vinnyvicious

Thank you! Those videos are really really good.

Does anyone recommend any good books on renderer architecture? One of the hardest things I face when writing renderers is making it not look like a big procedure. I have shadows, SSAO, PBR and I have tons of shaders and my renderer loop is huge. I'm hardcoding all the inputs and outputs. I never needed a scene graph or ECS because I'm working on simple and small games, but I wonder how one would organize the renderer in larger projects? Or even making the renderer more abstract so I can stop copying/pasting and tweaking stuff for every new game I make.

I've been watching some GDC videos and I was intrigued by the techniques demonstrated by Overwatch & Destiny: https://www.youtube.com/watch?time_continue=1196&v=7t0hLZd_8Z4 https://www.youtube.com/watch?time_continue=786&v=CXFLu8cityA On those marks, you'll see the animators explaining how they created pre-made poses for aim & roll situations. However, it's not clear how this is integrated in the engine itself. Does this mean that for each weapon, there are animation clips called lean_left, lean_right, lean_up, lean_down and those are played when the camera code detects one of those motions? What about movement forward and backward? Has anyone implemented something like this and could share how they did it? Thanks a lot guys! ps: My current weapon camera code looks a lot like the Destiny demonstration: a static model attached to the camera and feels like shit.

I was looking at FTE's GLSL renderer for ideas: https://github.com/mdeguzis/ftequake/blob/a83b8ca825255c343780ffc4fe99d96e9e9bce0c/engine/shaders/glsl/defaultwall.glsl#L97 They emulate Q1 software pretty well. But it's not clear to me how the ramp actually works. Do I have to average each pixel and see which one is closer to the colors available in my palette? Isn't that terribly slow?

I'm quite a noob when it comes to pixel manipulation shaders, so you will probably understand my fascination with this tool, even though it will look simple for most of you: https://kronbits.itch.io/pixatool I've been trying to create something similar inside Urho3D for the purpose of giving a retro-look to my particles, but I had no luck with the palletization. Does anyone have any sample shaders that change a texture to match a palette? Is it easier to do on a per-texture basis, or doing a full-screen shader?

https://github.com/SonyWWS/ATF

Sharing some interesting experiments with recoil: https://www.youtube.com/watch?v=E_Aj7Q8R2LA Sadly, animation is not affected in any way.

@Scouting Ninja, by using on the model you mean applying the formula on top of delta time and model transforms? Do you use different formulas for rotation, or you just apply on translation? Do you do any IK too? I'm not trying to emulate the real world, just apply a good "feel" to the weapons without having to animate them individually. I want to procedurally change some generic variables and have a set of those for each weapon in the game. It's not a realistic game at all (arena shooter). I'm just trying to understand how I can use easing formulas and other techniques to juice the whole animation.

I was reading CryEngine's codebase, and specifically, the procedural motion bits in the weapon module: https://github.com/CRYTEK/CRYENGINE/blob/26524289c15a660965a447dcb22628643917c820/Code/GameSDK/GameDll/Recoil.cpp Their recoil implementation is very interesting. It seems to be some sort of spring, applying the effect on the world position of the weapon model (and arms). No IK, no bones, Seems to be the whole object. Does anyone have any material on this subject? Or is this very unique? How everyone else is doing this? I'm very curious about this kind of thing. They also seem to add a jump bump motion procedurally, as well as sway and movement, but I still haven't found those pieces of code. I'm suspecting a classic Lissajous curve.

Does anyone recommend a minimal OpenGL + Bullet scene example? Something simple, like rolling a ball (marble blast style), but that covers most implementation caveats and details.

A freelance company produced tons of additional content for our project, but unfortunately, although the bone structure is the same as ours, they #*@!ed up on the naming convention. Now I have a bunch of models that I need to rename 30+ bones. Does anyone know any quick way of doing this? A Blender script, an Assimp tool, anything?

ping?

I've just finished most tutorial steps on learnopengl.com and I'm really excited about starting to write very small model viewer. Unfortunately, I have no idea how to combine all the shaders in the examples that were given. I mean: how can I do the shadow mapping shader also work with the diffuse/normal one? Is that what render passes are? How is this usually structured in a rendering class?

I'm not really worried about learning curves, or learning a new system. That's not really relevant and usually is fun. What really matters here is that if there's is good money to be grabbed. Does anyone know any public revenue reports from developers on Kongregate or other portals? Are we talking about few cents a day or more Flash-like numbers like 1 or 2 dollars a day, from ads?

If i blend Left and Forward, for example, how can i associate the blending amount from A to B based on the direction vector? As for mirroring, why would i need mirror bones? Maybe just invert the values?