Thomas, 2009-03-27, Coping with Optimizing

[Thomas]

We have been pretty bad at updating this work log, mostly because we have had so darn much to do. However we want to mend that and start writing more frequently from now on!

To start things off I want to discuss something that I have done quite a bit this week: Optimizing. Now this is usually a quite boring thing to do and just means changing something, writing down some number, changing something else, and so on until night falls. I usally end up with list such as this:

Code: (Select All)

Date:    080312
Desc:     30 small lights (raduis 2 - 4) and one large (radius 5) is rendered to a simple scene.

Results:
Test        Features                    Framerate
1        Deferred skybox                    ~96.5
2        Basic skybox after light render.        ~103.8
3        #2 + attenuation texture            ~85.0
4        #3 + no clamp                    ~94.2
5        #3 + no clamp, min(..) instead            ~94.2
6        #5 + skybox no longer additive blended        ~96.0

Usually I learn something in the process too... The above examples are different test with rendering the sky-box and light attenuation, trying to make it as fast as possible.

Now this may not sound terribly fun and to be honest it aint. However, I got a dirty trick up my sleeve, something that can keep me going writing out tests until my fingers bleed. I simply fantasize that I am mad scientist doing mad experiments.

"Thomas, you are one super nerd" you might think. And it is probably true. However, I have the utmost respect for these strange science folk who can dedicate their whole lives to the most wierd stuff. Like studying one single species of fungus or trying to figure out the body's metabolism by weighing and taking note of everything that goes in and out of oneself (yeah a guy did that). And when considering those efforts what I am doing does not seem that bad any longer.

And I can continue my "change stuff, test, write notes, repeat"-cycle with a smile on my face

[ShiftX]

Okay, optimizing might be an important thing. But what about the basic programming, (like when you pick up some object the enemy will be released behind the door or what is the players speed etc ...) How interesting it is ? How much time it takes ?

[Thomas]

(03-27-2009 05:10 PM)ShiftX Wrote:  Okay, optimizing might be an important thing. But what about the basic programming, (like when you pick up some object the enemy will be released behind the door or what is the players speed etc ...) How interesting it is ? How much time it takes ?

If there is interest I can write some other posts on how I deal with various sorts of programming. This was just a quick rant to give the work log a new start

[WindexGlow]

Just curious - couldn't this process be automated in the cases dealing only with number-changes? I understand there are many numbers to change, but I wouldn't think you guys just type in some random number - must be some sort of increment or equation you guys use.


And the part about mad scientist - I sometimes do the same

[Naked? No]

Trial and error my friend, trial and error.

[WindexGlow]

I know - but it seems like it could push them in the correct direction faster.

[Thomas]

Quote:Just curious - couldn't this process be automated in the cases dealing only with number-changes?

It is not a matter of changing numbers, but often algorithms or larger changes in the code. Otherwise I would be just watching TV while the comp did all the work

[WindexGlow]

Ah, thanks for clearing it up.

Perhaps a picture or two showing off some scenes with the new graphic effects..?

[Klayman]

While the game is far away from release could you make some screenshots how the game looks now and release them as bonus material in finished game? Are you going to add developers commentary (like in HL2:ep1/2, TF2, L4D...)? I would like to hear/read your design decisions.

I am doing this in my games. (except of dev. commentary)

[Daemon]

(03-29-2009 02:36 PM)Klayman Wrote:  Are you going to add developers commentary (like in HL2:ep1/2, TF2, L4D...)? I would like to hear/read your design decisions.

I am doing this in my games. (except of dev. commentary)

If you're looking for a good example of seamless [and interesting] dev commentary the best I can think of is 'The Chronicles of Riddick: Escape from Butcher Bay'.