Toots for kb@abraum.social account

Written by kb on 2024-12-25 at 10:14

Also hi to all the new followers. I rarely post, and if I do, it‘s usually either pretty obscure or about how stupid some of my code is, or both. Have fun :)

Feeling pressure now to write something short, witty and relatable now. Ugh...

Ok, how about this:

One of the weirdest side effects of those ADHD meds is that I‘m getting kinda good at UI/Frontend lately - as in, where does that patience when dealing with CSS (or even XAML) and making stuff pleasing to the eye come from 😀

=> More informations about this toot | View the thread

Written by kb on 2024-12-22 at 14:15

Update: SOLVED! ❤️

[It‘s that time of the year again: if anyone knows about one or two @darya/tags/38c3" class="mention hashtag" rel="tag">#38c3 tickets (Standard) that would like to have a new owner, and I would be happy to have them (and bring some A/V and event organizing experience if help is needed :) )]

=> More informations about this toot | View the thread

Written by kb on 2024-12-10 at 18:37

In today's "things that are obvious in hindsight" news:

Instead of the inverse transpose matrix, use the adjugate matrix for transforming normals. It leaves out the 1/determinant which you'd normalize() away anyway.

https://www.shadertoy.com/view/3s33zj

Only drawback: It doesn't flip the normals when the world matrix is inside out (eg. with negative scale). But you can counter this by flipping backface normals in the shader, and boom - everything always works, regardless what the user throws at it. \o/

=> More informations about this toot | View the thread

Written by kb on 2024-05-15 at 15:01

Anyhoo, if someone has a better idea or even knows that I'm wrong and has an actual solution, feel free to help :)

=> More informations about this toot | View the thread

Written by kb on 2024-05-15 at 14:56

Today on "easy graphics programming stuff made hard":

Additive blending of two images with alpha is impossible.

As in - if you have two premultiplied RGBA colors, and these colors are both HDR, i.e. can have RGB values greater than 1 *, there's no correct way to set the resulting alpha channel.

After some thinking and trying around, this looks the least wrong:

dest.xyz = c1.xyz + c2.xyz;

dest.w = c2>0 ? lerp(c1.w, c2.w, luma(c2)/luma(dest)) : c1.w;

=> More informations about this toot | View the thread

Written by kb on 2024-04-05 at 11:08

So, now that Windows officially doesn't have a mail application anymore, are there any good 3rd party ones? Back to Thunderbird?

=> More informations about this toot | View the thread

Written by kb on 2024-03-26 at 19:34

It‘s that time of the year again #revision2024 #demoscene #foh

=> View attached media

=> More informations about this toot | View the thread

Written by kb on 2024-03-01 at 14:41

(props to all the formats/codecs/programs out there that at least do CICP, you're appreciated.)

=> More informations about this toot | View the thread

Written by kb on 2024-03-01 at 14:38

If you're lucky (and I mean REALLY lucky) people put the color space(s) into the file name and leave the problem to you, the user.

And this also means that the absolute majority of software, including the big video editing products, is actually blissfully unaware of the color space anything is in most of the time. It either fits, or you're required to find the proper LUT or perhaps OCIO preset - otherwise it's just wrong.

How is that the standard that everyone's forced to work with, please?

=> More informations about this toot | View the thread

Written by kb on 2024-03-01 at 14:31

A thing that weirds me out most about diving headfirst into that whole color management topic is how very much at the beginning of it the whole industry is sometimes.

The latest example is LUTs. LUTs are like one of the most used tools for color grading and color space conversion. So imagine my surprise when I had to find out that literally none of the common LUT formats specify, like, the incoming and outgoing color space in any way.

I mean - that's kind of important information, isn't it?

=> More informations about this toot | View the thread

Written by kb on 2024-02-28 at 20:20

Audio programming protip: If you write something that has to run indefinitely, let it run for at least 25 hours. That's 2^32 samples @ 48khz.

(me? recently bitten by that one? nooooo)

=> More informations about this toot | View the thread

Written by kb on 2024-02-12 at 15:20

Hello graphics, and especially Direct3D bubble - any idea if setting the DXGI_USAGE_SHADER_INPUT flag on a swap chain comes with a performance penalty? (I only need it for screenshots, so I can live with one more copy I guess, but who knows)

=> More informations about this toot | View the thread

Written by kb on 2024-02-05 at 16:03

Ah, the joys of color management

=> View attached media

=> More informations about this toot | View the thread

Written by kb on 2024-01-29 at 11:42

I made this, and VS is still going.

(original: https://xkcd.com/303/)

=> View attached media

=> More informations about this toot | View the thread

Written by kb on 2024-01-16 at 10:25

And btw, any pointers towards any piece of software that has or, better, nails this would also be really appreciated. I looked around quite a bit but haven't found anything good yet.

=> More informations about this toot | View the thread

Written by kb on 2024-01-16 at 10:23

Hey graphics programming bubble, I have a question.

Does anyone know good practices for a color editor/picker widget that can do HDR and WCG? Like, here, pick any color out of the Rec2100 space? I got the tech part down (say hello to my unholy contraption made out of Windows Forms and DirectComposition, don't ask), what I'm looking for is just how the UI should look and feel. HSL? Or something something LAB or ICtCp? Color wheel or triangle or HS square? Do I mark out of gamut colors?

Help :)

=> View attached media

=> More informations about this toot | View the thread

=> This profile with reblog | Go to kb@abraum.social account

Proxy Information
Original URL
gemini://mastogem.picasoft.net/profile/111761780843390142
Status Code
Success (20)
Meta
text/gemini
Capsule Response Time
392.298995 milliseconds
Gemini-to-HTML Time
3.795561 milliseconds

This content has been proxied by September (ba2dc).