or huge learning curve of the tools. The nature of the art I am creating has many limitations and requirements. My job is to make pieces which are, if not “realistic”, certainly convincing, representational presentations. I cannot really call them “compositions” because I really have little guarantee on the details of how the audience member will experience them (its built as a dynamic, interactive experience). So I’m always working within the “stylistically realistic representational” realm, within an unpredictable environment. This unpredictability exists everywhere – how the character will be rendered, what graphical settings is the player set on, what level is he in, which frame of which animation, has the player seen them before… Also, there is a story which needs to remain coherant to the visuals. The model has to actually represent a character in the progression of the game (go figure). So this is what I mean when I say game art is “complex”.

In this art class I was specifically told not to make my work iconic or representational. It had to operate completely on abstract form. Again, my personal feeling is that no hunk of nonsensical plaster can be as cool as a convincing character. But in having to narrow my tool set down to the most basic visual elements, (and introduce the awkward limitations of actual physical objects) I was forced to really come up with a brand new way of thinking creatively which I had never thoroughly tapped into before.
A concrete example of what I mean. If the character I’m working on looks odd to me for some reason, my natural progression would be along the lines of “What cloths is he wearing? Should they be more futuristic? Is his body defined properly? Do his gloves make him look more like a ninja or a pilot?” when perhaps I should be simply asking things like “Is this balanced? How do the forms here and there relate? Which elements here are dominant?”. I mean, I’ll never explicitly think those things. But I think my eye will be keen to it.

To mention the game, my work has remained in the world scene. Its definately improving, but I’d really like to see it in game soon, with in game lighting and such. That would help… 0_o

Specular Maps and UVs0
thedenrei posted in Uncategorized on July 11th, 2006

Been doing a lot of texture work lately. My true focus is on improving the world scene. I’m really not at all satisfied with how things look now, and I think the current textures are contributing to that.  Other issues could be how the heightmap geometry relates to the buildings, and the level of definition in the building meshes.  I mentioned in a previous post my concerns about balancing distribution of detail among the different maps on a surface. I’ve been looking at textures from recent games I own, and reading what I can find on the topic…

A tutorial on creating a specular map for Quake 4 was helpful. They suggest putting extra highlight around the rims, thinking about how the light rolls accross the object.

When looking through the textures from Quake 4, I was suprised to see height maps included for many of the surfaces. It turns out that when the level loads, Doom just turns these into normal maps and can combine them with other already existing normal maps. Apparently they do this because it allows for an “easier workflow”. Which makes some sense to me. Doing it this way means that all the normal maps can be assumed to be 100% machine rendered. Any “by hand” normal mapping can be added in using the height maps.  Still, you’d think they might want to combine these before release in the interest of load times.
Another thing I’m not completely satisfied with is how the world geometry has it’s UVs defined. When working with game worlds, its important to keep a consistent scale on all the textures, so you don’t get patches with lots of detail and others with very little.  Its very noticable if this is not the case.  At the moment, I have a script which generates the UVs and puts them in a consistent scale.

string $arry[] = `ls -sl`;
for ($k in $arry) {
string $TargetObjName;
$TargetObjName = $k;
select -r $k ;
duplicate -rr -n AlternateObject;
makeIdentity -apply true -t 1 -r 1 -s 1 -n 0 -jointOrient;
polyAutoProjection -ibd 0 -cm 0 -l 2 -sc 0 -o 1 -p 6 -ps 0 -ch 1;
ConvertSelectionToUVs;
polyEditUV -pu 0 -pv 0 -su 0.01 -sv 0.01 ;
polyTransfer -uv 1 -ao AlternateObject $TargetObjName;
delete AlternateObject;
select -cl ;
}


I’m kind of horrible at programming/scripting. The idea here is to use an automatic projection to lay out the faces consistently, without letting it scale the peices. Because polyAutoProjection doesn’t consider the transformations on an object (in the parent or on the transform node) I create a temporary duplicate which gets parented to the world and frozen, then project on this new object, then copy the UVs back to the original object. Right now, if I move any of the verts on an object, I’ll need to rerun this script on that object. So any modifications to the UVs by hand after the script is run have to be considered disposable. I need a solution which can lay out the UVs for all objects in a consistent way like this, but I would like my by-hand modifications to somehow persist through modifications to the verts.
I also gave our music guy a call yesterday, and asked him how he felt about using some of his already existing work (which he had initially written for a different project of ours) for Outland Inc. Our music guy is one of my good friends from my hometown, Ann Arbor MI. He responded very positively.  I’m excited to look through his stuff with him, and discuss what would work where.

Layer Styles0
thedenrei posted in Uncategorized on July 5th, 2006

Having learned quite a bit from doing the planet and logo, I decided to try going back to some of my textures for the town. The biggest thing I’ve taken away is the power of layer styles (ie. when you double click on a layer and it brings up that little window with 1000 cryptic options…). I had known of them before, but thought they were mostly for doing noobish winamp skins and such.

This is based off a file I had been working on eariler, and was really not turning out. The picture on the left is without layer styling and on the right two have it turned on. The bread and butter are the “shadow” and “glow” options, and doing some fancy things with the quality contour curves (by that I mean choosing a non default one). I’ve used drop shadow to highlight the nooks and crannies. The curved line things are actually a single color, but by setting the quality contour curve to look like “/\/\” two hills, you get new and distinct lines, the color of which you can set. Also, for the glow options, I’ve set some noise, giving a bit of grain. The vertically curved lines still dont look good. Its too dense at the moment and could probably use a bit more blending into the wood below.

One “unknown” for me is how much detail to put into the color map, and how much to rely on other in-engine shading. Our engine supports normal and specularity maps at the moment. For example, here there are horizontal nooks (and crannies) which I’ve colored darkly, and highlighted with yellowishness. Should I instead leave the darkening completely to the normal map and the highlighting to the specularity map? I feel like I should leave at least a little bit of that information in the color map, from past experimentation. If you were to shine a light directly onto the surface (eliminating any direct shadows) the nooks would still be darker, even though the wood there has the same value as everywhere else. Is this what “occlusion maps” are for…? I guess I’ll just have to wait to get a more complete toolset from Chris (which is coming along very well) so I can just see how it looks and work intuitively.

I’m so good at renting movies. Since coming to school here in Madison, I tend to rent only when I’m not hanging out with anyone, so my picks are 100% my own taste. and I’m always rediculously satisfied with what I get. This time I got a whopping 4 titles: Ninja Scroll, Appleseed, Love Actually and Memoirs of a Geisha. Ninja Scroll I’d seen many years ago, so thats kind of cheating. And its really violent. My project in sculpture class turned out horribly. Been spending all my creative energy here, I guess. Whatever.

Height Map Pics0
thedenrei posted in Uncategorized on July 1st, 2006

What you are looking at is not geometry, but Maya’s (very lame) HeightFeild node. It takes in a greyscale texture and makes a fake mesh which can be used to preview the displacement for when you apply it to real geometry. Once I’m ready to make the map, I’ll break up a plane into tiles (this is 15×15 tiles, Outland only uses 9×5 for game play). Then, displace all the tiles at once (the tiles will inherit their UV space from the original plane). Because I’ll do all subdividing before hand, they will match up with one another perfectly. One problem with this is that the vertex normals between tiles won’t match up, and there will be a nasty seam. I’ll have to talk to Chris about a solution. Maybe someday, these maps will be generated at load time, then I won’t have to fuss with all this nonsense (0_o) but for now…

Logos and Height Maps0
thedenrei posted in Uncategorized on July 1st, 2006

I’ve put a good amount of time into the logo for our current project. Using this excellent tutorial by dinyctis as a starting point, I took a stab at creating what Chris had described at the end of our 2.5 hour meeting.

Situations like this are always kind of tough for me. We all came up with the name togeather, and the team has a feel in mind for what it should end up looking like. Sometimes that causes me to get in ruts, trying to guess and prod at exactly what everyone has in mind so I can deliver on it. I don’t address things creatively, I just guess at what people are thinking – which doesn’t produce the best results. I think this comes more from the respect that I have for my team, than any fear of inadeqacy in myself. Things like logos represent our group as a whole, it seems improper for me to just duck into my room and make it up. I think I’m getting better at handling this sort of thing though, I’m pretty happy with how the planet turned out anyway.

Looking through the tutorial, you’d think coming up with the planet would have been simple, but it was actually a little tough. The textures I had been using for the planets surface really weren’t working out as I had hoped they would. The texture dinyctis has is quite excellent for the task. But after a few failures, I got one that seemed to work quite well. My hope was that the planet would have the feel of the old west – like its harsh and undeveloped but at the same time it has potential for someone whos willing to sweat for it. I was even able to get different areas of the planet to show different climates. One side looks more harsh and hot, the other is more green and blue. I just did it becuase I thought it looked more interesting, but Chris suggested that “Maybe it’s a planet with an orbit that has one side facing an intense sun” which is a cool explination for it. Maybe the planet like, rotates very slowly so theres always vegetation on the back side which gets burned up as the planet rotates into the sun, then regrows when it enters the shadow again. I feel like I saw that in a movie or in a book or something. Ray Bradbury? Is that how it worked in Pitch Black? I think I had a dream with this in it…. if someone starts reading these posts and knows where this idea comes from please post.

I feel like doing this logo has also given me a strengthened sense of direction in terms of the look of the game world. Making world textures will be easier keeping this planet picture in mind. So far, things have been kind of disjointed in terms of like, realism vs stylized design. This planet seems to me to have detail of a realistic rendering, but lighting and colors of a particular feel. Which I’ll try to keep coherant with the rest of the game as it continues to develop.

The other thing I’ve been doing is working with Maya and Photoshop to generate some terrain using height maps. Its a really unforgiving process, but I’m getting better and better results. What I’ve got going now is using 2 maps. 1 is the more detailed, randomish map to get nice bumps and hills where I want them, then the second one gives the more generalized changes in height. So if I want a planes vs rocky area, that is decided in the first map. Then the second map gives broader changes in elevation. The reason this is effective is mostly because I can say 10% black is where water level is on the first map. Then everything below 10% will be under the water plane. The water plane does not receive any displacement from the first map. Then (after pointing all the vertex normals streight up) I apply the second displacment to both the water and the terrain. That way, I can keep the rivers looking just how I want them. This probably makes no sense at all as I’ve discribed it here. I’ll elaborate with pictures and such once I actually see the technique succeed.

Deadline for the next milestone is Sunday night. Things are looking good.