@AlexandreVaillancourt Fallout 4 had amazing graphics tho

Oh, well they have time and money to put on that now, and there is competition, so they have to...

(put money on making good graphics)

I like building Fallout 4 settlements. Building and managing are my favorite gaming topics. Plus customizing the main character. So much fun!

I liked the settlements system too. Wish the base game had more stuff to build tho...

@ArthurGibraltar Yeah, that's a very interesting aspect of games nowadays. I knew someone who played Karaoke Revolution only to unlock stuff and dress her avatar! Ok she also played because she liked it, but it was much more fun with the customization.

hahaha, nice!

Any time I play a game with character customization, I spend 15-30 minutes on a face I will never see under my helmet :)

@TheMattbat999 Yeah! At least dragon age has the feature of enabling the helmet only when you're in the wild, it otherwise hides it!

@AlexandreVaillancourt hmmm... cool. I gotta go. See yall!

Good night!

I feel like I'm in a quantum superposition of making games and not making games

I have no idea what to work on

I know that feeling.

I'm stuck in a loop of "this is definitely the project I'm going to be able to focus on for a long time and maybe even finish at some point" and "ooh, this would be a cool idea for a new project!"

I feel like I don't actually have an idea that I'm fully in love with

so I never get to get good at game programming

which makes me less confident, which makes me think about actually starting way less

@AlexMitan Part of why people recommend starting small is so that you can get a small feedback loop from starting a game to finishing on it. You don't need to fully love a game concept nor make a large endeavour out of it — do something you're interested in, do something small for it, and learn from the experience. Then do that again if you'd like.

The stuff you work on doesn't t have to be something you're fully in love with. What's important is you think you'll derive satisfaction and enjoyment from working on it, and you think you'll be able to bring it to some level of completion you'd be satisfied with. Leaving it when it's a half-finished prototype or concept exploration is fine as long as you learned & explored what you wanted with it -- move on to more important things if you don't want to bring it to 100%, for example to sell it.

@doppelgreener I feel pretty burned out, like my brain is shortcircuiting and I have to take painkillers

right now you ought to let yourself get some rest then :P

and yet I have nothing I'm satisfied with because it's all half-baked stuff that I'm alone with

it's 2 pm, it's not like it's late at night or something

besides, I have no idea what I'd do

I really really want to make something

I know that feeling

There have been times when I've had the drive to do /something/, but without a target for that.

as far as the half-baked stuff goes, what might help dealing with that is having clear goals for yourself. what were your goals for that project? did you achieve them? is there anything missing? if your goals for that fleet game were "experiment with ECS and this tree-based communication system" then you have achieved your goal for that project.

it is OK to do things just for study and learning and experimentation. it is also OK to build prototypes and choose not to progress on them.

Yeah, but the larger-scale goal was "find an architecture that I'd like to make a game in"

I like ECS, but I haven't fully experimented with it, in that I have yet to really USE-use it

i think you're being a bit too academic with that question :P

I just brought it to the point where it's similar to the first game with the shootyshoots

But I haven't taken advantage of the ECS by adding interesting behaviours easily or something

write games, not engines: 1. find a game you'd like to make, 2. write that game, 3. implement the architecture necessary for that game, 4. use ECS if that's the architecture you need, 5. implement the ECS functionality that creating the game requires if it needs any at all.

you will USE-use ECS when you aim to make a game (step 1+2)

because you will have practical, not theoretical, demands and constraints

o/

hello everyone

You're right... I feel like I'm anticipating "meta-problems", such as "if this won't get implemented quickly enough, I'll lose interest in the game"

every game has a hype cycle in its development: at the start you're going to be all excited and driven by that excitement because you're working on and learning totally new things every day, then things will feel like they're slowing down (such as when you're developing content, you might not feel excited that you're doing something completely new every day) and you won't have that excitement to drive you. instead you'll need discipline to keep moving forward.

I find that ECS itself is often not the right thing, but the reason for it over OOP is very valid. Subdividing classes into more modular components is important.

Too often I see bloated structures where everything relates to everything else in this giant tree of inheritance.

Hey everyone

hey william

Random question. I know you can make textures with a larger range on each color. Like 16 bit per color. But can you make textures with more channels? So still 8 bit per color, but like 8 channels?

I mean, one four component 16-bit texture uses the same memory as a eight component 8-bit texture.

I do like ECS so far, it feels like the game dev parallel of the visitor pattern, which I'm a fan of where I've used it

Just curious if the shaders knows about more than four channels normally.

I think there are more available, yes.

i know that e.g. the photoshop file format supports an arbitrary number of channels (between 1 and 56 according to its spec: adobe.com/devnet-apps/photoshop/fileformatashtml)

Also I feel like the most interesting ideas I have would require ridiculous levels of AI

I seem to recall the ability to set two channels each for RGBA in OpenGL but I'd have to find where I saw that.

Which will be interesting to work on when I get to it, but the sheer amount of AI I'd need to make just to get the game up and running intimidates me

i haven't written shaders before, but if you were making a 512x512 texture, could you mimick an 8-channel format by using a 4-channel format and making a 1024x512 texture instead?

(every 2nd pixel -- or the 2nd half of the image -- would be the additional 4 channels. that sounds like it's coming close to "make a second 512x512 texture" though.)

@WilliamMariager This might help: khronos.org/opengl/wiki/Image_Format

(assuming you're using OpenGL)

As far as I can tell, it's all 4 component textures.

I guess I'd have to do a double pass then.

I just had the random idea of also keeping track of non-visible light in a 2d lighting engine.

I use a fragment shader powered lighting system.

I'd like to track things like infrared as well.

At some point I might come back to the 2D lighting engine I was trying to make.

I could use alpha for that of course, but then I have no additional challenges.

I made a nice floodfill one that takes materials into account

If you had a white light source, it could hit red glass and the light would continue as red on the other side.

Basically had a resistance layer.

So you might be able to do that with what I had in mind

For each lightsource, render everything it might hit to an occlusion map

Then, map that from Cartesian to Polar coordinates

(so that the top is the closest to the light source)

Then, fill from the top

When it hits something that filters the light, continue flooding with that filter to the rest of the column

This was a version where I did single-pass but with persistance :P

So basically a single source of light would spread out

And then fade out slowly if moved

It's a bit artistic, but it was a nice effect.

Here's a small enclosed room.

I I did a multi-pass one as well where the effect was instant. It was more realistic, but obviously also more taxing.

And had a limited range.

Something like 4 tiles per pass, so you'd have to do light_resolution/4 passes to get accurate results.

Here's a multi-pass with RGB colors

Was trying to do 2d source approximation for doing more advanced single-pass lighting effects like normal mapping.

I basically sampled a 3x3 area of light cells and then approximated where the source was, which gave me a direction and strength.

It wasn't perfect and I never could get it to look quite right.

@Garan, this image shows what I do

In the top left

The left-most is the resistance layer, the middle is the light source layer where each tile that provides light is added and finally the right image is the output. I then use that and overlays the scene.

I see. Can you increase the brightness of a tile to turn it into a lamp of sorts?

Well, basically any tile that has light is a lamp

So if you have a light/torch/lamp entity, it'll force a certain light level on it's space and this will propagate via the floodfill

For directional lights, you'd have to be a little creative. I did consider something like the bressenham line algorithm

Also, light propagates around corners and through materials.

It's for a certain type of game and not for everything.

I made it for a Terraria clone back when Terraria was just released, because I wanted to see if I could make something like it.

it looks like this operates on a basis of objective lighting levels, e.g. characters' eyes don't adjust to lower/higher lighting levels; being a terraria clone explains that :)

not that either way is right, but i was considering a few minutes mentioning if you'd considered subjective lighting

You might also be able to do a mostly baked lighting- cache it and only update if a tile has been changed.

The issue then is a moving lightsource.

Yeah. The point of this engine, especially the single-pass that slowly spreads, was that it wouldn't need to do any performance tricks.

It was highly efficient, doing a single fragment shader pass per frame.

It could be upped to like 4 or something for a snappier effect without having much effect on the performance.

The resolution made it fast as well, considering it was 1 pixel per tile.

What kind of propagation do you have in mind?

The kind I described above, with a mapping to a radial map and then a flood to figure out what's blocked.

Ah

It's really the same thing as several raycasts, just a lot easier to calculate.

I did play with the idea of making the light sources directional. Just not sure how I'd do ambient and combine two directional lights into one if a cell was hit by both.

Again, going for a single-pass approach with infinite light sources instead of a limited one.

There's a lot of material on using regular point lights. Not so much for cellular stuff like I was doing.

Well, I'm off again. Just had the question about more channels pop into my head. :P Talk to you all later.

ttfn o/

Hey y'all!

@TheMattbat999 Hey

Hello again ...

So I was thinking about the directional celluar lighting.

And I have a way of doing it :P Just way too few channels.

I'd need a channel for each direction I wanted to support ...

So 8 red, 8 green and 8 blue. :P

Absolutely

It would only be in 8 directions though.

I guess if something moves in one direction, it could spill a little into the others.

Could you approximate the angular distribution with a polar function, instead of storing each direction explicitly? (Similar to light propagation volumes)

What kind of polar function are you thinking? A key feature of the lighting is that it doesn't have point lights.

Effectively a Fourier series in the angular domain.

Hello!

@TheMattbat999 'allo :)

How are things?

Can't complain. :) On vacation atm.

Nice. Just like a stay-cation or what

For now, yeah. May take a trip next weekend. Just decompressing so far though. How 'bout you?

Running low on.. inspiration for my game, Crying Moon. Now kinda sitting around until i come up with something or gotta head for church.

Other than that, fine.

@DMGregory Enjoy!