Developer Diary #14

[Mal]

ART

This months art discussion focusses on the subject of grading. You may have heard about film grading, when watching the Lord of the Rings extras, or perhaps just puzzled over the look of O'Brother Where Art Thou or other popular films with a noticeable colour treatments.

Grading is the process of adjusting colour, also referred to as colour correction, or colour timing. The same theory of treating colour can be applied to games. (Compare screen shots of Quake and Gears of War). With film there are a number of ways to apply your grade. The traditional methods of applying colour correction with chemicals, or coloured filters over lenses, is being replaced by digital processes. If you have the Lord of the Rings - 4 disc set, there is a great little introduction into digital film grading. Digital provides the director with fine non destructive control over all aspects of the image.

Grading is used to give a film a certain look or quality which may enhance the experience of the film, much like a painter would use colour and shade to denote mood and evoke an emotional response in the viewer. Another use is to reduce differences when filming on different days/weather conditions and combine separate filmed or CG elements into the same scene.

So how can we perform grading in games? We could use a post process run time effect, by using the frame buffer, and applying a filter, however this has computational overheads, so it's costly at run time, and may not be achievable on lower spec machines or shader models. Another solution is to apply the grade directly to each texture used in the game. This has no extra run time overhead, and provides fine control as each set of textures (landscape textures, and character textures for example) can be treated slightly differently if required.

With careful planning and organisation of source textures, it's an easy process to grade via a batch file which opens each texture applies the filter and saves it off. This is the method I've used this month to generate our grade.

Let's look at the approach to finding the grade of choice. Obviously batching all the textures takes a bit of time, so we don't want to do our testing that way we'll reserve that for the final treatment. Fortunately we can mimic the final result by using a screen shot. In Photoshop I'll experiment with different approaches to adjust the grade, until I find a workable solution to the look I want. The same process (when finalised) can then be used to batch the textures.

Here is the original Screen Grab.


Through the grade I'm looking to create a more cohesive image and create a 'look' for the game.
In Photoshop I used the Gradient map tool, This tools adds a colour gradient into the light and dark colour range in the image. So with the Gradient map you can isolate and tweak the highlights, mid tones and shadow areas. This on it's own will remove too much of the original colour, so the screen grab is duplicated first, the Gradient map is applied, then the result of the grade is blended with the original image and finally the opacity reduced.

Here are two colour strips which demonstrate the colour shift as a result of the Gradient map process described above.

Before


After


Graded Screen Grab result.


The graded image, processed on the screen grab not the actual textures. To batch the textures I have to create a Photoshop action, this stores the Gradient map, blend mode and opacity settings. The action can then be run on a selection or an entire folder of images.

Below is the result in game after the textures have been processed.


That is a screen grab with the new batched textures, not just another Photoshop graded screen grab. I wouldn't trick you like that, honest! Seeing as you've been so good and I'm feeling generous, here are a couple grabs from the game with the new grade in use.





Until next time then...

[Mike]

Design

Looking at Mal's post, I've got picture envy again. He's doing all the sexy photos of the game running; Lee put that pic up last month of the logical elements displayed in the game editor; so I could post...what...? It's getting to the point where I really am going to post a screen grab of the spreadsheet that contains all of our item data. I'm not joking either, I'll do it. I'm a desperate man.

Bah! Pictures? We don't need no stinkin' pictures! I'll dazzle you with my words instead. I...um...make good with the words.

Right, what's been occurring in GamesfactionLand (the gaming-est place on earth) this month? We've been carrying on with what I talked about last month really; building the level up and seeing which bits we need to add to get all the systems working properly; and that's been coming on nicely. You can now play the level from start to finish with all the objectives completing properly, giving you points and opening up new objectives, running cut-scene cameras to show you important events during the level and even displaying basic versions of the story texts that will pop up during the mission. It's all coming together very well and is showing which bits are working properly and which bits need some tweaking, which is just what we hoped it would do.

I just mentioned the story texts there and that could probably do with going into in a bit more detail. In a game like this, you've got two places you can tell the story, in the levels themselves and in the bits outside the levels. That might seem blinding obvious, as those are the only two areas of the game, but not all games use both of those for story-telling. Take the big game of the moment, Bioshock. That tells its entire story in the levels themselves; the stuff outside of the levels is only really the main menu where you load and save the game. In our game, the menus and other parts outside the game levels take a much higher priority. You set up the squad that you'll take into the mission there and you also get briefings on the upcoming mission - which is where the bulk of our story-telling will take place. Now, we're not writing War and Peace here, so we don't want to overdo the story elements, but we're still trying to create a game world that's interesting to be in, where the player wants to find out more about what's going on in the world around them and where they feel like their actions are part of a greater story; where they're making a difference.

I find the best way to do that is to come up with far more backstory than you need, have all the details of the world, and the characters in it, worked out on paper and then decide on the bare minimum you need to relate to the player in order to move the story on in a timely fashion. That way you work out what the important points are, telling them to the player without bogging them down in tons of tedious exposition, while still giving the feeling that there's more there to be found - that the world is larger than what's going on in the level that the player happens to be on at the time.

So, we've got the briefings, where the bulk of the framing story will be told, and we've also got messages that pop up in the levels themselves. Those will mostly be in place to tell the player about gameplay features; new objectives, things that need to be done before the current objectives can be completed, cancelled objectives and so on; as well as the occasional message who's only purpose is to advance the plot. As we're going for a purely text-based system, we have to be really careful to cut all of our in-game texts back to the bare minimum; we want people to be able to read them even while being engaged in fierce combat. In that situation, the way a character says something becomes as important a part of portraying the game world as what they are saying. Think of Halo for example; you get most of your information about the objectives and plot from Cortana. But she doesn't spend all her time talking, she only speaks when it's absolutely necessary and mostly then to tell the player what they need to do next. Most of the feel of the bigger picture is communicated in the way she speaks; the phrasing she uses, colloquialisms and, of course, in the actresses' vocal performance. We can't include that last element, as we're going to be purely text-based, but we can have a good stab at the others.

To that end, I've created a couple of characters who will be the player's ambassadors to the game world. The way they speak will add colour and the differences in their personalities show up conflicts of opinion during the time the game takes place, fleshing out the world and hopefully making it seem a more real and rounded place.

See? I don't need any pictures to add excitement to my posts, do I Doc?


"That's right, Marty"

There you have it.

[Lee]

This must be a first; we've all done our diary entries on time...

Speaking of picture envy, I know what Mike means. I'll have lots to show in the next couple of months though when we start working on some funky shaders and our effects system. I can't wait to get onto those jobs, but rightly we're focussing on getting the last little bits of gameplay out of the way first. As the only programmer, and as there are so many tasks that need to be done, I've approached each system by getting it working 'just enough'. This is good for getting things in quickly and seeing if they are going to work, but it leaves a legacy of incomplete systems. We're now trying to switch into a mode where we systematically complete systems. Not 90% finished (and the other 90% later), but completely finished. For some of the systems, that is just impossible due to the number of dependencies. For instance, currently I'm adding all the additional features to our instants (think magic spells but without the, err, magic bit) but a large part of their appeal will be their visual appearance: lots of funky particle and lighting effects and spot-on animation and timing. These systems are yet to be worked on so there is obviously only so far that I can complete the instants. Even when that is done, we need to record all of the sound effects but that's a whole other story. As you can see, with some things there really is no option but to have lots of plates spinning at once. It is a symptom of our decision to develop our own technology at the same time as making our first game. None of these issues will exist for future games thank goodness.

So, after finishing off the instants, I'll be starting on a job I've been putting off for a while. When we prototyped the gameplay, we used a very iterative approach, layering on new systems and adding new requirements. This leads to code that is less than ideally structured, but it achieves its job which is to test out the ideas. Now that we know we have the systems right, the game playing like we want, and we can fully specify every last edge-condition, it is time to bite the bullet and start the main logic afresh. I think it is always a good idea to write prototype code with the intention of binning it later. The rewrite won't take too long, and will result in significantly more maintainable and flexible code. And then, when that is done, all the game-play code will well and truly be finished and I can get on with the visual aspects.

I'm going to resist the temptation to twitter on, but one last thing; we did our first movie of the game last week, so that we could send it over to the guys writing our music. Annoyingly, it was before we totally changed the whole look as Mal described It made us think about getting some movies together for future diaries and I'm sure it is something we'll do. It'll be good to visually show the progress over the closing 6 months of this project.

[AndrewG]

Wow, great looking screenshots 

I love colour correction!!

[Mal]

Cheers Andrew, It's coming together nicely and will only get better.