What are you working on? (Game Making Thread)

[BizarreMonkey]

Yeah I don't make commercial projects ever, lel.

I've gotten a huge amount of work done, and my team members have also done some good stuff, I'll unveil it all soon on the game's thread.

THAT being said, 6/6 is unlikely, there's a lot I have planned and I don't wanna force myself into rushing this thing, so it'll be released on 7/7, I don't see it taking longer than that.

 

[coyotecraft]

I'll be doing some story outlining in the car. Which, for what it's worth, might as well be doing nothing at all. Story of my life.

 

[BizarreMonkey]

Not true! When I used to have to travel an hour to school then an hour back by train I used to draw so much and get so many concepts and ideas down, including but not limited to bossifhgt mechanics in PFC, characters profiles, comics and storyboards. All of these helped in ways!

As for what I got done, it's not complete yet, but its what I'm working on.
http://www.youtube.com/watch?v=JfoC2FjhhdE
She at first seems slow, but then she turns out to e super fast. Despite not having external spells or mechanics, she's already very frightening in combat.

 

[Xilef]

This is unrelated to any existing projects, but I've been learning how to program the GBA (Gameboy Advance).

http://www.youtube.com/watch?v=FSyfgSycAHwhttp://www.youtube.com/watch?v=05_Ul33pipE
http://www.youtube.com/watch?v=ks-LugT1XDY

It's a really lovely piece of hardware for C programming. There are some really bizarre quirks though, the big one I've faced is that with direct rendering the video memory is addressed as a 16bit value, but each pixel is an 8bit palette index. Writing to the video memory's hi or lo byte will erase the other byte, so my image loader has a pixel-feeder that reads pixels from the image and puts them into video memory 2 bytes at a time.

My plan is to implement as many of these tutorials as possible; http://lodev.org/cgtutor/

 

[Fayte]

I watched all the vids and had legit no idea what I was looking at.

 

[Xilef]

The first two videos are the GBA in mode 4, where a palette of 256 colours is used for the background and the VRAM is split into two 8bit pages.

In the first demo, I am decode a PCX image into video memory and its palette into the palette memory of the GBA. Pressing Left/Right modifies the palette to be negative, dark, greyscale, no-green, no-blue, no-red, yellow'd.

In the second demo, I decode a PCX image of the Quake2 PC colour palette, this is basically a 256 colour palette but with a reference image that arranges it to fake 64 brightness levels of the 256 colours and have a table of colour blending.
Pressing left/right will cycle between loading a PCX image into VRAM and loading two PCX images into memory, blending their palette indexes using the Quake2 blend table, and then writing the new indexes into VRAM, creating a "blend" between two images.

The third demo is in mode 3 graphics, where no palettes are used and the VRAM is one single 16bit page, meaning everything is rendered directly to the screen - so you will see the pixels slowly change. It is an intense mathematical calculation for displaying "fractal patterns" from two sets.
The GBA isn't even close to being powerful enough to calculate these kind of patterns, so I lowered the quality of the render massively - which hasn't helped as much as you'd think it would.


The goal is for me to get very comfortable with utilising the GBA hardware, figuring out what it is good for and what it isn't good for and learning about all the hardware registers. The first demo proves it is very good at modifying colour palettes, the second demo proves that it isn't very good for blending large images, even with the blend colour pre-calculated, and the 3rd demo proves it's absolutely awful for complex calculations and the limiting factor is definitely the processing power.

EDIT: New demos
http://www.youtube.com/watch?v=UWlsDjprPWA http://www.youtube.com/watch?v=BDMe8IiZ4sY

 

[BizarreMonkey]

Back from down south and progress on PFC has been good since my return, done with all of act one sans the End of Act Cinematic.
Here! It preview:
http://www.youtube.com/watch?v=Uk9WPfNaxwg

 

[Xilef]

More GBA demos.

http://www.youtube.com/watch?v=UpxuvB37lzw http://www.youtube.com/watch?v=irZi1drTzFA

My first attempt at making this ray-caster was hilariously bad; slow, buggy, jerky and nasty crashes.
These are at half width resolution as full-resolution is much slower. Lots of room to optimise, like using 8.8 fixed point units in some areas rather than 16.16 and using more look-up-tables.

 

[Nathaniel3W]

Recently I might have said that Xilef was a little crazy for diving that deep into this stuff. Writing one's own raycaster and writing code just to load a picture onto the screen is a bit farther into the weeds from where I like to work, but hey, to each his own. Working with C on that very limited hardware will certainly teach one to optimize.

Reminds me of an article I read about Sergey Aleynikov. He was that computer whiz from Goldman Sachs who unfortunately was convicted of stealing trade secrets. The FBI had no idea what it was investigating. The lead agent on the case was interrogating him, saying, "You uploaded the stuff to your subversion server. That's the sketchiest thing I've ever heard of." Anyway, one of the things that made Aleynikov so good at what he did was the fact that he learned programming on outdated Soviet-era computers. And they were shared computers, so your code had to run really, really fast, or else the computer lab would kill your calculation and let the next guy have his turn.

As for what I've been working on... I didn't like my old world map. It was too big, took too long to load, and was too rambly... I guess that's not a word. It was the first open area of the game and it didn't give the player any direction. It also wasn't as varied as I would have liked. I've been learning a few things and I thought I would give the world map another try.

It still needs lighting, trees, and some fixes to the hexagon overlay and smoothing of some of those small features. But I think it's going to be a much better map.

 

[Xilef]

Recently I might have said that Xilef was a little crazy for diving that deep into this stuff. Writing one's own raycaster and writing code just to load a picture onto the screen is a bit farther into the weeds from where I like to work, but hey, to each his own. Working with C on that very limited hardware will certainly teach one to optimize.

It's all a learning exercise, it would be madness to write even a homebrew GBA game at the level I've coded these demos.
I'm still staying shy from writing in assembler.

The most crazy thing that I am doing is fixed-point arithmetic, which this next demo (and my optimisations to the previous demos) now use at all sorts of ranges. Previous demos have been in 16.16 fixed point, which take up the full 32bit bus of the GBA when operating in high-speed ARM 32kb memory. Does cause some visual artefacts from optimisations losing accuracy.

http://www.youtube.com/watch?v=MVmjRVrXDQQ

GBA isn't suitable for ray-casters, the CPU clock is too low. Doom-style 3D should be faster - would be interesting to re-port Wolfenstein 3D to the GBA using a 1D span raster like on the SNES port of Wolf3D; should be faster and higher-resolution.

 

[Nathaniel3W]

I remember playing Final Fantasy Tactics. I tried giving a ninja the equip guns skill. And I tried giving an orator the dual wield skill. Neither combination let me go into battle John-Woo style. That disappointment was one of the guiding factors in my decision to include both pistols and long guns in Himeko Sutori.

 

[Xilef]

You've really embraced the purple-outline style. Looks great, makes me miss some of my old prototype ideas that specifically featured purple outlined graphics.

I'm curious about your character system, how the character sprites and colours are prepared. Specifically, have you ever considered changing the art style of the characters? They certainly pop out against the background. I have a feeling that Half Kaiser works best on bright, colourful backgrounds and now that you've really developed an awesome style of the world, will you go back and revise the character style to marry them both together? Is there a technical challenge behind that?

My thinking was basically "this might look cool with hand-drawn pencil characters" - going towards Dragon's Crown style. Perhaps a palette change in the characters would suffice, maybe a paper-texture effect as well applied to them to create noise.

 

[Nathaniel3W]

Thanks Xilef. Yeah, I tried to move away from the black outline. Now it's kind of a purplish outline during the day, turns redder at sunrise and sunset, and it's darker at night.

Many, many times I've considered changing either the sprites or the backgrounds to get them to match better. There isn't really a technical challenge. It's more a problem with me being the only developer and artist working on this, and being completely out of budget. Honestly though, if I were to change anything at this point it would be the 3D background. The sprites were by far the harder, more expensive, and more time-consuming asset to add.

The game will be completely moddable though, so maybe if early sales look promising, I could hire someone to help me with that paper-and-pencil look and players could choose between original pixel characters or the new pencil characters. A while back I even did some experimenting with a postprocess effect that would give me kind of a Valkyria Chronicles look:

Edit: Looking back on that now, I guess it could have used some really strong antialiasing to get rid of those jagged edges. I'll add that to my list of things to try on my next project.

 

[BizarreMonkey]

Now the collision script works, and in good time, I now have ten days to do LOTS OF WORK!

This is it, this is what all of my training has been for, this is why I have bought snacks galore, there is an endgame, and I intend to reach it.

Game MIGHT be delayed to 7/17/17, mostly so I have some time to market it, and get what else needs be done, done, but we'll see!

End of Act 1 Cinematic is done sans some voice acting. The OST sans one track my composer is currently remastering, is also done!


My artist, while currently not in possession of a connection to send files, will be back in action on the 28th/29th, and she's finished the rest of the HUD stuff, given Tiramisa a sad pose (ears down) and is in the process of giving Svoli her hood up bustset, which will basically be an overlay/edit on what currently exists.

So until I've gotten a game build ready, don't expect to hear from me again, it is time I cease my social interactions online and get busy.

 

[Nathaniel3W]

I've learned a lot about color grading over the last few months. Now we're able to change color lookup tables based on time of day, giving us a lot more control over how everything looks at all times.

That shows a fast "preview" quality lighting build, resulting in blobby shadows and some artifacts. But the colors are pretty much correct.

Edit: Actually, I'm pretty sure the player character is not picking up the world lighting settings. Her white shirt should be pink or purple here. I need to figure out what exactly is going on there. :huh:

 

[Xilef]

I'm back with another GBA demo. This time; it's a ray-caster with sprite rendering.
http://www.youtube.com/watch?v=nG2-p3QCkUg

 

[Nathaniel3W]

Cool. Very Wolfenstein. Have I seen those textures somewhere before? Also, didn't the original Gameboy have a kind of an FPS? Yeah. Just Googled it. Faceball 2000. Pretty crazy they were able to get that working at all.

 

[BizarreMonkey]

Got a shit ton of work done on PFC lately, I plan to have it release-ready by the 14th, the remaining three days I'll market it like a mofo, then release it all proper on the 17th to the public.

http://www.youtube.com/watch?v=roZPS7m6U60
A couple small things to fix, also need to add a gumball to get HP back, but yeah--

I'm more or less done with acts 0, 1, 2 and most of 3, Gotta do the Sty and Naxon, that's it, though.

Then I'm up to act 4, which is getting a lot of stuff added, such as a fight versus Esperia as both Kara and Svoli, a bunch of optional sandboxxy content, and finally the journey up to the London tower to defeat naxon himself, once and for all, then I'm up to Act 5, which once I refactor the massive battle with Esperia, it's all new content from there, then lastly is Act IV. Which is Pep's mind, fighting XEPP and lastly the ending.

seems like a lot but if I can get done each day the amount of work I did just yesterday I'll be done before the 12th.

 

[ZenVirZan]

reopeneing my project for the first time in forever
i want to make a thermoptic camo ability

dicked around with shaders for a bit but couldnt achieve what i want
i think i may need to do it manually

Spoiler: end goal made in photoshop

each character's segments (head, body, arms, gun) are just drawn straight to the screen right now. i think i may need to create a texture for each character which the parts are drawn on to before theyre pasted on screen. i dont know how expensive this would be though.

if theres only going to be like 6 characters at once, would i be able to get away with making entire screen framebuffers for drawing characters to save doing some math, or should i really find the most optimal way first time? im not sure how different sized framebuffers play with each other

EDIT: seriously considering restarting the project from scratch, its all written so messily that its probably the reason i stopped working on it in the first place

 

[ZenVirZan]

if i have a 2d tilemap and want to write my own raycasting solution, how would i do it?

the system i have atm is i iterate along the line in small segments, and for each segment i check the tiles near it in a 3x3 around the start point of the segment, and run a line-through-box test on it to find if it intersects

my implementation may just be a bit jank though because going up and right works fine, but down and left randomly misses and i cant tell why