Video Game Blurs (and how the best one works)

[-]

tron21net@reddit

The website cannot be viewed as the FQDN has DNSSEC enabled and yet cannot be resolved behind properly configured DNSSEC resolver: https://dnsviz.net/d/blog.frost.kiwi/dnssec/

For those of you who are able to view the website then your DNS resolvers either don't support DNSSEC or are horribly misconfigured.

[-]

Ohh wow! All the years that this blog exists and is hosted on Github pages, I have never seen a connection failure or the thing you described. Indeed, the main domain and the subdomain point to different places.

I followed https://docs.github.com/en/pages/configuring-a-custom-domain-for-your-github-pages-site/managing-a-custom-domain-for-your-github-pages-site closely during setup (the `CNAME` case) and my Registrar has absolutely no DNSSEC enabled, infact, it's a paid option I do not pay for.

I am not sure what is happening here. Can you elaborate on, if I don't have DNSSEC enabled, what is this error implying and how can I fix it?

[-]

tron21net@reddit

I narrowed it down to a temporary error with some ui-dns.com name servers. Even though frost.kiwi is not showing to have DNSSEC enabled, as mentioned in a later post.

[-]

mallardtheduck@reddit

Not sure if something's changed since you posted that, but it's covered by an "opt-out" in the NSEC3 records (for both the .kiwi TLD and the .github.io domain that the CNAME references). The purpose of the opt-out flag is to allow insecure delegations for subdomains.

[-]

tron21net@reddit

Strangely enough I just retried the site and it now resolves for me.

I checked bind9 logs and was getting:

broken trust chain resolving 'frost.kiwi/DS/IN': 2620:10a:80ab::1#53
broken trust chain resolving 'blog.frost.kiwi/A/IN': 217.160.82.64#53
validating blog.frost.kiwi/CNAME: bad cache hit (frost.kiwi/DS)
broken trust chain resolving 'blog.frost.kiwi/A/IN': 2001:8d8:fe:53:0:d9a0:5240:100#53
validating blog.frost.kiwi/CNAME: bad cache hit (frost.kiwi/DS)
broken trust chain resolving 'blog.frost.kiwi/AAAA/IN': 2001:8d8:fe:53:0:d9a0:5315:100#53

[-]

chazzeromus@reddit

is that neo tokyo

[-]

Frost-Kiwi@reddit (OP)

Yes it is <3

[-]

WuPaulTangClan@reddit

I have been scared of anything remotely related to programming the GPU but this makes me want to dip my toes in it.

Any good suggestions on where to go from this article? Some WebGL playground in the browser with a tutorial in hand?

[-]

lunchmeat317@reddit

You might want to checkout webglfundamentals.org if you're looking for WebGL. There's also webgpufundamentals.org for WebGPU, which allows GPU compute!

They aren't interactive but the content is very good and worth a read.

[-]

Boojum@reddit

Shadertoy!

It's basically a WebGL playground for shaders. Click through, find an effect you like, open it up to see the source and start mucking with it in the browser. Make an account and you can save and share your own creations, comment on others, etc.

[-]

lunchmeat317@reddit

This is the type of content this subreddit needs and deserves. Thank you.

It's actually interesting because although it's not the same thing, I think that some of these core techniques are also used in audio processing (convolution reverbs abd whatnot). I'm interested in the downsampling mentioned in the article (especially the fact that it grows logarithmically), and I love the fact that articles like this one make me think. (There's a lot to learn and maybe I'm off-base about the audio similarities, but that's also an opportunity.) I'll definitely be rereading this one.

[-]

wd40bomber7@reddit

I always heard the "simple" strategy for a bloom filter was to write to a lower resolution texture and then upscale it allowing the GPU upscaler to handle the "blur" basically. I wonder how that compares in look/perf?

[-]

Ameisen@reddit

You don't need to upscale, you can just sample it as-is, and the magnification filter set for the sampler should take care of it - unless you want a specific upscaling algorithm.

[-]

BlindTreeFrog@reddit

Feel like I once read a similar article reviewing the technical details of how Dues Ex: Human Revolution did it's neon glow, but I can't seem to find it now.

Or at least, I don't think this article is the one that I'm thinking of:
https://www.adriancourreges.com/blog/2015/03/10/deus-ex-human-revolution-graphics-study/

[-]

balefrost@reddit

If we're collecting interesting articles along these lines, there's this one about Elden Ring: https://mamoniem.com/behind-the-pretty-frames-elden-ring/

There's also this seminal GPU Gems chapter about the use of glow in Tron 2.0 (a game I remember very fondly): https://developer.nvidia.com/gpugems/gpugems/part-iv-image-processing/chapter-21-real-time-glow. That was state-of-the-art... 20 years ago!

[-]

AegisToast@reddit

Wow, legitimately excellent article. The interactive parts are really well done, it reminds me a lot of the old RedBlobGames articles

[-]

iamcleek@reddit

that format is absolutely hostile to reading.

[-]

Venthe@reddit

My friend, amazing writeup.

[-]

The_Northern_Light@reddit

Nice visualizations 👍 thanks for the share

[-]

TankorSmash@reddit

This is stupid good, holy jumping

[-]

SlovenianTherapist@reddit

amazing article

[-]

skatopher@reddit

You are going to have to convince me I want blur in the game before I care if it’s efficient. I hold space for the idea someone can do it right, but in practice it just seems like we are wasting resources to make the scene uglier. I disable this on level 2 of every game I’ve ever bought. I kick myself for not just doing it by default.