Yeah, looks like they are doing weird stuff to load the actual content, not sure what that is about, maybe loading stuff dynamically from GitHub or something?
Looking at the source of the page, they are using some JS library called markdeep that seems to render the content, so if you don't have JS activated, it won't work. Similarly, overzealous or misconfigured adblockers would probably make it not work as well.
Well, I myself use adblockers for all my browsers and I sometimes add my own rules that blocks more than I actually wanted. Are you saying it's impossible to misconfigure the blocklists in adblockers?
Just trying to help you out with figuring out what's wrong. If you don't want help or to fix the issue, maybe refrain from posting next time.
AMD's RDNA 2 is expected to have hardware ray tracing acceleration. Do you think that Vulkan will adopt support for it and everything in this guide will be identical (ie, relatively hardware independent)?
Shameless plug: We created a simple game engine with this technology. I am planning to upgrade it to cross-vendor ray tracing support with the upcoming Vulkan SDK. Cheers.
No, I haven't thought about it, but it makes sense to also provide a video. I'll probably do that tomorrow and link the video in the README.
The engine comes with a tiny demo, mainly showing off reflection and refraction. At this point we don't have any binaries available online as most people interested in this project are typically developers. Building it is comparatively easy, just follow the guide referenced in the README.
I don't mean to sound cynical, but, this is a long and detailed tutorial. They've clearly spent time working on it. IMO they should have tried to come up with a final result that is visually more appealing. Otherwise, people come to the tutorial and it's like, why go through all this work to recreate something that just looks meh? (a poorly lit model of a house floating on a grey flat rectangle) IMO, just a plain old Cornell box[0] would have looked better (and not been more complex to achieve).
Going to go out on a limb and guess that it's because the people who would actually go through these tutorials are people who already know what can be accomplished with raytracing and want to get their feet wet with doing so in Vulkan.
The average person who's looking for "What can raytracing do for me" is going to get their answer from Control, Battlefield, Metro Exodus, or even the Quake II and Minecraft demos, and not some developer-centric Vulkan tutorial from Github.
I'm a single data point, but as a graphics engineer working on DDGI and having worked on SSR/SSAO, SDF shadows etc... I really do not care about the visuals of the tutorial haha. Clear greyscale images are honestly perfect for understanding exactly what's going on. The tutorial is meant to explain the mechanics, not dazzle you with a material shader.
I think it's a good decision, because it makes it clear that this just teaches you to use the Vulkan extension and doesn't have a lot of useless fluff. The screenshot is a good way to communicate this and set the expectations of the audience.
Vulkan is a very low level API and even putting a single triangle on the screen is a huge wall of C++ code, so it's necessary to keep the domain part of the example as simple as possible.
I agree with you. That image makes me distrust the tutorial like maybe it's not actually using ray tracing. That image certainly does not look ray traced. There is no radiance what so ever. It looks forward rendered with standard phong shading and shadow maps.
As a graphics engineer, the more compelling use cases of ray tracing are things like visibility tests, which can be temporally stabilized by adding a low-discrepancy noise to each sample and used to perform leak-free global illumination. Samples can be used to feed other aspects of the pipeline like generating soft shadows, or atmospheric effects, etc.
It's an API tutorial, for professionals in the field. Ray tracing is just one tool in the toolbox, not the alpha and omega of how the frame gets composed.
