Saurian Sorcery

[main cosmos]

there's this game called Tunic that has relatable characters in it

[obsidian wolf]

ooooo ty

i really should consider (past or present) trying to simplify my models a bit more

it's easier to make the rest of everything when matching the fidelity that way

though my first thought is that game relies a lot more on lighting than i wish to

[obsidian wolf]

this still needs some work but its a fun start

[severe spruce]

cool stuff!

is your pipeline forward? have you considered just throwing msaa on it?

[obsidian wolf]

its deferred

[severe spruce]

daww

[obsidian wolf]

i should try to implement fsr2 at some point, yeah?

[obsidian wolf]

https://sketchfab.com/3d-models/rafaj-the-mulberry-warlock-13fd93cd06164e4c892478e57306a24c
i found this model online

and the idea of it actually fit generally really well

so i downloaded it and tried to clean it up

actually turned out pretty well

[obsidian wolf]

part of it is causing me to look back at some of those tweets

as i understand, mr. stalberg has some method of determining outline sensitivity

so for his grass texture, it outputs a low sensitivity, and the rock outputs a high sensitivity

grass on grass = few outlines, on rock = more outlines

i have this issue here where it's missing some outlines that i'd want, like on the back sides of this band

but if i set the depth outline to anything that'd actually show those, it'd be way too sensitive, so i almost want to force outlines sometimes

doing outlines on material borders could help here, but i think i'd need to verify that it wouldn't fuck me over in other places

[main cosmos]

tails

[obsidian wolf]

o_o

are u pointing out the sonic similarities

that's the first thing my friend said too if so 😭

[main cosmos]

lmao

I am indeed

[obsidian wolf]

https://tenor.com/view/tails-dancing-microbro-sonic-gif-26270469

i got sent this

whatever bro if somehow this does actually end up being a game

i can grift to the sonic furry crowd

[main cosmos]

Huge market potential tho

[obsidian wolf]

https://sketchfab.com/3d-models/week-2-salamander-458daf7c543a4cb7bf430b7063ab3d9b
man this is from the same guy, its beautiful

i really need to figure out how to do animation like this with my system, to make it more jolty

much more lizardlike than the two pose smoothstep interpolation that i'm doing, but idk how practical it is

[main cosmos]

couldn't they just be using a very small time delta between the two poses

hmm actually idk, I think you're right

[obsidian wolf]

no the way that you can tell is because the tail isn't bound by it

that, my system cannot do

if i wanted to support it, i'd have to support keyframes. supporting keyframes is very hard from an editor standpoint

if i had it, i could just set the two for the animation cycle for the tail, then like 4-5 for the head like this is probably using

[main cosmos]

per-joint keyframes?

[obsidian wolf]

yeah

it wouldn't be the hardest thing ever, like i can imagine most of it which is a good sign, but still a minefield of timesinks

[main cosmos]

just try it, what's the worst that can happen? ||||

[obsidian wolf]

the reason why i'm set for now on not is because i really want to actually make some amount of content, and this isn't stopping me from doing

i've always sucked dick at making content

i always make shit that gets me to the point of making content

and i make a single thing with it, and i'm like yes, okay this works, and then usually i move on

[main cosmos]

ye

[obsidian wolf]

it's not intentional though and i'm trying not to do that anymore

[main cosmos]

you could make the whole game without animations frfr

though it adds to the feel and theme (like audio), so it has some importance

[obsidian wolf]

i will say that idle poses add a fuckin lot

the croc still doesn't have idles

[main cosmos]

hehe wdym completely stiff statues feel fake

[obsidian wolf]

and i just stopped placing him essentially because of that

i have the intent to modify him a bit, or at the least give him gear though

[dry monolith]

🤔 what you system does instead?

[obsidian wolf]

i have a library of poses, and a pose-mode widget to set them

but poses set every single bone indiscriminately

[dry monolith]

so you have a keyframe system, but one keyframe has to contain all transforms for whole skeleton?

[obsidian wolf]

yes

it's not harder from a renderer side, but the UX side is a lot harder

[dry monolith]

it's suboptimal data-wise I'd say

storage-wise

[obsidian wolf]

for sure

[dry monolith]

otherwise I don't see a problem

you are probably missing one most useful thing in animation workflow

it's the inverse kinematics

[obsidian wolf]

i don't think so, but i've only speculated about how IK works

[dry monolith]

make an IK rig and animating becomes easier mutiple times over

[obsidian wolf]

oh

in that sense yes

i was gonna do IK at some point, but i was expecting it to work by applying some transformation after posing

so set pose, then apply IK and ideally IK applies minimal enough change to keep the pose in tact

but from that sense, i should probably try to do it soon

so i don't waste time doing all of this by hand

[dry monolith]

IK applies maximal change within constraints

the goal is to reach some point in space and you have to solve a chain of parented transforms with joint constraints and IK solves this problem

[obsidian wolf]

isn't it minimal change? i'm not sure i understand otherwise

minimal change to reach the point in space

[dry monolith]

I actually don't understand what you mean by minimal change so I probably misunderstood you

[obsidian wolf]

yeah i'm using it in not a math sense, i just mean that it won't completely destroy the pose

like if a pose is set, then a reasonable IK is set (like setting foot position), it'll still keep the leg pose

[dry monolith]

there's no guarantee the change will be minimal, it's just that there are multiple solutions and IK solvers are sensitive to initial state

so solution ends up resembling initial state

[obsidian wolf]

one thing i'm curious on, that you might know, is it usually required to make joints and joint constraints to get decent IK?

yeah that sounds right

[dry monolith]

well those constraints describe the system so I guess if you don't want solver to simply stretch the arm to reach the point you need to constrain it

I can easily see FABRIK doing this without any constraints

[obsidian wolf]

i'm looking for ease of use at the cost of quality, sounds like FABRIK is what i want from looking at it

[dry monolith]

ye fabrik is easy, transform forward, constrain, transform backward, constrain

do N iterations until it looks good

https://www.desmos.com/calculator/ky7guavklk

simple variant for length constraint

[obsidian wolf]

neat, this is super useful

hopefully i can do this next weekend

[obsidian wolf]

https://sketchfab.com/3d-models/trogdoooorrr-the-burninator-7fb6c05106cc404eb95b4a8f9feceb9b

[severe spruce]

TROGDOOOOOR

amazin

[main cosmos]

Random thing someone linked on another server
https://gfx.cs.princeton.edu/pubs/DeCarlo_2003_SCF/DeCarlo2003.pdf

[obsidian wolf]

i will read this tmrw

ty

[obsidian wolf]

spoiler: im not doing IK

(yet)

[obsidian wolf]

was hoping i could get this working before the weekend was over, but only a resemblance

weekend was taken by touching grass 😩

[main cosmos]

what's this

tile generation?

idk the term, but I think I can see what it is

[obsidian wolf]

yeah

tilesets

this is the input, can add models and select which corners are inside, which corners are outside

basically i dont want to model levels, so i can either make them procedurally or just very quickly in-editor

this is why i wanted to do it

https://images-ext-2.discordapp.net/external/hzjGqjattu_5vSWhxKbvHfxtFUEps-5KZsro1_YXFTI/https/pbs.twimg.com/media/FBk2PhcXsAETv1x.jpg

[main cosmos]

ah I 'member

[obsidian wolf]

fixed it during my lunch break, the missing sections are just because i didn't add something that can fill that space yet

[severe spruce]

did the fix fit in your lunch break or did your lunch break expand until the fix was found

[obsidian wolf]

uhhh i don't want to do anything until this PR gets looked at lmao

so i found something to do

[obsidian wolf]

not a big fan of this tileset but

lotta functionality at least

i wanna add tiles that span multiple squares

fun fact, this codebase is at just under 18k lines currently

[obsidian wolf]

🥴

[obsidian wolf]

i think i want to grid-ify the enemy movement a bit more but im not sure how

https://tenor.com/view/zilliqa-zilchess-chess-gif-20509231

:^)

[obsidian wolf]

i sorta want to do a frog with a big sledgehammer

that very obviously stretches his limbs

like he basically just throws it with elastic arms

[main cosmos]

dat frog looks like a turtois

[obsidian wolf]

yeah he found a tortoise's shell and made it his new home

[dry monolith]

battle shovel

[obsidian wolf]

mannnnn

i should really download like unreal or godot or something and play around with animation players

see what the graphs do and what functionality is needed from them and for them

but god does that sound boring

[dry monolith]

what graphs?

[obsidian wolf]

state graphs

[dry monolith]

you usually have a timeline with keyframes that's about it

[obsidian wolf]

that's just for single animations

[dry monolith]

what's multiple animations?

[obsidian wolf]

but for example if i want to add an attack sequence

[dry monolith]

an animation is an interval on global timeline

[obsidian wolf]

i don't want to define that in code

[dry monolith]

or you can think of it as individual timelines, it's equivalent

[obsidian wolf]

animation assets aren't the problem, choosing which animations to use is the problem

[dry monolith]

but that is gameplay programming

[obsidian wolf]

i am doing gameplay programming

[dry monolith]

in unreal you have blueprints

and state machines and shit

behaviour trees

[obsidian wolf]

yuuppp i gotta figure out what functionality i need to replace that

[dry monolith]

I am personally unfamiliar but know of those from random youtube videos

[obsidian wolf]

same, i always avoided 3d when i used the game engines

[dry monolith]

well those are just programming patterns made into GUI

so that an artist may fiddle with them using buttons and knobs instead of typing text

there are at least two animation related features that are somewhat essential - IK and animation blending

animation blending is basically bone influence weight mask and weighted average between TRS with it

what IK is you already know

it's usually stabilizing refinement step after blending

but is also useful in animating by hand

[severe spruce]

really nice!

[obsidian wolf]

much more happy with the appearance of this tielset 😄

the blender file for this btw

[main cosmos]

at 5 seconds it looks kinda like a hot dog

[obsidian wolf]

holy fk its the hotdog tileset

[main cosmos]

frfr

[obsidian wolf]

the way this works is that there are 8 corners, and each corner has a value (here, the red stone is 1, and the sandy value is 2), then when it's time to build the code just iterates through and rotates around to check for viable candidates and selects one at random
i'm reworking this slightly, there's sometimes cases where tiles can represent multiple corner values, for example in this case

four of the corners are clearly representing sand, and the corners closest to the camera are clearly open air, but the back two corners could represent anything solid

so i'm changing the corners from being values to being bitmasks, (then checking that two are equal is just corner_bits & other.corner_bits != 0 for each corner) but there's an interesting side effect of this where the "empty" value has to become a bit as well

that took me a minute to come to 🙂

also, if i do this in the future, i should really find some sort of way to automate generating all of the relevant cases, there are a lot

like at least 100

[obsidian wolf]

this is finally enough coverage to make the maps i drew out over christmas break

🥴

[obsidian wolf]

jk

missing 1

[main cosmos]

thing, I don't know why

[obsidian wolf]

huh?

[main cosmos]

(I have a brainworm)

[obsidian wolf]

i hate thes eshadows

i think i've never seen shadows look this bad because i usually blur them

and slight shadow acne doesnt look bad if it's blurred, just looks like overattenuated light from the angle

the bottom part is it escaping the frustrum i believe, so ignore that part

[obsidian wolf]

god this looks so much better with a background

[obsidian wolf]

made the FPS local so my PC no longer gets tanked when i draw a block, also allowed vertical flip to cover more cases (though my tileset isnt' designed with that in mind)

[severe spruce]

thats p cool looking

[obsidian wolf]

added 3d pathfinding

[dry monolith]

a star?

[obsidian wolf]

yeah

[dry monolith]

a star that's using a star?

[obsidian wolf]

can't tell if punning with the target or actually asking a question LMAO

[dry monolith]

punning

[main cosmos]

Make a star and do a star on it

[dry monolith]

first one was legit question

[obsidian wolf]

yeah its just astar with some funky stuff for ramps adding vertical neighbors

though it doesn't seem to quite be bugless :^)

[main cosmos]

bugles 📯

[obsidian wolf]

trumpet

[main cosmos]

sorry I could only find the curved trumpet emoji

[obsidian wolf]

foiled once again by not using enum class

curses

also foiled by not fuckin making this constructor explicit, dont know how this escaped for this long

i wish constructors were explicit by default >w<

oh wtf it was literally only my 3d vector class that didn't have explicit for that

[obsidian wolf]

well, s'all fixed, outside of the asymmetrical ramp usage which i don't really care about atm

[obsidian wolf]

i sorta wanna do funky stuff like this for foliage

not entirely, but at least some alien elements

[main cosmos]

Looks like a thingy

[obsidian wolf]

actually crazy that you can pick up on the resemblence

that's literally exactly what i was making

[main cosmos]

it reminded me of this thingy I made in blender

[obsidian wolf]

how'd you make that thingy?

i made my thingy with the skin modifier, tried deleting edges with geometry nodes at first but it had bad results so i just did it by hand

[main cosmos]

uh I think I used a wireframe modifier, then applied some other modifiers

the wireframe made it look like the red one in the center

I'm a big blender noob so I can't fully recall what I did besides mess around until I saw something interesting

[obsidian wolf]

issa thingy

[main cosmos]

you figured out my blender secrets

[obsidian wolf]

with these two simple tricks

[dry monolith]

there's also some disabled by default addon that I can't remember the name of which allows you to replace each polygon with a mesh

so you can easily model fences and stuff

[obsidian wolf]

geometry nodes exists now which you can use

[dry monolith]

I guess so

[obsidian wolf]

i still think they're incomplete but they feel pretty useful now

[obsidian wolf]

this actually gives similar quality results for a lot less effort lmao

also there are more fun things to do with it

[obsidian wolf]

cleaned up version of a 3d model i found

[obsidian wolf]

in game

i need a tree

[obsidian wolf]

now that i fixed the shadow acne, the pixel flickering, a lot of the shadow issues, and added a non-black background, things are so much nicer to look at

who would've guessed

[cedar rune]

I quite like the art style :)

[obsidian wolf]

ty is coming together a little bit

when i have some time im hoping to get characters moving around again, which will be neat because this is the first time the environment feels real

[obsidian wolf]

money making

[main cosmos]

gameplay 👍

[obsidian wolf]

still a ways to go until i have a testable level (i want to add some sort of attack to the fruits, hence the lizard health bar), but for sure 😄

[cedar rune]

What is this, someone making a usable product instead of unfinished engine after engine

Heresy

[obsidian wolf]

WIP guy

he will have a bow

but until then he will just look like the emote

[main cosmos]

like a 🏹, or a 👔 (I couldn't find a bowtie emoji)?

[obsidian wolf]

🏹

[main cosmos]

a bowtie would look cool too

[main cosmos]

what is Guy's last name going to be

[obsidian wolf]

he doesn't even have a first name yet lmao

mr. hanther

running dnd campaigns has given me the cursed ability to pull absolutely random names out of a hat

[main cosmos]

lol do your players like to ask random NPCs for their name

[obsidian wolf]

i think it happens almost every time

[main cosmos]

my brother is running a campaign right now and he called two goblins we encountered Boblin and Noblin

those were prepared names

[obsidian wolf]

i liked my series of goblin names that used no vowels

my particular favorite was "n'd" because my players thought i choked when i said his name the first time

[severe spruce]

based & nased

[obsidian wolf]

also i had pattern recognition going in my brain for the hands but could not remember what the image was

this was it lmao

[severe spruce]

glad that i was not the only one seeing it

[main cosmos]

us players did CBT to some poor ogres in our last session

it was pretty epic when I rolled a crit though

[obsidian wolf]

does this mean he fucking dies or does it mean he turns into a ogre with degenerate pleasures

[main cosmos]

it meant I got an epic finisher animation on an ogre

(double meaning not intended )

[dry monolith]

what workflow are you using btw?

low poly + subdivide?

[obsidian wolf]

yeah

[dry monolith]

can you unwrap a mesh without having to apply the subdivision modifier?

in blender

I don't know never tried

[obsidian wolf]

yeah subdivision doesnt really hurt uvs

[dry monolith]

btw sucks that blender has no material painting

so you have to paint one pbr texture at a time

[obsidian wolf]

but i applied in this case regardless to fix up some verts, do the armor, and do mats

im not familiar

[dry monolith]

they really should upgrade texture painting

if they implement material painting then substance painter may lose a big chunk of people

[obsidian wolf]

i never used painter :(

but i think i know what you mean then

[dry monolith]

what I mean is basically configure a brush that paints to multiple textures simultaneously

so you can paint whole materials

[obsidian wolf]

im one of the few people on earth thats used substance designer for over 100 hours but never opened painter

and yeah gotcha

i also wish you could bake textures as part of the export process without doing anything custom

that way procedural texturing in general worked better

[obsidian wolf]

with or without? 🙂

[main cosmos]

save the horns for the upgraded version

[obsidian wolf]

lmao it would work well for that

1,2,3

dont think i'm doing unit upgrades though

[main cosmos]

then decide whether to give it horns based on how badass you want it to be

[obsidian wolf]

it's interesting how neutral it seems to be

i asked 2 other people and neither of them decided either

[severe spruce]

they aren't very prominent

[main cosmos]

they both look good so ultimately it's up to wherever fits the game imo

[main cosmos]

make them rainbow-colored

[severe spruce]

if they would be more central to the design, then maybe they would be more impactful

but i like it without

[severe spruce]

maybe a ram character with horns?

[obsidian wolf]

yeah i think im going without, other designs can get horns and this boy is kinda slick

[obsidian wolf]

this boyo is posed

gonna see if i can export poses from blender directly

[obsidian wolf]

btw i like the name saurian sorcery 🙂 though idk how many people know what saurian means

someone give a take on it

[obsidian wolf]

Saurian Sorcery (Tower Defense Game)

[obsidian wolf]

been updating my skinning system a bit

i only supported basically 1 instance of skinned models, so if i wanted to have more than one of a type of lizard it wasn't possible

went from having a SkeletonCPU struct and a SkeletonGPUstruct, which is how most of my objects work that have some interface to the renderer

to now a SkeletonPrefab in addition to those two, which is what i usually have for gameplay side objects, which contain information for building entities

which kind of breaks the pattern and incentivizes me updating some other asset structs around it, but the reaosning is that i only keep one SkeletonPrefab around, which holds the poses

so TypePrefab structs are generally immutable, serializable asset data, TypeCPU structs are instances of Prefabs which runtime (update functions) needs, and TypeGPU structs hold handles to the GPU data

i think this makes sense?

also in some cases, TypeGPU represents data which is trivially uploadable to the GPU, but it works for me because in this case, binding handles is similar enough to uploading trivial data

once i finish cleaning this up, i'm going to import poses from blender/gltf

main reasoning is so that i don't lose data in the future, but also so i can take advantage of blender features, like not needing to implement fully featured IK with constaints and everything, being able to pose before i go through the export process, etc.

oh also lastly, i'm being explicit with my animations, constructing a list of states and i'm going to explore what the pose controller system will be able to do within those states (for example, is an ability where a lizard charges and swings considered one state?)

no fuckin clue yet

[main cosmos]

I haven't done animation yet so I dunno, but I'm still watching 👀

[obsidian wolf]

aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa

i shouldn't have done this all in a day, i'm going insane

[obsidian wolf]

just had a memory of adding in temporary code somewhere

and now i can't remember where

[dry monolith]

https://tenor.com/view/dementia-gigachad-gif-25845229

[obsidian wolf]

poses in blender on the keyframes, i import the .gltf to engine here and convert the 3 poses to an idle sequence

[dry monolith]

so the takeaway is that using poses to animate leads to messy workflows

[obsidian wolf]

abnormal yes, neither gltf or blender are set up to handle this sort of animation, that friction is certainly real

messy (or do i regret it), idk, try and find some other projects where one guy NIH's a 3d engine like this

animation is hard

[dry monolith]

what do you mean neithyer gltf nor blender are set up to handle this sort of animation

[obsidian wolf]

they have no concept of a "pose" beyond having the ability to express them

i think i could've automatically imported these pose sequences as animations, but i don't know if that would be good

[dry monolith]

name a few pros of using poses to animate

it should be clear why there's no such system anywhere

[obsidian wolf]

they mostly come down to being simple,
from ux: you can interpolate between any two at any point and it generally gives decent results
from engine: easier to set up the tools to manage them

[dry monolith]

ok but those points are not unique to the system

[obsidian wolf]

because very few people try to make their own engine where they also make their own characters and animate them all in 3d

[dry monolith]

the keyframe system is generalized form of the pose system

[obsidian wolf]

yeap

if i made anyone else use my code i'd certainly try to generalize it

but you're not going to be able to implement keyframe animation from the starting point of hardly understanding gltf and having a horribly hacked version of it in your engine in a day

[dry monolith]

anyways I don't want to keep this conversation going it feels like I'm attacking you for no reason

[obsidian wolf]

nah i like defending things

i mean it's certainly true that i might be in the category where given all knowledge, it would be best to just suck it up and do full keyframe animation rather than spending time managing the constraints

but i went into animation as a whole with almost no knowledge on any of it beyond general graphics programming

[dry monolith]

my primary concern is that you're probably wasting resources by sticking to it

[obsidian wolf]

and given the question of "should i switch now", the answer is probably not

[dry monolith]

eventually it's gonna have to go because it scales poorly

there is also a problem of usability that you have to pose every single bone for every single pose

and you cannot change two poses to interpolate between them if there are a few more pose frames in between

[obsidian wolf]

so the question is less

are there drawbacks to this

and more, is there something better

if it's just dropping the pose (all keyframes) limitation

[main cosmos]

the animation section of game engine architecture was so long that I skipped most of it 🥲

[obsidian wolf]

then what do i get out of that? because i can't animate like real games

infeasible with how much time real games spend on animation, both authoring and programming sides of it

[main cosmos]

ngl the fact that you have your own animated assets that actually look good is quite epic

[obsidian wolf]

yeah i coulda had star game jaker

[dry monolith]

in code I think it's pretty few locs

[obsidian wolf]

this is what u did for saying "le lizard and fruit looks so cute xdd"

[main cosmos]

ebin :DDDD

[dry monolith]

    struct Skeleton
    {
        std::vector<Bone> bones;
        std::vector<Animation> animations;

        //compute a set of bone transforms using animation index and time
        void ComputePoseTransforms(Transform* pose, size_t anim_idx, float time);
        //turn a set of bone transforms into a set of matrices used for rendering
        void ComputePoseMatrices(glm::mat4* matrices, const Transform* pose);
        //apply inverse bind transforms to finalize the process
        void ApplyInverseBindTransforms(glm::mat4* matrices);

        float GetAnimationDuration(size_t anim_idx);
    };

the thing that implements the bare bones

[obsidian wolf]

it's all UX

[main cosmos]

the book goes deep into the various ways of representing animations with lots of code ex🦐les

[dry monolith]

what part of ux

there should be basically no difference between most of the ux for both systems

and you also can leave the animation process itself to 3rd party tools

at least for the keyframe based one

[obsidian wolf]

80% of it is "how do i author content", making sure that it's possible to edit it and use it

if it's, make a flail animation in blender, make an idle animation in blender, set idle state, on falling, blend to flail state

1. what does blending to the flail state look like
2. is it literally just a question of quality then? because the animations in blender are going to be made with 2 poses, and adjusted with additional keyframes to correct problematic areas

[obsidian wolf]

shrimple idle/flail animation (ignore the attacks)

[dry monolith]

1. if you look carefully at the code I posted, the keyframe based system always computes a pose from keyframes and then turns them into set of matrices for rendering, so you should be able to interpolate between two arbitrary animations by interpolating the transforms of two poses obtained from them, before turning the final interpolated pose into matrices

2.idk you animate bones lol

there will simply be less bones to move around because you can place a keyframe anywhere on the timeline

[obsidian wolf]

1. now we have multiple animations playing at once
   either way, it's not so much that i think it's better, but switching systems requires a reason for me, possibly some knowledge of an impassible obstacle, which is more a question of looking at a large set of animations i'll need for this game and considering it between these two systems (which i'm going to do now)
   unlike the traditional GP way, i am practicing programming in service of an actual product, rather than engine dev purity, so i have to consciously ignore seeking improvement if it doesn't serve the bottom line

[dry monolith]

I wouldn't say that you have to play multiple animations at once

you may obtain 2 poses, one is the end of source animation and the other is the beginning of the destination animation, and interpolate between these for some given transition time

so it's same as playing just one

interpolating between 2 playing animations would probably look better though

again the keyframe system is the generalized pose system

so what one can do the other can as well

[obsidian wolf]

yeah, but the restrictions of the pose system allow you to skip the first key frame of the second animation and replace it with your start pose

because the second key frame of the second animation is all synced, the interpolation is even across properties

more so, you just use that transition logic for everything

[dry monolith]

I don't get it

[obsidian wolf]

if it's a question of "how do i get a good idle animation into my game", it's doghsit, there's no point

if it's a question of "how do i convert an animation system's state machine into poses", it's a lot better

[dry monolith]

I don't get it 😩

all you said is still applicable to keyframe system

[obsidian wolf]

because the second key frame of the second animation is all synced, the interpolation is even across properties
you don't have this guarantee

[dry monolith]

you can do this if you copypaste the pose in the editor

but I don't see how it's mandatory

the interpolation will always smoothly transition between any pose

[obsidian wolf]

if you're just saying you can artifically apply that limitation to yourself

is that what you're saying actually

[dry monolith]

maybe

[obsidian wolf]

ok

[dry monolith]

I don't understand the purpose of it

anyways how about we wrap this up, I need to go

if you have anything to add, I'll read it later

[obsidian wolf]

if you blend anim 1 and anim 2 (i miswrote)

and you cut out d for the interpolation step, then c-e will be a quick interpolation, c-f will be a medium interpolation, and c-g for the third track will be a slow interpolation

i.e. random parts of your animation will still be showing the old animation while other parts will switch to the new animation very quickly

but if b,e,f don't exist, the interpolation between c and g will be even across all tracks

either way, if you treat all animation as transition between two states (where cycling is a case where the end state is equal to the start state), then you need to animate all state transitions

you can do this trivially by blending between these two states

and you can make the system more robust by allowing more keyframes

but because there's a lot of state transitions, that's a lot of keyframes if you want to have really good animation. if you don't need really good animation, you can skip programming this step and just rely on the simple blend

it should be clear why there's no such system anywhere
either way, this isn't my idea, its from overgrowth https://www.youtube.com/watch?v=LNidsMesxSE, not necessarily suggesting you watch it, only that it's available if you have the desire yourself 🙂

fun convo tho, i appreciate the interest

https://tenor.com/view/monkey-reaction-confused-baby-monkeys-gif-22432417

[dry monolith]

I think there's a misunderstanding somewhere

[dry monolith]

in keyframe based system the whole timeline is a continuous set of poses, and so it is with pose based system

due to interpolation being in place

so you could take any pair of time t0 and t1 and sample a pose and interpolate between them

a pose is nothing but a set of TRS per bone

[obsidian wolf]

are you just saying that you can add additional keyframes at e for track 2 and 3?

[dry monolith]

track?

[obsidian wolf]

in this example, track meaning the horizontal line that corresponds to some animated property

[dry monolith]

I am saying that timeline is a continuous curve and each keyframe is a control point

you can sample a pose using time

this fact is true for both systems

you add redundant control points which is a flaw in my eyes

[obsidian wolf]

see here: i do not care about engine dev purity

i do not care about flaws unless they are impacting my progress tangibly on the actual product i'm working on

that's a bit of a lie

i do care, but i don't like caring and so i am consciously trying to not care

[dry monolith]

how do I stop being passive aggressive

[obsidian wolf]

i believe this is where you have to make a bit of a stretch to empathize with my goals

[dry monolith]

I was trying to make a sentence but it always comes off as getting personal wtf

[obsidian wolf]

LOL

man you should see my conversations with my and my lead

he's a dry swedish man and so all of our conversations would come across as an argument between two people who hate each other

well

[dry monolith]

basically to put it bluntly I'm not trying to persuade you in any way

because I can see that there's a certain degree of sunk cost fallacy at play

finish the product then maybe consider it again

I wanted to bring up that the workflow could become suboptimal because of how many things you need to tweak per pose, and it grows with each individual pose added to the animation

as amount of bones times amount of poses

it works for overgrowth well because they rely on the procedural generation a lot

[obsidian wolf]

hm, you know this is how i'm adding poses, right?

[dry monolith]

it simulates forces and does IK

yeah but then when you need to edit it you need to propagate the changes manually

across all poses inbetween

[obsidian wolf]

what

can you rephrase

[dry monolith]

say you need to edit something in first pose and last pose

now you need to edit all of them

[obsidian wolf]

🤔

that's how i interpreted what you said

[dry monolith]

wait what does it do

you added a keyframe to one pose, correct?

[obsidian wolf]

i modified the tail on the first pose in that second video

where the first pose maps to one of two poses by my idle animation from my engine skeleton data

[dry monolith]

okay but does it apply to all of the poses?

[obsidian wolf]

no, but thats definitely not desirable

[dry monolith]

if the change is in first pose of the animation and last pose of the animation you're going to have to edit the poses inbetween them

[obsidian wolf]

there are only two poses per animation

[dry monolith]

I see

my point is moot for your use case then

[obsidian wolf]

low quality

[dry monolith]

that should settle the debate

[obsidian wolf]

sort of, maybe you're more convinced than i am now xD but yeah

it's a little funny because like

right now, looking forward, there's a shit ton of work, so really it'd be sub-par for me to spend more than this weekend on this animation problem

i can totally imagine a future where i finish a bunch of things though, and i look again at the animations

[dry monolith]

that's the sunk cost fallacy part I was talking about

[obsidian wolf]

and they dont' meet my standard

but i can't really focus on that future, if i'm in that future, i'm succeeding

[dry monolith]

it's reasonable in your case though since you seem to be time constrained

[obsidian wolf]

yeah i don't really work on this on weekdays unfortunately

and i don't want to bring anyone else on because i'm not trying to make money from it

so they're incredible time constraints xD

[obsidian wolf]

made another guy this morning

[severe spruce]

spiffy lookin fella

[obsidian wolf]

i like his little sash, but it also kind of looks like he draped orange toilet paper upon himself

[severe spruce]

do not lick the orange toilet paper

[main cosmos]

Looks like a wise monk

[obsidian wolf]

yeah i can't decide if i want to lean into it

thinking about what object he should hold

considered a floaty water orb

[severe spruce]

scroll?

[main cosmos]

Monks use quarterstaffs ofc

[severe spruce]

four quarterstaves

[dry monolith]

https://en.wikipedia.org/wiki/Khakkhara

both

[obsidian wolf]

tin stick

that's actually a really nice design though

the one thing that i've been considering more recently compared to when i started

is thinking about the casting/attack animations before i settle on any character design

[main cosmos]

Obviously the main thing you're missing are lusty argonian maids

[obsidian wolf]

like here, it's not exactly easy to figure out how a lizard that holds a pot should actually attack

a lot of mage characters in games have attack animations that essentially just look like they're throwing a baseball

maybe i just make every character with a staff swing it xD

[severe spruce]

swinger characters, hmm

[obsidian wolf]

considering yoinking the blurred normals from teardown

@main cosmos i'm assuming there's some depth discontinuity rejection on the sampling for that, any details?

[main cosmos]

it only blurs normals that are near the camera, for one

it probably does a depth discontinuity check too, though I can't remember for shure

if you have teardown you can just renderdock it urself

[obsidian wolf]

mayhaps

[main cosmos]

it's a little tricky to run under renderdoc, but I can help if u get stuck

you have to use the global hook instead of the normal way

[obsidian wolf]

if i'm feeling adventurous one day 🙂 or really need some details

i think it's fairly simple, especially since i don't need perfect performance

it scaling on distance to camera is probably the only other detail

[main cosmos]

yeah, it ain't very clamplicated

[obsidian wolf]

i sort of do the opposite rn

instead of blurring normals, i outline them

so hard edges like this are black instead of smooth

[main cosmos]

since u have a non-voxel game, can't you just explicitly model the bevel

[obsidian wolf]

voxels are certainly the best case for blurred normals, i don't think it would be as good in my case

[obsidian wolf]

effort

[main cosmos]

lel

[obsidian wolf]

bevels aren't just a modifier

there's a lot of topo maintenance that you have to do on them

[main cosmos]

I think outlines might look better for your game anyways

guess it's not too hard to whip up a normal blurring shader to see

[main cosmos]

I can confirm that teardown has a check to avoid blurring normals over depth discontinuities

the whole shader is basically 25 loc

[obsidian wolf]

ah shit

just realized that i'm probably going to have to have some extra shit on skeletons to use for gameplay far in the future

obviously its a bit of a dream to have to never edit skeletons

but knowing kills the dream 🥲

reimporting skeletons is very scary

[severe spruce]

i too am afraid of skeletons

[obsidian wolf]

💀

[obsidian wolf]

i need more indicators to show what's actually happening here

he's set up rn to attack in a diamond shape around him, and hits in a 3x3 area

[main cosmos]

didn't know those apples were 3D models

I thought they were 2D sprites lol

I think that speaks for the aesthetic of your game

[obsidian wolf]

https://sketchfab.com/3d-models/lowpoly-fruits-vegetables-d3be8fed96eb48be88b47bbe8d2951e1

spot em ;)

i actually have other spawners set up, i just haven't added the button to instance them yet lmao

which takes like 2 sec, 1 sec

[main cosmos]

I think one is the apple

lmao they have the dark banana

[obsidian wolf]

banana rot mechanic

my model loading is on god fucked

🙏

asset loading returns a reference now and modern memory management got me fucked up 💀

nvm? am i an idiot? apparently clearing my entity registry is somehow deleting a model?

[obsidian wolf]

ya im dumb

but also

i fucking hate templates, more specifically the error messages that they produce

one of the only things about c++ that legitimately bothers me

having to comb through 60 line, 1500 character error messages makes me feel like i'm living in the middle ages

like this being the simplified version

naw, fuck off

[main cosmos]

aw hell naw

btw check the output log for a better clue

[obsidian wolf]

the output log was worse

i know what the error was

[main cosmos]

it will look worse, but actually point at your loc that caused it

[obsidian wolf]

yeah, i usually just scan over both for a file, then for a line number if it's there

don't even bother trying to read it

my asset editor sucks

not really but the memory management of it sucks

pov: engine dev, you want to click a button you haven't used in a while

[obsidian wolf]

surely i wouldn't spend 90 minutes debugging something and making no progress

right

nvm false alarm i'm continuing being an idiot, but the 90 minutes did actually fix things

ok this only took like 30 seconds of actual asset editing

[obsidian wolf]

Saurian Sorcery

gameplay thought, fruits move too fast right now to attack them, but slower feels weird. i thought about having some sort of breakable obstacle that would spawn slowly in sequence from the thing that you're defending

basically like minions in league of legends, something for the fruits to get stuck on. this might come from allowing the fruits to attack the lizards though, which is planned soon, so it's just a note for now

[obsidian wolf]

i do not understand something

i have a r32 uint texture

fout_id = uvec4(fin.id, 0, 0, 0); (vout is the same as fin)
that i write to with nonzero data

but the value in the texture remains 0

i do not know all of the state that could be fucked up here

[obsidian wolf]

i am posting about this in questions since i am out of things to look at

[severe spruce]

so the turtle is a doctor

the eternal struggle between apples and doctors finally in a computer game

[obsidian wolf]

something something a day keeps the apples away? 🙂

the idea with him was for him to make things attack him, he looks very masochistic though 😄

[main cosmos]

are the apples zapping the turtle

[obsidian wolf]

yes

[main cosmos]

it's zapping time

[obsidian wolf]

he even dies

[cedar rune]

zapples

Why does he have two health bars

[obsidian wolf]

i like that

you talking about the yellow part?or the black line

[cedar rune]

OH its a 3D health bar

[obsidian wolf]

yeah

[cedar rune]

It looked a bit confusing but I see it now

Neat

[main cosmos]

ah

[obsidian wolf]

this is only slightly broken 🙃

though, i actually didn't get any crashes upon testing all of this for the first time, so this is still relatively successful

[main cosmos]

is that the boiling oil thrower

[obsidian wolf]

of some sort 😄

[obsidian wolf]

fixed da tings

looks better on 144hz! hmm...

thinking about if i care about that discrepancy

[main cosmos]

Tower suggestion

[obsidian wolf]

i have thought before about adding both octopus, and snails lmao

[obsidian wolf]

trying to figure out what this one will wear

and use for a weapon

[main cosmos]

that guy looks like an acid/thorn thrower

[obsidian wolf]

i think i should represent women at least somewhat so i think it's gonna be a woman xD

woman lizard

i think they need a belt of some sort or something though

dark souls tattered garb

[main cosmos]

put a bandolier strapped across her body

then give her a big gun

[obsidian wolf]

[obsidian wolf]

pretending to be a concept artist

[obsidian wolf]

oh btw

[main cosmos]

dominance: asserted

[obsidian wolf]

huh?

it was meant to be a thx for the idea :)

i wonder if that tone is not read now though 🤔

[main cosmos]

T-pose = dominance assertion

[obsidian wolf]

oooh

i forgor it was even like that 💀 im too used to it

i wonder now if character animators are more blind to t poses

[obsidian wolf]

this is looking good

[main cosmos]

twitchy

I think the idle animation is a bit intense, but maybe that's the low framerate of the video talking

[obsidian wolf]

no lol

its just bugged

[main cosmos]

oh mao

[obsidian wolf]

the jumpy frame is intended

[obsidian wolf]

multipart attacks

:wikid:

[main cosmos]

Honestly quite incredible

[obsidian wolf]

https://tenor.com/view/honestly-quite-incredible-frog-incredible-fascinating-gif-23673132

[obsidian wolf]

i'm sort of wondering about my gameplay code

case study would be, for the assassin, their attack was supposed to weaken units hit

so i have a struct for numerical stats

struct Stat {
    umap<???, StatEffect> effects;
    float value();
};
struct StatEffect {
    enum Type { Type_Base, Type_Multiply, Type_Add, Type_Override };
    // ... value, stacks, etc.
};

[main cosmos]

weaken in what way?

[obsidian wolf]

make them take more damage

and my abilities have callbacks at start casting, trigger, and end casting, where i manage stuff like cooldowns and what not, so clearly we need something to go in the trigger callback which adds a stat that's used for a damage multiplier

something like

void ability_callback() {
  // ...
  enemy.damage_taken_stat.effects[this_ability] = StatEffect(Type_Multiply, .2); // takes 20% more damage
}```

but it needs to have a limited duration, so we need a timer class, i already have one that i use for all my emitters in the existing gifs, and even the abilities use it

[main cosmos]

hmm I feel like Risk of Rain would be a good study on modifier mechanics

[obsidian wolf]

risk of rain has horrid source code

like it's dog shit awful

[main cosmos]

lmao rip

what about ror2

[obsidian wolf]

err i meant 2

idk about 1

[main cosmos]

ok same

[obsidian wolf]

2's is available because of modding

[main cosmos]

clearly it works though :^)

[obsidian wolf]

my timers work like this

        add_timer(scene, name, [](Timer* timer, void* data) {
            auto payload = (EmitterTimerTimeoutPayload*) data;
            payload->scene->registry.destroy(payload->entity);
        }, payload, allocated_payload = true, permanent_timer = false)```

[main cosmos]

I was kinda hoping they'd have a nice system for this, but I guess not

[obsidian wolf]

so i can just remove the effect if i have the key into the stat effect map when the timer triggers

there's two questions

do i ever need to do something to the timer other than just let it tick

and 2. does it really matter what the key into the stat effect map is

because my abilities have UUIDs (random u64s), so i could just use those, and even offset if i need to have multiple timers

really i should just continue with what i'm doing and not try to solve problems that don't exist

i.e. just add a timer and let it tick, if i ever find a reason to interrupt it, then worry about it then

and just use the ability UUID as the key for the stat effect map

[obsidian wolf]

they are tremendously OOP brained

like

man i don't even want to think about it, all i remember is when i tried adding a burn effect there were like 6 different "Burn" classes

[main cosmos]

Lmao you definitely have a better idea of how to do this than I do then

[obsidian wolf]

the reason why i don't really think it works in ror2's case

is because honestly, the development doesn't go that quick

like they aren't on the high end of item count, complexity, or anything like that

something like noita i'd mad respect the source code of even if i didn't understand it at first

[main cosmos]

RoR has a funny lizard character that reminds me of your game

Also the art style in general

[obsidian wolf]

the poison thingy?

[main cosmos]

Yeah

[obsidian wolf]

yuh

[main cosmos]

Poison boi

[obsidian wolf]

i like the game, just not the code

[main cosmos]

Wit da leap

[obsidian wolf]

acrid

[main cosmos]

I bet @urban lantern would have some good ideas about making an ability system

Heh, get teleported

[obsidian wolf]

for why?

[dry monolith]

was he not here already?

[obsidian wolf]

naw

[main cosmos]

idk anymore

https://tenor.com/view/iamgoingmentallyinsane-insane-crazy-cat-kitty-gif-21273974

[obsidian wolf]

deccer was not

they were in the thread before your projects and i think joined for a bit at the start

[urban lantern]

no clue what risk of rain is

[main cosmos]

https://store.steampowered.com/app/632360/Risk_of_Rain_2/

Steam

Risk of Rain 2

SURVIVE AN ALIEN PLANETOver a dozen handcrafted locales await, each packed with challenging monsters and enormous bosses that oppose your continued existence. Fight your way to the final boss and escape or continue your run indefinitely to see just how long you can survive. A unique scaling system means both you and your foes limitlessly increas...

Price

$24.99

Recommendations

153803

Metacritic

85

▶ Play video

[urban lantern]

hmm ive seen that somewhere some time ago

[obsidian wolf]

i been tryin to figure out how to add an ability to lizards that isn't just their basic attacking

there can be

• just a button on a UI that you click
  -> with a cooldown?
  -> driven by mana that you get from attacking / being attacked?
  -> available on kill?
• automatically with the above conditions
• the one that i'm most fond of currently, a special tile in every area that you can place a lizard on to make them "cast"

i also probably want the enemies, or at least special ones, to have the same ability and casting sytem, so if it was player controlled, then it would be some AI heuristic

and if it was a tile, then there would be a tile for the fruits and for the lizards

i think i'm gonna start with the tile idea

since it's the most unique, and the others will be a fallback if i don't like it for some reason

[obsidian wolf]

spent a bit making a model for it lmao

also fixed a lot of the shit normals on the models

still not perfect lmao

the tiles are very visible 😖

[main cosmos]

depending on how cluttered you expect the map to get, the "manual ability trigger" could become too much micro to effectively use

however, it can work if you don't expect the player to have more than a handful of towers at a time

[obsidian wolf]

i'm a little micro freak

[main cosmos]

(i.e., not bloons)

idk if you've played the company of heroes series, but it's great in terms of micro and per-unit abilities

[obsidian wolf]

nah i haven't

[main cosmos]

basically you only control like 6 squads (no individual unit control) at max, and some of them have manually-activated abilities like grenades and such

[obsidian wolf]

the main reason i didn't like sc2 was the build order geekiness

[main cosmos]

but it's pretty manageable, partly because abilities cost resources

[main cosmos]

yeah that game has too much emphasis on mechanical skill imo

[obsidian wolf]

im sayin the opposite

[main cosmos]

CoH places more emphasis on individual unit micro

[obsidian wolf]

wait what's the difference

[main cosmos]

SC: micro an army and an economy
CoH: micro a few squads

[obsidian wolf]

oh

[main cosmos]

it means you give a lot more attention to individual units compared to SC

[obsidian wolf]

i've always imagined a game where you control like 3 league of legends champions at the same time doing slightly different task as both

entirely unexplored and super fun

it sounds at least in that direction, so i'm interested to try it

i spent like 2 years of my life thinking about what parts of it people potentially wouldn't enjoy and how to avoid that

le classico wanting to make a super game

[main cosmos]

if you make it an mmo it will inevitably be a success

[obsidian wolf]

truly, i'm integrating networking as we speak

googling "how to make an MMO with RPC networking protocol"

[dry monolith]

MMO with HTTP

[obsidian wolf]

@main cosmos'

oops didn't mean to send but is COH1 worth it?

or only COH3

[main cosmos]

Haven't played coh3 yet but it seems good

Coh1 and coh2 are both worth it, though maybe coh1 is better for single player

Coh2 has steam workshop support which is quite nice

And matchmaking if you want to do PvP or PvAI

[obsidian wolf]

no interest in pvp

i saw some reddit comments saying that coh2 just had a couple more problems

[main cosmos]

Both are solid choices tbh

But I think the community content gives coh2 an edge overall (coh1 has it too, but it's harder to install mods)

[obsidian wolf]

i would just get 3 but 60$ lol

reminds me of how at my company all of the execs say 70$ instead of 60$ now

what a splash

[urban lantern]

@obsidian wolf can you show @grave nacelle your animation editor thing again? they are too shy to ask 😄

[grave nacelle]

lol yes i would like to see

[obsidian wolf]

skeleton editing?

[grave nacelle]

the graph part

[obsidian wolf]

hmmm

i'm still not sure which part, i have two things, one for sequencing poses, and another for posing skeletons

when i import a model, the keyframes in the animation all get put into the "Extras" category

[grave nacelle]

the part in which you can parametrize animations to use them

variables state

and etc

[obsidian wolf]

and i manually move them to the pose category by dragging or doing the checkbox + migrate button

the timing in the pose categories is time to interpolate to it

it's very hardcoded, but it's designed specifically for my game, because i don't have enough time to make or use a complex animation system

can also preview them to mess around with timing. one thing i wish i did here was differentiate between cycling and non-cycling states

there's no graph though, see below for reasoning 😄
-> because i don't have enough time to make or use a complex animation system

i also used to be able to edit the poses in engine, but i haven't reenabled this because right now i'm just importing everything from blender

the animation system is mostly from this
https://www.youtube.com/watch?v=LNidsMesxSE
so if you're not using it or something similar, it's pretty hard to reference this stuff honestly

this is what the results end up looking like for ~10-12 poses per character, they're not perfect but about as well as i can do without it being my job xD

if this isn't what you were lookin for, that's ok, its still fun to talk about 🙂

[grave nacelle]

no it wasnt, i was looking for a graph editor for animation but it interesting anyway

[obsidian wolf]

yeah, i wanted to do that too, but decided just to hard code my graph into code

limiting keyframes to poses allows me to treat state transitions as just pose interpolation

[obsidian wolf]

i have seemingly learned that chameleons are the only lizards that catch bugs with their tongue

and now i have to decide whether i care

chameleons are so neat why do they get so many unique features

[main cosmos]

They got the p2w edition of life

[obsidian wolf]

truly

shilling vid https://www.youtube.com/watch?v=KJtaIqahi3I

[spring sentinel]

Ew

[dry monolith]

yes indeed, better use http

[spring sentinel]

No

Too many game engines use crap like RPCs and replication

Those don’t scale very well

You need to go low level with your netcode and pack your bits into your packets manually, counting them accordingly. You need to set up your protocol to be resilient to lost packets and not halt the simulation just because delivery wasn’t ACK’d.

Bandwidth is going to be a concern with any MMO. Don’t put in an abstraction layer. Every byte matters.

[obsidian wolf]

tbh i sort of do want to make something multiplayer at some point, but certainly not an mmo lmao

[spring sentinel]

First problem is that you’re trying to slap networking on it after the fact

[obsidian wolf]

what

[spring sentinel]

Usually you want to start netcode early on and design a game around it

[obsidian wolf]

i dont want to make this multiplayer

[spring sentinel]

Oh

[obsidian wolf]

yeah i know how bad of an idea that is

[obsidian wolf]

heavy WIP still but damn this was hard to make

need hands + feet, texturing (beyond just gradients), and the clothes was just testing so they're poorly made, but i think it looks not shit anymore?

i get the feeling that chameleon characters are generally avoided because of how weird the proportions are on those creatures

btw this is why i havent done the hands and feet

[severe spruce]

have you considered making this character inflict sleep

a coma chameleon if you will

[obsidian wolf]

🙂

the unironic drive of it was that i wanted someone that would grab fruits with his tongue

and apparently chameleons are the only lizards that grab distant things with their tongues

[urban lantern]

that looks pretty cool

[obsidian wolf]

holy muck, i just wrote a bunch of gltf writing code and it worked on the first time

i'm a freaking professional

[spring sentinel]

Can it save out @urban lantern’s cube?

[obsidian wolf]

the round trip deccer's cube test

[spring sentinel]

Ya

[obsidian wolf]

no 😄 i could, but hierarchies are extra work and i dont need them

trying to automate my WCF tileset

[spring sentinel]

What’s with the wasted space

[obsidian wolf]

so i can make more tilesets, but less manually

[obsidian wolf]

my editor for the tilesets exposes some dots to press to show what gameplay tile they represent

[spring sentinel]

Attach this to my server

It supports tile-based games

🙂

[obsidian wolf]

pretty sure that's a networking task, assign it to the server folks (you) 🙂

[spring sentinel]

Oh just use my API

Have a collision map on the server. Pass in the target tile coordinate you wanna move toward on the next tick and a you’ll get a stream of responses on what tile your character should be standing on every game tick back.

Just interpolate for rendering in between with whatever animations you like.

[obsidian wolf]

for saurian sorcery 2 ;)

in approximately 50 years

[spring sentinel]

Ha

[main cosmos]

new mechanic for saurian sorcery 3

[obsidian wolf]

for the merch

tho that's actually a very neat idea

if only i could disassociate from the "japan" meme

[main cosmos]

Reminds me of this incredible meme

[obsidian wolf]

lmao

[obsidian wolf]

well i have all of the edges now

unfortunately 490 is a lot of meshes to make

so i'm going to try to generate faces instead of just edges

but idk how yet

[obsidian wolf]

i posted about this in the questions forum because it's a super niche problem, so visibility is good 🙂 if you have any ideas, pls gib

[obsidian wolf]

abilities :^)

[main cosmos]

I just noticed that the health bars have shadows lol

[obsidian wolf]

hehe, everything does, my materials don't have an option to exclude

(in this it's the range preview)

[urban lantern]

that looks super cool already

i also like the style of the whole thing

[obsidian wolf]

i'm sort of thinking about having gameplay mechanics not reference entities at all

so for example, the buff ability above would only buff a specific tile, not the orange guy that casts it

probably more effort than it's worth though

the current restriction is that abilities can only target squares, but their effects can then be attached to units

[cedar rune]

do your guys move

[obsidian wolf]

moving units from tile to tile costs money

[cedar rune]

I see

So that wouldnt transfer the buff

[obsidian wolf]

right, but you could buff a square, then swap units and transfer the buff

[cedar rune]

Thats interesting

[obsidian wolf]

so there's basic attacks, and an ability that placing units onto this platform casts

and the way i'd have to set up self-buff abilities if i did it that way, is that it'd probably buff the last square that the unit was on, then teleport them back to it

[obsidian wolf]

finally made a tree

[urban lantern]

you have a hand for those kind of things

[main cosmos]

and a head

[severe spruce]

and my axe

[obsidian wolf]

meep

not perfect, main issue is the arrow blending back into the quiver (which i have a fix for, it's just bugged), and we'll see how spawning the actual arrow projectile goes, but i like this 🙂

the cast is supposed to be him throwing something, perhaps a large bear trap or a smoke grenade, not sure yet

[obsidian wolf]

i just realized they alternate hand sides in the idle animation

[obsidian wolf]

damn it's really crazy being able to reimport an animated mesh and not lose any data

takes like 3 minutes to fix that

[obsidian wolf]

nevermind i should not speak

[main cosmos]

Horrors within human comprehension

[obsidian wolf]

i have a teeny tiny problem, my ability casting works by having an ability targeting function, where when a target is found it casts a spell
works well if a spell is instant, but the more delay it has, the more likely it is to miss

it happens near the end of the vid above

there's two (and a half) ways i could fix this that come to mind

1. instead of using the fruits current position, i have their pathing data, i could approximate how long it'd take for the ability to land (either pre-trigger casting time, or pre-trigger + travel time), and estimate where they'd be at that point in time

• should be fairly accurate
• fairly representative of how targeting actually works

• can get kind of complicated, especially as abilities and effects grow in complexity (e.g. a unit has a move speed buff for 0.5 more seconds.. do i account for it if the ability will take 1.5 seconds to land?)

2. i could update targeting at trigger-time (e.g. right when the arrow releases in the case above)

• would be more accurate than currently is
• very simple

• lizards usually rotate to face their target, and turning to face it is usually done during the precast time, so this would mean they could cast in a direction they aren't facing
• only mediocre accuracy

and the extra half option is to give lizards an "intelligence" stat, where i implement a tiered targeting AI system, and it selects the targeting behavior based off of it...
big raging crocodile? constantly missing attacks, but maybe the old wise wizard gives an aura that can help him out with that
old guy using a bow? knows the perfect square to target to maximize damage

• unique to my knowledge
• creates a lever to balance some things around
  ? probably not very intuitive (especially for minmaxers)

• hardest to implement

[main cosmos]

Btw what age of empires 4 does is make projectiles curve towards the target (it seems like they follow some sort of parametric curve)

[obsidian wolf]

melee characters have this problem too, but #2 + that would work

its just a little odd because targeting is per-grid square only, and so that'd be a very large shift especially if not near the start of projectile life

[spring sentinel]

These are some of the same challenges missile guidance systems need to resolve.

[main cosmos]

but in a video game you can just cheat and make t=1 be the location of the target

[obsidian wolf]

instead of all dat im gonna do water this week or weekend i think

hardest part will be that i probably want to do reflections 💀

and i have no clue how, i'm guessing i could do SS fairly easily, but i might just render twice from the camera flipped across the water plane lmao

[main cosmos]

rt

[obsidian wolf]

literally the only reason i'm not going to intergrate any hardware RT is renderdoc

and software RT is too much work

[main cosmos]

But you'll still have Nsight, RGP, and RRA

[obsidian wolf]

renderdoc my beloved

[main cosmos]

bababooey

[obsidian wolf]

but i might just render twice from the camera flipped across the water plane lmao
i'm probably gonna do this xD

hmm wait does that even work

if the water plane isn't completely flat, idk how people distort it

i should probably just google around

[main cosmos]

Well it works for flat planes

UE has planar reflections as an option

[obsidian wolf]

i need at least a lil distortion, it'd probably be minimal tho

[main cosmos]

you can probably cook up some fake distortion to look like ripples or whatever

just by sampling the reflections with an offset

[obsidian wolf]

ye

i don't even think i draw a mesh for it

i also wanted to add a lizard bench for lizards, similar to TFT

mostly so they have a place to respawn and so the shop is less messy

not ugly water and that are soon tm 🙂

[dry monolith]

also need to discard fragments above the plane, historically there were additional clipping planes in gapis for that, idk how it's done in modern days, could be manual discard calls

and pretty much you can't do planar reflections for curved surfaces

that's where probes begin to shine (very dimly and with artifacts)

[obsidian wolf]

hmmm

so SSR

[dry monolith]

adding normal maps and shifting uvs with them can simulate ripples somewhat plausibly but will expose parts where geometry cuts off at the waterline

[obsidian wolf]

the waterline will be masked off by foam

[dry monolith]

SSR is just bad, okay with probe fallback, but still meh

[obsidian wolf]

all of the references i liked for water are basically flat

i'm not doing no sea of thieves water, so maybe i can just do the planar hack and hide the artifacts

[dry monolith]

that's how it was done in 200X-s

e.g. half life 2

[dry monolith]

yeah foam works good

[obsidian wolf]

i basically just need a little movement so it isn't a static frame

[dry monolith]

can also mask away the ripple perturbations near the shore

[main cosmos]

a e s t h e t i c

[obsidian wolf]

what's the algo then? just march against planar render g buffer in reflection direction?

[dry monolith]

no just render using reflected camera and clip geometry above the water

[obsidian wolf]

yeah but how do offset

adding normal maps and shifting uvs

[dry monolith]

just offset literally

read normal, add normal*parallax_factor to uv

[main cosmos]

just add water one

[obsidian wolf]

oh

hmmm, i'm struggling to grok it, is depth not a factor?

i guess not

[dry monolith]

it'll look fine as far as I remember

[obsidian wolf]

well no point to overcomplicate it then unless i run into problems

i'll just try that

[obsidian wolf]

mobile game aesthetic lmao

[dry monolith]

https://github.com/SimonDarksideJ/XNAGameStudio/wiki/Riemers3DXNA4advterrainoverview
https://github.com/SimonDarksideJ/XNAGameStudio/wiki/Riemers3DXNA4advterrain11ripples

this is a github page that revives an ancient tutorial for XNA and D3D9 which has some pages on how to do planar water with reflections and refractions using method I described

I found the method there btw

https://web.archive.org/web/20071116032551/http://www.riemers.net/eng/Tutorials/XNA/Csharp/series4.php

or you can directly use the wayback machine if you prefer

[obsidian wolf]

hmm, nice

[obsidian wolf]

i may just

ignore reflections for a while lmao

add some noise waves with this and call it a day for now, i consider trying to experiment with hue shifts in the water to make things fancy

[main cosmos]

I hab idea

you can use the foam texture with some kind of noise

so it'd be noise that's only visible where the texture isn't black (i.e., in areas where it currently is)

[obsidian wolf]

are you just saying add the current of the image and the white parts of this?

[main cosmos]

uh I mean maybe multiplying the "foam density" (the texture you generate now) by noise

[obsidian wolf]

it'll probably be an add but yeah

im not sure

i want to try a lot of things because it's kind of random what looks good

[main cosmos]

the "adding" thing would be something else

like global foam that doesn't just affect the shore

what I'm talking about is specifically the shore foam, trying to make it more natural looking

[obsidian wolf]

i'll smoothstep after the add

so it'll affect it

perhaps not enough

[obsidian wolf]

this is a random debug vis, but it looks kinda neat

[obsidian wolf]

ok

my experiments with hue shifting look really bad

it might look better with gradients on the tile set, which i've considered doing

[severe spruce]

bruh this is looking better every day

[obsidian wolf]

glad it's not looking worse every day

[severe spruce]

we'd still have some time until it would look bad

[obsidian wolf]

ok i'm satisfied

[main cosmos]

Mmm worter

Btw another feature you could add is a fresnel effect where the water subtly changes color with the viewing angle

[obsidian wolf]

i might do that

but it'll have to be based off the water plane i think to avoid it being too detailed amongst the stylized look

[main cosmos]

what I did back when I fiddled with water was make it turn slightly greener as you looked more directly at it. It was a very low frequency effect

If you ever add reflections then you can add proper fresnel and then it'll look perfect

[obsidian wolf]

this is a neat screenshot

[main cosmos]

what's that

looks like someone testing different water shaders

[obsidian wolf]

that's what i get out of it

[obsidian wolf]

i do like it, i'm exaggerating the effect here to evaluate it, but when it's subtle it looks really nice

i'm not so sure the right one is better though

[main cosmos]

I think it looks more plausible

[obsidian wolf]

i haven't found many good references

[main cosmos]

some random pic

that's a very low view though

[main cosmos]

I just googled "caribbean water"

[obsidian wolf]

a big problem with references is that that the sky and shallow reflections play a huge role

another one is that most real references are on beaches

[main cosmos]

yeah the depth has a big effect on the greenness of the water

[obsidian wolf]

a lot that i see look darker at more direct angles purely because there is less reflection

[main cosmos]

that's if the water is very deep