Webgpu

I have put a pin in the Renderer for now, it’s in a good spot do that I feel. 📌Bonus Content! Hello, me again. I know I only posted this yesterday and I’m aware of what the previous sentence says, but… earlier today I opened this post on my desktop and noticed how aliased things were (the retina display makes it less obvious on my Macbook). I had disabled MSAA when I was working on SSGI because there were a lot of resources that needed to be multi-sampled. For a deferred renderer, other techniques for anti-aliasing are better such as Temporal Anti-Aliasing (TAA) and AI boosted methods like Deep Learning Anti-Aliasing (DLAA). The latter isn’t something available to me (yet). ...

This post is going to be dense. It’s the summation of the last month or so of work on StaticRectangle (sans PBR materials covered in my last post). Let’s start with the results and discuss how we got there. Pop Out h toggles the Controls ~ opens the terminal Mouse look with left/right mouse buttons w, a, s, d to move Space Bar fly-up, Left Ctrl fly-down Change the Render Pass in Render Options Enable Show Lights in the Viewport to select Lights to tweak properties etc. Select objects to change properties Foward vs Deferred Rendering To properly implement indirect lighting (and the other screenspace effects I had bodged til now), I needed to move away from a primarily forward renderer, to a hybrid-deferred one. ...

I’ve been quite busy the last couple of weeks, the other day a popup on my MacBook alerted me to an increase of 46% in screen-time. In my last devlog I had turned my sights from shadows to shaders and had implemented a Blinn shader. The next logical step was PBR materials. Change Log Static Rectangle V.10 ✦Implemented PBR material ✦Implemented EnvLight and HDR image-based lighting [1] ✦Added an optional 'skybox' render pass ✦Added a final composite pass for tone-mapping etc. ✕Removed AmbientLight (superceded by EnvLight) [2] ~Refactored lights to descriptor-based approach ✦Implemented screenspace reflections ✕Fixed race conditions in pipeline creation of the LineRenderer ✦Improved materials to support UV properties (scale, offset, tiling etc.) ~Refactored UI controls [1] I’ll briefly touch on this but it really deserves it’s own post which I’ll do next time. Same goes for Screenspace Reflections (SSR) [2] I had only just added this when I was working on the Blinn shader. I also just finished refactoring the Light too, these things happen. ...

So shadows… What?! Shadows again? Yep, I’m sure you’re getting tired of hearing about it, I’m getting tired of thinking about them… But here is what I’ve done in the last few weeks. Static Rectangle V.09 ~Decoupled the filter radius from the SpotLight penumbra [1] ~Optimised shadow sampling by adding early out [2] ✦Increased sample count for shadow edges from 16 to 32 ✦Implemented "Randomly Rotated" poisson disk sampling to address some banding/muddy shadows [3] ✕Fixed pervasive SSAO issue where samples were flipping at face boundaries [4] ✦Added a simple AmbientLight that supports just colour and intensity ~Refactored Light class to support the AmbientLight ✦Added a ShadowCasterLight extension that both SpotLight and DirectionalLight extend ✦Added a Shader DAG node to the SceneGraph ✦Added the Picker pass for rendering selectable objects for user interaction ✦Added Outline renderer that renders outlines around selected objects ✦Added SelectionManager to manage selection state [1] I moved shadow softness to an explicit property on the light itself. In retrospect, this wasn’t the good idea I thought it was, it meant lights without penumbras could only have hard shadows without coming up with some other solution. Dumb. [2] I now do 4 samples first to check if the fragment is in shadow. Most fragments are either fully lit or fully in shadow, so those now only need 4 samples instead of 16. [3] The first fix was randomly rotating the sample kernel, which helped but introduced noise. So I swapped to using a blue noise (ironic) texture to drive the rotation instead. [4] The samples flipping directions at face boundaries was causing hard edges/obvious faceting. It was most obvious on curved objects and I had noticed it on the AT-ST model but shrugged it off. When I dropped a sphere into the scene I couldn’t ignore it. Shader Time After putting shadows to bed (for now…) I turned my attention to shaders. ...

I’m at a point where I’m doing a lot of work, but the progress seems minimal. This happens often and is not at all unusual, especially projects like this that develop somewhat organically. You’ll often find yourself working on a component only to rework it when you look at an adjacent component or even the same component with fresh eyes. While this isn’t unusual, it does make blogging work in progress difficult. I’ll either have non-functional code or some simple test scene which isn’t interesting or difficult to explain out of context. ...

Life got in my way a wee bit lately (a trip back to New Zealand, a Windows 11 update bricking an SSD and finishing up the work year to name a few recent events), though I have been tapping away at StaticRectangle. SSAO I decided to tackle SSAO (screen space ambient occlusion) after reading through the learnopengl SSAO article, it didn’t seem like too much extra work on top of what I had just completed for shadows. ...

I struggled getting shadows working, there was a real moment where I wanted to cut my losses and give up. It was particularly frustrating because shadow mapping itself isn’t a difficult concept to comprehend. You render a depth map from your shadow-casting light’s perspective, capturing what the light sees–and equally importantly, what it doesn’t. Then, while rendering pixels (fragments technically) during the main render pass, you check if that pixel falls within the light’s bounds (based on clipping, range, falloff, etc.). If it does, you transform that fragment’s world position into the light’s space and sample the shadow map. You compare the fragment’s depth (from the light’s perspective) against the stored depth in the shadow map–if the stored depth is less, something closer to the light is blocking it, so that pixel is in shadow. ...

OK, so drawing lines is hard. Well maybe not hard per se, but certainly tedious. Looking into it (starting with Matt DesLauriers’ infamous ‘drawing lines is hard’ article linked above), I had the general gist. To draw the lines I want, the way I want, I need to segment them and render triangles. Like this, kind of… No real surprise there, honestly. Your GPU is a super-powered triangle crunching machine. While line primatives (line-list, line-strip) do exist, they’re not very useful–as far as I can tell. They’re typically aliased, hardware dependent and 1px wide. ...

Before we move on, in my last post I took a detour from the lighting/texturing pipeline to fix some issues in my Transform implementation. The debug scene was helpful for finding and fixing issues with transformations and it inadvertently highlighted a different issue. Multi-Sample Anti-Aliasing Without a technique known as ‘anti-aliasing’ edges can look jagged/stair-stepped because each pixel is either fully inside or fully outside the triangle you’re rendering with no in-between. ...

Far from the first and unlikely the last, I’ve taken a slight development detour. Firstly, I have migrated this dev blog from WordPress hosted by my web host to this statically generated one hosted on GitHub. I won’t bang on about it, but I was finding WordPress quite cumbersome and exceptionally slow—not just to load on the client side, but to do anything with it. It made the blogging process quite frustrating for me. ...