Correct sRGB Dithering

This is a brain-dump inspired by a thread on twit­ter about cor­rect™ dither in sRGB, mean­ing, to choose the dither pat­tern in such a way as to pre­serve the phys­i­cal bright­ness of the orig­i­nal pix­els. This is in prin­ci­ple a solved prob­lem, but the dev­il is in the details that are eas­i­ly over­looked, espe­cial­ly when dither­ing to only a few quan­ti­za­tion levels.

So, this top­ic came up on twitter:

I had pre­vi­ous­ly spent some time to wrap my head around this exact prob­lem, so I shot from the hip with some pseu­do code that I used in Space Glid­er on Shader­toy. Code post­ings on twit­ter are nev­er a good idea, so here is a cleaned up ver­sion wrapped up in a prop­er function:

Con­tin­ue read­ing

Followup to Atmospheric Scattering—Part 1: Overview

This post is the first in a series to fol­low-​up on my 2012 GPU Pro 3 arti­cle about atmos­pher­ic scat­ter­ing [11]. What I showed there was a full sin­gle-​scat­ter­ing solu­tion for a plan­e­tary atmos­phere run­ning in a pix­el shad­er, dynam­ic and in real time, with­out pre-​com­pu­ta­tion or sim­pli­fy­ing assump­tions. The key to this achieve­ment was a nov­el and effi­cient way to eval­u­ate the Chap­man func­tion [2], hence the title. In the time since then I have improved on the algo­rithm and extend­ed it to include aspects of mul­ti­ple scat­ter­ing. The lat­ter caus­es hor­i­zon­tal dif­fu­sion (twi­light sit­u­a­tions) and ver­ti­cal dif­fu­sion (deep atmos­pheres), and nei­ther can be ignored for a gen­er­al atmos­phere ren­der­er in a space game, for example.

I have writ­ten a Shader­toy that reflects the cur­rent state of affairs. It’s a mini flight sim­u­la­tor that also fea­tures clouds, and oth­er ren­der­ing good­ies. A WebGL 2 capa­ble brows­er is need­ed to run it. Under Win­dows, the ANGLE/Direct 3D trans­la­tor may take a long time to com­pile it (up to a minute is noth­ing unusu­al, but it runs fast after­wards). When suc­cess­ful­ly com­piled it should look like this:
Con­tin­ue read­ing

So, where are the stars?

In my pre­vi­ous rant about dynam­ic expo­sure in Elite Dan­ger­ous (which hon­est­ly applies to any oth­er space game made to date), I made a rough cal­cu­la­tion to pre­dict the bright­ness of stars as they should real­is­ti­cal­ly appear in pho­tos tak­en in out­er space. My pre­dic­tion was, that,

  • for an illu­mi­na­tion of sim­i­lar strength to that on earth,
  • if the sun­lit parts are prop­er­ly exposed,
  • and with an angu­lar res­o­lu­tion of about 2 arc min­utes per pixel,

then the pix­el-val­ue of a promi­nent star should be in the order of 1 to 3 (out of 255, in 8‑bit sRGB encod­ing). Since then I was curi­ous to find some real world val­i­da­tion for that fact, and it seems I have now found it.

ISS_and_Endeavour_EV+0

Con­tin­ue read­ing

\usepackage{cmbright}

The ‘mod­ern’ look­ing sans-serif font I use recent­ly in \text{\LaTeX} equa­tions on this blog is called ‘Com­put­er Mod­ern Bright’, and actu­al­ly is not so mod­ern at all: Designed by Wal­ter Schmidt in 1996, it is still to date the only free sans-serif font avail­able for \text{\LaTeX} with full math sup­port. Type‑1 ver­sions of this font are avail­able in the cm-super pack­age, but I did­n’t need to install any­thing, because appar­ent­ly, the Quick­La­TeX Word­Press plu­g­in has them already. The only thing to do was adding just one line to the preamble:

\usepackage{amsmath}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{cmbright}   % computer modern bright

I also turned on the SVG images fea­ture that was added with ver­sion 3.8 of Quick­La­tex, so the equa­tions are no longer pixel­lat­ed on reti­na dis­plays or when zoom­ing in! Neat, huh?

Elite Dangerous: Impressions of Deep Space Rendering

I am a backer of the upcom­ing Elite Dan­ger­ous game and have par­tic­i­pat­ed in their pre­mi­um beta pro­gramme from the begin­ning, pos­i­tive­ly enjoy­ing what was there at the ear­ly time. ‘Pre­mi­um beta’ sounds like an oxy­moron, pay­ing a pre­mi­um for an unfin­ished game, but it is noth­ing more than pur­chas­ing the same backer sta­tus as that from the Kick­starter cam­paign.

I came into con­tact with the orig­i­nal Elite dur­ing christ­mas in 1985. Com­pared with the progress I made back then in just two days, my recent per­for­mance in ED is lousy; I think my com­bat rat­ing now would be ‘com­pe­tent’.

But this will not be a game­play review, instead I’m going to share thoughts that were inspired while play­ing ED, most­ly about graph­ics and shad­ing, things like dynam­ic range, sur­face mate­ri­als, phase curves, ‘real’ pho­tom­e­try, and so on; so … after I loaded the game and jumped through hyper­space for the first time (actu­al­ly the sec­ond time), I was greet­ed by this screen fill­ing disk of hot plasma:

ED001

Con­tin­ue read­ing

Journey into the Zone (Plates)

I have exper­i­ment­ed recent­ly with zone plates, which are the 2‑D equiv­a­lent of a chirp. Zone plates make for excel­lent test images to detect defi­cien­cies in image pro­cess­ing algo­rithms or dis­play and cam­era cal­i­bra­tion. They have inter­est­ing prop­er­ties: Each point on a zone plate cor­re­sponds to a unique instan­ta­neous wave vec­tor, and also like a gauss­ian a zone plate is its own Fouri­er trans­form. A quick image search (google, bing) turns up many results, but I found all of them more or less unus­able, so I made my own.

Zone Plates Done Right

I made the fol­low­ing two 256×256 zone plates, which I am releas­ing into the pub­lic so they can be used by any­one freely.

Cosine zone plate with constrast weighting

Cosine zone plateCC0

Zone plates with contrast weighting

Sine zone plateCC0

Con­tin­ue read­ing

Yes, sRGB is like µ‑law encoding

I vague­ly remem­ber some­one mak­ing a com­ment in a dis­cus­sion about sRGB, that ran along the lines of

So then, is sRGB like µ‑law encoding?

This com­ment was not about the col­or space itself but about the spe­cif­ic pix­el for­mats nowa­days brand­ed as ‘sRGB’. In this case, the answer should be yes. And while the tech­ni­cal details are not exact­ly the same, that anal­o­gy with the µ‑law very much nails it.

When you think of sRGB pix­el for­mats as noth­ing but a spe­cial encod­ing, it becomes clear that using such a for­mat does not make you auto­mat­i­cal­ly “very picky of col­or repro­duc­tion”. This assump­tion was used by hard­ware ven­dors to ratio­nal­ize the deci­sion to lim­it the sup­port of sRGB pix­el for­mats to 8‑bit pre­ci­sion, because peo­ple “would nev­er want” to have sRGB sup­port for any­thing less. Not true!Screen Shot 2014-03-06 at 19.02.54I’m going to make a case for this lat­er. But first things first.

Con­tin­ue read­ing

Followup: Normal Mapping Without Precomputed Tangents

This post is a fol­low-up to my 2006 ShaderX5 arti­cle [4] about nor­mal map­ping with­out a pre-com­put­ed tan­gent basis. In the time since then I have refined this tech­nique with lessons learned in real life. For those unfa­mil­iar with the top­ic, the moti­va­tion was to con­struct the tan­gent frame on the fly in the pix­el shad­er, which iron­i­cal­ly is the exact oppo­site of the moti­va­tion from [2]:

Since it is not 1997 any­more, doing the tan­gent space on-the-fly has some poten­tial ben­e­fits, such as reduced com­plex­i­ty of asset tools, per-ver­tex band­width and stor­age, attribute inter­po­la­tors, trans­form work for skinned mesh­es and last but not least, the pos­si­bil­i­ty to apply nor­mal maps to any pro­ce­du­ral­ly gen­er­at­ed tex­ture coor­di­nates or non-lin­ear defor­ma­tions. Con­tin­ue read­ing

Branchless Matrix to Quaternion Conversion

(EDIT: This arti­cle is a more in-depth write­up of an algo­rithm that I devel­oped around 2005, and first post­ed to Mar­tin Bak­er’s Euclid­ean Space web­site. That time was the height of the Intel Net­Burst archi­tec­ture, which was noto­ri­ous for its deep pipeline and high branch mis­pre­dic­tion penal­ty. Hence the moti­va­tion to devel­op a branch-free matrix to quater­nion con­ver­sion rou­tine. What fol­lows is the com­plete deriva­tion and analy­sis of this idea.)

The orig­i­nal rou­tine to con­vert a matrix to a quater­nion was giv­en by Ken Shoe­make [1] and is very branchy. There is a way to elim­i­nate these branch­es and arrive at a com­plete­ly branch-free and high­ly par­al­leliz­able code. The trade off is the intro­duc­tion of 3 addi­tion­al square roots. Jump to the analy­sis sec­tion and the end of this arti­cle, or con­tin­ue fist with the math bits.

Con­tin­ue read­ing

Cloud rendering and the relativity of whiteness

Here are some philo­soph­i­cal and ren­der­ing-relat­ed ques­tions that I took home from the last vaca­tion. What’s the col­or of clouds? The stan­dard answer would be, white.
What’s the col­or of snow? Again, white. Ok, then look at the fol­low­ing pic­ture, where the snow seems con­sid­er­ably whiter. This is the case in almost all pho­tos that I took.

There is an image on Wikipedia from the same gen­er­al area on which the bright­ness dif­fer­ence between clouds vs snow is even more pro­nounced. If you look at the direct­ly lit parts of the snow and con­sid­er it white (#ffffff), then the direct­ly lit parts of the clouds are at most 50% grey (#bbbbbb). Is that an evi­dence of air pol­lu­tion? Unlike­ly! (At least not in Tyrol).

Con­tin­ue read­ing