next »

[JWki]

I'll get back to it soon enough! c: Did a jam recently and exhausted myself. Considering doing some prototyping in Unity for the game I'll be making with my own engine in the end.

Looks good! Is that the brand new colour picker that Omar tweeted about, or one of the third-party ones?

Thanks, it's the new one that Omar tweeted about. I did this pretty much the instant he merged it. I REALLY really like how it turned out, it pushed my urge to roll my own UI lib back a bit.

[oahda]

Sounds good! c:




Here's the result of today's prototyping! I basically wanted something like Zelda's timeshift stones but in a bubble, and where things don't just pop in and out, but can be cut in half by the bubble.

Of course there can be things in the red area that don't show up in the blue bubble too:

[JWki]

Nice! Stencil masking or something completely different?

Also I can now paint with textures:




Just need to extend this to paint like a full material texture set at once into multiple targets and tadaa, can do full material layering.

[ferreiradaselva]

3 days without programming. That's why I'm happy, yall.

[oahda]

Nice! Stencil masking or something completely different?

Tried that at first, but didn't turn out that well. Anything that was visible through the bubble from the camera's perspective would get rendered no matter the distance from the camera. Tried to fix it with depth buffer trickery and stuff but it soon got very convoluted. Now I'm just checking if the fragment is within the radius of the sphere instead.

Also I can now paint with textures: Just need to extend this to paint like a full material texture set at once into multiple targets and tadaa, can do full material layering.

Cool! I've never implemented a painter, but I've done similar things in shaders anyway, so I think I understand how it works. E:

3 days without programming. That's why I'm happy, yall.

Me too, really, if not more. c: Lots of Netflix.

[bateleur]

Now I'm just checking if the fragment is within the radius of the sphere instead.

You also seem to be doing something slightly cleverer to get the white faces where the sphere cuts a "blue" object in the foreground or a red one in the background? Part of it is just rendering as white any fragment very close to the radius, but how do you get white pixels in places where there isn't a triangle at all? Or do all the meshes have double-sided geometry with a different material on the inside?

[oahda]

Now I'm just checking if the fragment is within the radius of the sphere instead.

You also seem to be doing something slightly cleverer to get the white faces where the sphere cuts a "blue" object in the foreground or a red one in the background? Part of it is just rendering as white any fragment very close to the radius, but how do you get white pixels in places where there isn't a triangle at all? Or do all the meshes have double-sided geometry with a different material on the inside?

It's unfortunately down to actually rendering the backfaces, yeah (but I don't think it's too much of a problem because only select objects are going to be governed by the bubble, and only a few at a time, and I could check first if it's entirely inside of the bubble and if so enable culling again) but I'm just checking the normals relative to the view angle in order to identify backfaces (if dot product is negative in my case) and change the colour in shader instead of having manually applied a different material.

[gimymblert]

Sounds good! c:




Here's the result of today's prototyping! I basically wanted something like Zelda's timeshift stones but in a bubble, and where things don't just pop in and out, but can be cut in half by the bubble.

Of course there can be things in the red area that don't show up in the blue bubble too:

Perfection 

[JWki]

Someone else in another forum made a nice frosting effect for dear imgui so I tried getting a similar effect - didn't take very long and looks nice:

[oahda]

Perfection 

<'3

Someone else in another forum made a nice frosting effect for dear imgui so I tried getting a similar effect - didn't take very long and looks nice:

Oooh, bringing dat Aero back!

[JWki]

Here's something exciting: Loading meshes from FBX.



Took me a while to mess around with all sorts of transformations for submeshes, eventually got them right I think.

[JWki]

Metallic surfaces look pretty amazing with a cook torrance based shading model, even with very simple image based lighting that's not using proper convolution yet (atm just sampling the cubemap multiplied with metallic which is super incorrect).

[Schrompf]

Still looks awesome IMHO. You could probably emulate the roughness with a simple texCubeLod(). Bonus for pre-convoluted mipmaps of said cubemap.

[oahda]

Cool stuff! I'm restructuring my stuff to get a better build system, so I can't work on anything else at all until I finish that.

[JWki]

Still looks awesome IMHO. You could probably emulate the roughness with a simple texCubeLod(). Bonus for pre-convoluted mipmaps of said cubemap.

Yeah convolution into mipmaps is the next thing to do, both for the diffuse part and the specular part.

Then I can work on build systems too again.

[gimymblert]

Still looks awesome IMHO. You could probably emulate the roughness with a simple texCubeLod(). Bonus for pre-convoluted mipmaps of said cubemap.

There was a paper to do that more or less correctly by biasing the sampling directly (no convolution) the lod by the BRDF, If anyone where I can find it back. ALL I find now is real time convolution using importance sampling which is the whole deal (analyzing and baking convolution into the mipmap) and not just a LOD sampling.

[JWki]

Still looks awesome IMHO. You could probably emulate the roughness with a simple texCubeLod(). Bonus for pre-convoluted mipmaps of said cubemap.

There was a paper to do that more or less correctly by biasing the sampling directly (no convolution) the lod by the BRDF, If anyone where I can find it back. ALL I find now is real time convolution using importance sampling which is the whole deal (analyzing and baking convolution into the mipmap) and not just a LOD sampling.

You can translate the roughness into a lod level if that's what you mean - you do that anyways even with convolution because you usually store the different roughness levels in mip levels, but even with just regular mipmaps you get a blurrier -> "rougher" reflection with higher roughness levels if you do that.

[gimymblert]

Yeah I know but the convolution is expensive to compute RealTime™ to stay Physically plausible, and then if you just linearly sample the LOD you don't have the right lighting response, which can be odd in varying lighting condition (mismatch in reflected energy), the paper was gear for real time refreshed cube map (think moving car reflecting the environement), they just bias the mipmap sampling (since you couldn't convolve in this condition) with the brdf to get a good enough approximation (though ther was still some energy mismatch in some condition).

[JWki]

Yeah I know but the convolution is expensive to compute RealTime™ to stay Physically plausible, and then if you just linearly sample the LOD you don't have the right lighting response, which can be odd in varying lighting condition (mismatch in reflected energy), the paper was gear for real time refreshed cube map (think moving car reflecting the environement), they just bias the mipmap sampling (since you couldn't convolve in this condition) with the brdf to get a good enough approximation (though ther was still some energy mismatch in some condition).

Oh okay I think I know what you mean now.
I haven't seen the paper though so yeah that sounds interesting.

[JWki]

Learned about the fact today that in C++ you can take pointers to specific overloads of functions, apparently.
This is nice because it allows for this:

Code:

namespace myApi
{
    void Foo(int);
    void Foo(float);
    
    struct Interface
    {
         void (*FooInt)(int);
         void(*FooFloat)(float);
    }
}

EXPORT 
void my_api_get_interface(myApi::Interface* interface)
{
    interface->FooInt = static_cast<void(*)(int)>(&Foo);
    interface->FooFloat = static_cast<void(*)(float)>(&Foo);
}

which is useful for dynamically loaded modules and ABI interfaces in general - the implementing module can still work with the overloads so only outside binaries have to deal with the additional verbosity.