AuthorTopic: What's your process for creating game environments (aka "tilesets")?  (Read 2989 times)

Offline Sunjammer

  • 0001
  • *
  • Posts: 19
  • Karma: +0/-0
    • View Profile
There's a lot of wisdom and analysis out there about making sprites but creating pixel environments isn't touched upon nearly as often.

Most articles that exist on the topic have a focus on how to create individual tiles which I've come to feel is misleading in that it places the focus on making an environment one tile at a time, which doesn't seem to work if you're trying to achieve any kind of artistic vision for the mood and atmosphere of the environment using color and light.

For those of you who make tilesets for games, what's your process like? Do you block in colors and values first as though you're establishing an environment like a traditional/digital painter would?

The Pros to this approach as far as I can tell are that it'd be easier to get the colors you want down and establish the value relationships in the scene and get the proper mood and lighting. The Cons would be that it'd be hard to take these blocky shapes to a final level because unlike painting, blocks of solid color don't communicate the final scene as well - so much relies on the rendering.

(Tales of Phantasia example)


Another possible approach would be to draw out all the architecture of the scene exactly as it will appear in line art and then color and render it afterwards. The Pros here might be that it'd easier to imagine the final appearance of the environment and nail down the architectural details, while the Cons might be that it could be harder to visualize how to put color into it and it might focus too much on minutiae rather than nailing down the feel of the level.



Starting with an individual completed tile does have the benefit of giving you something fully rendered to work with which can help you visualize how the rest should look, since pixel art appears vastly superior with the full ramp of colors rendered in, versus when it's just solid blocked-in value.

Regardless of your approach, how do you manage color ramps throughout the environment? If you did Process 1 would you create ramps for each major hue segment of the environment, or just add colors as needed when rendering the textures in? How much do you think in terms of traditional painting where you make the piece complementary or monochromatic or analogous? And how much of that translates to devising ramps?

The entire concept of 'color ramps' is very useful for sprites but almost seems like an artificial and awkward principle for 'painting' game environments. Of course ramps exist in all game visuals but maybe it's not as useful to conceptualize it that way when creating tilesets than thinking about it like painting a scene.

How important is it to have concept art before pixeling a game environment?

I don't know if there's been much discussion of these things before but I haven't been able to find it  anywhere. Maybe this topic is just mysterious to me!
« Last Edit: January 25, 2015, 07:08:53 am by Sunjammer »

Offline Decroded

  • 0100
  • ***
  • Posts: 1285
  • Karma: +3/-0
  • Oh hai
    • View Profile
Most articles that exist on the topic have a focus on how to create individual tiles which I've come to feel is misleading in that it places the focus on making an environment one tile at a time, which doesn't seem to work if you're trying to achieve any kind of artistic vision for the mood and atmosphere of the environment using color and light.
I agree.

For those of you who make tilesets for games, what's your process like? Do you block in colors and values first as though you're establishing an environment like a traditional/digital painter would?
Yes I like to paint instead of line-art.

...more comments:
How important is it to have concept art before pixeling a game environment?
The whole creative process is a weak area for me, something I really need to put more work into.
Years ago a great friend and artist said I had an issue with "visualization" as he described it.
Perhaps some more creative people can have some good suggestions but I've just been thinking for my next game environment I might try a process like this:
1) look around for a theme, perhaps something with some kind of background story or whatever
2) save a bunch of references of anything along the same theme to get ideas flowing
3) write down a bunch of ideas and start sketching things in the theme (probably directly in pixel art), mess with mixing bits and pieces of various references and try to create things along the same theme. at this point just having no real thoughts about what is going to go where in the environment
4) consider some level design and block in structural elements
5) drop in previous sketches and start working the designs into the scene

I also really need to start engaging in the Creativity thread we have going here, time willing.
Something you might want to jump into yourself.

Regardless of your approach, how do you manage color ramps throughout the environment? If you did Process 1 would you create ramps for each major hue segment of the environment, or just add colors as needed when rendering the textures in? How much do you think in terms of traditional painting where you make the piece complementary or monochromatic or analogous? And how much of that translates to devising ramps?

The entire concept of 'color ramps' is very useful for sprites but almost seems like an artificial and awkward principle for 'painting' game environments. Of course ramps exist in all game visuals but maybe it's not as useful to conceptualize it that way when creating tilesets than thinking about it like painting a scene.

My advice is forget completely about ramps and palettes.
Instead just add colours as your workflow needs them.
I personally start with very few colours of just light and dark to block in the main forms, then start adding more as the materials require and as I add more detail with smaller brushes.
At this point I'm usually messing with colour temperature and saturation a bit according to materials and lighting.
Usually I end up with a washed out look and I use the software's tools to increase the overall contrast, mess with saturation etc.
Once the scene is becoming detailed and has good colour balance, you can start thinking about doing more advanced colour techniques like creating hue variations and bounces through your ramps.

There are many artists here of course who I'm sure bypass lots of the steps I need to take.
I'm sure some people work off predefined palettes too which I need to have more of a go at (palette suggestions anyone?).
I can only imagine once you've got alot of experience you can visualize fancy ramps and just start painting.
« Last Edit: January 26, 2015, 09:35:10 am by Decroded »

Offline Ryumaru

  • Moderator
  • 0100
  • *
  • Posts: 1683
  • Karma: +0/-0
  • to be animated soonly
    • ChrisPariano
    • View Profile
These things need not be exclusive. Organic things such as trees benefit from being more "painted" architecture more "drawn". If a style is supposed to be more or less organic you can sway the percentage of these two major aspects to taste.

Formulas are necessary in efficient game asset creation, but I don't think it should be across the board. Each artist should find their own paths so that as much artistry as possible can remain intact.

In terms of colors, having at least 1-2 shadow colors shared across most assets is a good way to keep things unified, but be sure to change how often it is used based on the surface type and local color. This is especially useful in very low color, especially when some tiles need to share colors because reasons.

I agree that the focus on individual tiles can be misleading. I've been known to sketch something out completely organically, only to then mostly redo the setup in a tile friendly environment. This isn't the most efficient, but it can lead to solutions that would not have come up otherwise. It's common to be working up around 9 tiles at once.
 

Offline hawken

  • 0010
  • *
  • Posts: 214
  • Karma: +0/-0
  • 8bit fantasy
    • hawkun
    • View Profile
    • my portfolio
I find that drawing out "features" (unique objects, trees, nude greek statues etc.) and "building blocks" (bricks, stairs, ladders, grass, stones etc.) really help. You flesh the level out in a tile-map editor with your building blocks (much like how 3d level design is done) then consider your features as "prefabs" that you can drop in to pretty the place up. It helps to have a game plan of sorts (but I also think free-forming a level does lead to some creative surprises).

As was mentioned here already, it's beneficial to have a style or two you can apply to unify the tiles.

I'm currently working in 8x8 tiles and it is both maddening (looking for a tile) and full of happy accidents.
twitter: https://twitter.com/hawkun
Pirate Pop Plus for 3DS, WiiU & Steam!

Offline astraldata

  • 0010
  • *
  • Posts: 391
  • Karma: +1/-0
    • View Profile
    • MUGEN ZERO
I too have been interested in this subject for a long time. Here's my take on it:


==============
   Metatile Chunks
==============
Many game artists / level designers back in the day of the NES (and SNES / Genesis especially) used tile "chunks" to build their levels. These "chunks" were simply groupings of individual (smaller) tiles arranged in a grid blob of varying numbers of tiles wide and tall -- generally they were always square or rectangular. These blobs would build large things like trees or statues or columns and arcs, etc. so that they could be quickly placed in the levels. Each individual tile in these chunks could have flags set as to their solidity and other properties (i.e. one or two parts, for example, of the larger tile chunk is two parts [the higher end and the lower end] of a two-tile-wide slope). The Sonic games are an extreme example of this tile system at its most functional.

That being said, if you were an artist back in that day, you might have designed your assets in large squares or rectangles to help you establish a mood. This applies to everything from RPG tiles to platformer games on the SNES, for example, because any sort of scenery could be drawn in those large square/rectangle chunks as long as they fit in the SNES's sprite/tile 'atlas' on the cartridge and were able to be stored in memory as the game was running.

I think these tile "chunks" might have also been called "metatiles" but I'm pretty sure the terminology varied from company to company since most of that stuff was proprietary back then.

======================
   Multiple 'Canvas' Tiles
======================
The main idea of the "tile-chunk" approach is to draw out your world in many smaller "canvases" (larger than a few tiles in width and/or height usually) on the same tilesheet when you need to have lots of large and varied scenery (such as in platform games) -- especially if your scenery must NOT look repetitious and you have memory to spare. An extreme example of this is the backgrounds in oldschool arcade beat-em-ups like the Simpson's Arcade Game or any number of other Konami titles in arcade machine cabinets.

======================
   Modularized Metatiles
======================
If you don't have memory to spare or if you need lots of tiling variations (i.e. in the case where you have many jointed pieces and angles with lots of shape/style variations that need to remain modularized), you would establish a base texture for them, and develop all tile variations from that (which ultimately will bring us back to all those single RPG grass-tile tutorials found all over the net.)

The modularization technique, contrary to popular belief, does NOT always have to be applied to a single tile -- it can also be applied to a larger tile-chunk / meta-tile in order to make it appear tileable. In actuality, different games use different methods (single-tile vs. meta-tile variations) for modular tile variations.

Star Ocean: Blue Sphere on the Gameboy Color is a great example of using modular tiles and tile chunks together to represent amazing-looking worlds while keeping the modular tiles extremely varied and tileable. There's a thread here on Pixelation that details some of the tilesets in the game. You'd be amazed at how tiny the tiles are that are used in some of the meta-tiles, and yet it's because of their granularity that they are able to be so flexible -- they can be used to build much larger and more varied assets with much less memory.

===================
   Tileset Design
===================
How you design your tileset really boils down to how you ultimately choose to derive variation in your tiles. An individual tile (or metatile) can only ever be either A) modular and flexible or B) unique but static and there really isn't any inbetween for a single tile in a tileset.

It's important to keep both approaches in mind. In most cases -- at least in the early days, where the SNES and Genesis were involved -- performance was most often prioritized over graphics artistry since the 'fancy' hardware was still pretty new to programmers. Eventually, graphics became a higher priority once the underlying technology and performance issues were better understood. Once tile count stopped being as much of an issue anymore, metatiles were utilized a lot more often -- however, these were almost always still used alongside oldschool tiling / modularization techniques to stretch the number of tiles that could go toward forming larger (metatile) tile-chunks.

===================
   The Oldskool Approach
===================
Even with today's virtually unlimited memory, if you want to go with a grid-based graphical layout for your levels, you'll generally still use both types of tiles, just with modular tiles being more geared to terrain or pipes or anything with variable size/shape/configurations throughout your levels.

In design cases like that of the tree in your screenshot, tile chunks were specially created for the bulk of the tree -- and, actually, the lighting was done by hand by the level designer (and not the tileset artist) using some very much rudimentary tiles provided by the artist. I'll give an example:

Here you have the original image, now aligned to a 16x16 grid (most likely what the level designer used when designing this level in the game):



Below (in yellow) are the unique tiles and (in blue) are any flipped and/or tiles repeated more than once in the image in various places that were used to achieve the lighting effect here (with their respective yellow tiles highlighted in yellow elsewhere):



This effect is mostly a transparent overlay of tiles that outline the tree tiles and also attempts a 'cutout' effect for the cast shadow effect to simulate a halo of light and a shadow. I can guarantee no painterly layout was used for this process. Just a slick level designer using some tiles haphazardly made by the tile artist to achieve this effect. It's very likely the two worked closely together and the level designer said "hey, it might be cool to give this tree some lighting" and the artist said "yep, I'll throw some tiles your way to see how it looks" and thus you got the image above.

==============
   A case study
==============
Put the images above on a 16x16 grid (switch to an 8x8 grid for the smaller tiles) and you'll see the tiles yourself.

After studying those images (perhaps as frames in Graphics Gale you flip back and forth between) you'll see just how haphazardly the tiles were drawn -- as stated before, much more focus was given to drawing them modularized than designing a specific 'mood' for the level.

Even the lighting on the window was simply reused overlay lighting tiles, from down beneath/around the tree's root area. The key to this design is the level designer was simply creative in how he used the existing tiles to achieve the effect he was going for. You'll find that many games in the past (and even today!) still rely on creative level designers to utilize the provided modularized art assets in creative and interesting (and sometimes in even unintended!) ways!

==============
   The Takeaway
==============
My advice is to stop worrying so much about what your assets look like and instead focus on what you can do with them. The main question you need to answer is this --> Do the assets I have serve my intended purpose for this level? If not, follow up with this --> What assets do I still need to accomplish my vision of this level?

In the end, if you don't have a strong vision for a level, no amount of drawing fancy-looking tilesets will save it.
I'm offering free pixel-art mentorship for promising pixel artists. For details, click here.

     http://mugenzero.userboard.net/