Theo Verelst Local Diary Page 44

I've ditched the usual header for the moment, I think it doesn't help much anyhow.
This page is copyrighted by me, and may be read and transfered by any means only as a whole and including the references to me. I guess thats normal.

Home (alt)   Tripod Home   Local Diary list   Tripod Diary List   Previous Diary

This page is under contruction, so check back later, too.

  Tue Jan 22 22:08,  2008

   Moved to Bay Area? Certainly not, I only wish! "This is Amsterdam man....". Cute spot though.

I listened to a WBGO newyear webcast (it´s a New York radio station with webfeed, which is member supported, and yes I made somebody pitch in, too...) with the Count Basie orchestra in San Francisco, which made me feel good.

The way I did that is fun, I used a Nokia Communicator (the new one) to receive an internet webcast in realplayer form, which sounds pretty good. In fact I used internet that way, too, because the server is still where the internet currently is, in the new place there´s no connection.

What, moving again? Nasty subject. In a way, at least I´m moving allright. And the wireless internet from the phone works very fast in fact, about the speed of the earlier Adsl I´ve used some years ago, amazing, though maybe not so good for the local elektrosmog. ´Use with measure´ or something.

   local view on a wintermorning

As a added remark: of course when you´re reading and viewing this page the server has come back on line! So the entertainment and informing of interested people may continue.

Mixed pictures

So what camera(s) did you I for the recent pages? Well interesting question, in fact I fully abandoned the HP 5 megapixel camera I made thousands of pictures with and use two others: the Sony 2k High Definition camera, and a Nokia communicator with 3 Megapixel camera. Of course the Sony is the most Hifi camera, it can really capture a lot of good light for a film or photo, or a photo from film (HD video) as I often do. But the new Nokia has a very decent cam built in which does more than a bit of those photographic ´optimisations´ such that the photos look juicy and all.

I think it is a funny game to not say which processed picture like I show mostly on these pages are from which of these two cams, so people can guess. I like trying to get preferably a great number of classic and recent photographic types of pictures so as to have a good palette of choices instead of a camera precooking always a certain style.

So unless I mention it (but it is possible to look at the picture names to get ideas) it is quite possible the images are processed, and in some cases such that normal photoshop users would be lost to revreate those effects, totally. And I mean I was a poweruser of Photoshop years ago, and times even paid for it, so I know what I´m saying! And I like currently (having no money) to use Open Source Software, and make it myself for free,too, so that the results on these pages are most likely processed by Cinelerra, Cinepaint (a recent linux version) or Gimp, or a combination thereof. In some cases like processed frames from HD I use first cinelerra for adaptive frame averaging (in 32 bit floating point per colour component!), processing and saving of a HDR (High Dynamic Range) TIFF image, subsequently cinepaint in 16 bit per colour component mode for HDR quality processing like gamma correction, and brightness control, sizing with bit accuracy improvement, and possibly sharpening, of which the resized result is then saved in high quality jpeg, which I feed through gimp to get a progessely coded jpeg with quality versus size control. In other cases I simply use gimp under windows to make a screendump from the high quality mode / Purevideo accelerated windows media player (also a fairly recent update) and save it in a fitting progessive jpeg accuracy.

Well, that´s where I was some time ago, the more advanced, but not necessarily better processing would involve mixing amounts of equalized image versions with the original, special effects (though seldom), multilevel analysis and selective colour enhancements (also using layers), and all of this preferably in HDR, or it soon becomes a color quantisation hell (8 bits processing gives accumulated errors). Oh I also tend to use curves to adjust brightness and contrast, but since the HD cam I often prefer a hopefully correct amount of gamma correction.

Magazine or book looks from the 70s, and general film impressions (which require more accuracy than most monitors will give you) are favorite effects of me to aim for. Though I don´t normally introduce artificial grainyness, and am aware of the power of analog processing of actual negatives as compared to even high resolution HDR images...


I´ve in the meanwhile tried to look at BBC space relayed broadcasts, which worked when it wasn´t windforce 10, at which point I had to support the not so fragile beam with the antenna with a broomstick.

The theater being wrapped up.

The sub and a main speaker on the backseat,

amps and the other main speaker in the trunk.

a temporary setup.

     A video from Barbara doing kitchen painting, click on the image.
     (mpeg4, 640x480, 36 seconds, 5.1 MB)

Pretty German-ish light.

Hard winds require (improvised) stabilisation of the dish.

A start to make a frame to extend to almost ceiling-height which is to carry a full layer of heavy sound isolation material.

Where this page was partially made:

A little improvised workspace. The table is Ikea, and I for the first time tried furniture oil on it, twice, to hopefully make it watertight in 2 weeks. The tower on the left is an experimental beamer setup which works good. The floor is isolated using soft and hardboard on hard concrete, to create less resonances, the blinds I custom-made to exactly fill the windowpane. The computer setup looks a bit modest, but it has fully portable internet, it´s a Dell M90 (some years old) with an actual 1080HD screen, which the quadro card can drive good, it has a long USB cord to an excellent Lexicon Omega 24 bit AD/DA converter, hooked up to my main amplification setup, and there is an ethernet connection to a good internet pipe (over a megabyte (not bit) per second download speed) and to a major server (where this page is on) of which I remotely use X windows, too.

Oh, and the notebook has a recently installed Fedora 8 (32 bit) with a perfectly working firewire High Definition camera link which even allows live 1080 HD monitoring (using dvgrab -noavc -f mpeg2 - | mplayer -cache 8196 - , giving stable and non-jerky but not totally perfect HD viewing), a working Nvidia Quadro card driver for fast graphics, also allowing the latest Blender combined with the latest Gelato to work well, it has cinepaint, cinelerra (but this version is slow), and the portable audio department is great because I now can plug in the Lexicon and use Jack or even PulseAudio (about which more later) with Rosegarden to do multitracking (works great) and several synthesis programs, and I recently tried out JackRack with some LADSPA effects on a live guitar setup which also works with great sound.
And hopefully I now can mount.cifs a server partition transparently again, which didn´t work on FC5.

The server (A 3.3 GHz Athlon 64 machine) where you read this from has also been upgraded to Fedora 8, but is unfortunately has a problem running dvgrab, which I want to use for a webcam. Also PulseAudio, which I did the upgrade for mainly, has problems, altough it does work, and I even got a remote audio stream to work, but it isn´t where I want a multi machine 32 bit good quality Jack, unconverting, distributed audio system to be, yet. I´d rather have no resampling than not working user permissions and general messups...

This photo is HDR-processed from phone-camera 3 Megapixel original:

Mobile Web Serving

I was pleased to find it is possible to run a webserver on a mobile phone:


For a user with certain permission, it is possible to take a picture with the camera when it's webserver is running, from a remote location, that picture doesn't end up on the camera's memory it only appears on a web page, but an audible photo sound is there when the photo is taken. No GPS info, no flash though. Example:


The server runs physically on the actual phone, it is an adapted apache server with a lot of python scripts to present all kinds of menus. It does take a little extra power!

I'm working on using not mobile serving but mobile viewing using a good quality camera on the Fedora server side of which the live BirdCam is visible on the mobile phone, after encoding for 10 seconds the mobile phone shows a little downscaled 460x480 (DVD picture size) movie of 10 seconds of a birdhouse recorded in mp4, with audio!

The page I use for it now is this one I use various coding sizez and qualities, and even at times TV is encoded, so you might get 10 seconds of quite viewable CNN or so...

On yet another side of things, I made my analog audio mixer´s Linux board finally connect to the actual internet over the router/modem the such that it can download files from the internet (cnn headlines on the machine display, or software updates, mix data, device setting, ...) but also it can be reached from the internet, by using

a page is routed to your browser which comes from the mixer´s Linux control board, if it is switched on (mostly at the moment it is switched off), mind that the IP address above may change, just use to get the address, and then add the :8080 to try it out, and in principle you might be able to control the volume of a channel from where you are on the audio setup the mixer prototype controls. Uuueh, this actually works, I have used a mobile phone internet connection to change the actual volume of a analog channel pair (feeding the audio amplifiers) ! Needless to say  such control will not be generally on the internet, because the system is VERY powerfull, and should not be put loud by funny people on the web, of course...

Wed Feb 27 12:13:49 CET 2008   I´ve (obviously) switched this setup off until I make it a bit more safe, that is, I took the internet connection down until I made a few changes, so please check back later.

Some more Mathematics: the mathathematical sound wave server

Some pages ago I did Taylor expansions of a non-trivial integral which is probably ´a bit much´ for any normal person out there except edified enough engineers and some scientists or mathematicians in a connected field. The added value of putting such for me momentarily not so life or science essential but interesting experimentation on my diary pages is that for advanced scientists it does make clear maxima is doing good and that I did some relevant mathematics when you consider that many things in science are based on mathematics, especially when those things are well contained and offer power otherwise not available, like practically speaking seemingly unimportant limits to infinity, or Taylor expansions to mathematical or scientific laypeople.

To be more interesting to non-experts, lets see what a powerfull mathematics package like Maxima (or Mathematica of course, which costs money) can do with more everyday formulas like the equation of Pythagoras:

   x^2 + y^2 = r^2

being the equality equation describing a circle in 2 dimensions.

On another side of things I made a test setup to render mathematically generated sounds on the server!

I made a cgi script to run the mathematics package maxima, and some more programs and tools automatically to generate a short audio file with a given mathematical function as the wave rendered in it.

For instance I could take the maxima formula sin(440*x) which would generate a perfect (in 16 bits CD quality) sine wave in the generated sound file which can be downloaded and listened to, but if I feel a bit experimental I could make very complicated formulas, like on this page or just play around with fundamental waves like sine waves multiplied by an exponent: sin(440*%pi*2*x)*exp(-10*x) (or: sin(440*float(%pi)*2*x)*exp(-10*x) to make Fortran not have to crash because it doesn´t know symbolic PI, this is the result).

All this can make wonderfull sounds, which the server automatically puts in mp3 form, although it may be better to make the wav file downloadable itself, too, because it seems various windows players don´t like the 3 seconds long mp3.

The url is (carefull, I´m working on it)*x) where the part after the question mark is the maxima formatted formula, with x being the progressing time in seconds. The resulting mpeg is stored in and can be freely downloaded and used. Currently you must make sure yourself the formula is normalized, in other words is never greater than +1 or smaller -1, and as I said x runs from about 0 to 3 and is time is seconds.

The setup isn´t as easy as it might seem, and the maxima package is capable of making and processing very complicated mathematics, so the number of possible waves is sheer infinite. And the symbolic manipulation is very powerfull such as to do formal differentation and even integration, which works in practice, for instance changing the above formula in diff(sin(x*440),x) gives  440*cos(440*x) without effort for maxima.

Generating the waves is done by generating Fortran program code from the maxima formula, after evaluating it, and then running a linked C program with a loop over 44100 * 3 samples to create the .wav sound file, so each time a new formula is entered, a compilation takes place after maxima is run, and after that the resulting program is executed to create the wav file, and after that a program to turn this into mpeg, all automatically, and within seconds.

People may say: that´s a strange detour here, isn´t it? Uuuhr, nope. The compilation of fortran and the C loop and small wav library link take a little time, but this pays of big because the evaluation of the formula for varying time steps becomes just about as fast as a contemporary computer gets doing this job brute force (without interpolating or repeated evaluation schemes) because the gcc and gfortran compiler are excellent and give very fast executable program code, the 44.1 thousand samples per second can be quickly computed while one waits in this case for the page to download, and so formulas can be rendered quickly. The maxima step is because of all kinds of functions and processing steps and reorderings that can be done with it, but rendering numerical results from maxima would probably be prohibitively slow. The Fortran code generator comes standard with maxima, just like the tex typeset output, so standard maxima can be used, and thus far the sequence of program being called works good, including for large formulas, like this one:


Of course the page is totally in buildup c.q. experimental, probably the URL translation isn´t even ok (yet).

Wed Feb 27 12:13:49 CET 2008   I´ve (obviously) switched this setup off until I make it a bit more safe, at least some obvious ceavats should be gone like adverse maxima commands as part of the formula. Please , check back later.

The Blues cannot be had

Because it is always going to hit back in the hand, twice as hard, preferably. Mean statement? No the pope and the new age and such are mean, like ´mother´ in a negative connotation. Some sins and lifes cannot be forgiven, especially when they haven´t even ended are are on purpose to hurt people. And some people only play at the highest level and take everythings they can find, and require that they are kicked out of their place, and probably will only listen to either The Reaper, curses from God or hitting them twice as hard as they do other people until the highest level is taken back for the free and at times mildly good people.

A guitar-played blues, one take, no elaborate production, 192 kilobit/sec mp3, 3.5 Megabyte, 2 min 24 sec.

Mike Stern will perform with the YellowJackets this years, isn´t that great. And in fact a recording has been done, too. I should join, too!

This is where I performed several times on keyboards/synthesizers in the past, the Koornbeurs in Delft where, for other reasons at some point I became member of this protestant non-corporal, non-initiation rite club, and probably still would be if I´d have had money the last times. Probably I played the blues there, too, but certainly elsewhere in that town, too.

What about the title of this section? I think in general the blues won't be had, meaning people won't be having control over it or debase it in some general sense. And that it wins of the new age sh* called music hands down just like other real music, and of the nig problems having surrounded the good people.

Edited HD Film (Video)

The following are stills from the 3 scenes in a test HD film with processing and editing, having used Cinelerra. I´ll put on a more compressed version of the result, because my version is about 140 megabytes for 34 seconds, which is great on the 1080 screen and the 720 colour reference beamer (A 720 HD HD1000 from Mitsubishi), also motionwise, and while not becoming as dead as a doormat anywhere in the process.

(couldn´t you pixel-*, uh -mess these to a bit smaller format, Theo? Sure, but all that pixelmessing, heh...)

But this camera is not really HD, is it!? Eh, it is known to have a 2.2 Megapixel ClearVid advanced CMOS sensor with 14 bit signal processing in it, so it should be normally 2k 16:9 capable, maybe there is this issue which makes lets say 1 megapixel pixels into 3 x 333 kilopixels RGB, I don´t know, except I happen to know the Clearvid uses 4 pixels (RGGB) per image pixel, but wether there are 2.2Mega RGBs or 2.2M any colour I don´t know.

The issue with stills is at least also that the motion and noise present in the 25megabit per second MPEG2 transport stream (which responds probably a more neutral than other compression formats) which are deliberate, such as colour dithering and sub-pixel motion and averaging of this, taking one frame doesn´t give a good impression of the movie shown on a sufficient resolution and fast enough LCD or (even better) on a DLP projector. I have made adaptively frame averaged pictures using cinelerra, which do look better but I cannot do partion motion estimation, which would be better still, but only whole frame motion compensation.

Visible by the PureVideo hardware accelerated windows media player full-screen controls, the above is also an actual still from a playing video, though shrunk afterward, from a HD recording from a moving vehicle! So the above road at dawn is actually moving at up to 30/40km/h or so.

This of course is not easy to film with a modest high end consumer HD camera or normal computer mpeg players, it requires using the right settings and a lot of knowledge about the equipment, unless one is lucky.

Remember, more than a few of the above images were taken with non-stationary camera position and/or from moving scenes, of which the following is a nice example, which is a large parking lot where I filmed holding the camera in hand from a moving and turning car, with no shutter applied, so all motion per frame comes through to the film. Great impressions are possible that way, one frame doesn´t do good justice to that:

All of the above images were processed from the original HD video material (and obviously most were scaled down), click on the images to get the original HD frame (from before processing)!

Clearly, I´d like a 4k HDR cam and major (striped) disk space to film with for a change...

What is he saying here? Well, I notice all the differences in the processed images, too, also on good CRT, various quality LCD and well behaved DLP projection, probably the issues about pixelation and distribution of illumination and colour sampling raised here are not because I didn´t understand the latest standardisation process about some video or photo format yet, but because I play with the whole idea and liked "North and South" on a late 70s colour TV better then the motion- and pixel-messed new shows over a digitally compressed media box channel. Because that wasn´t so messing dead and annoying.

The new age?

From pretty soon after I started I tried to record good quality HD video, like I´m well trained in ordinary (576i) video which worked good, but I also wanted to see how far it all would go rather more than the shots in let´s say Dirty Harry or Jaws II or so than most of the new cr*p or the obvious encoded video on the web or the infamous media encoder box or so, no matter how usefull either can be. So that the image is sharp enough is a given, although that is quite hard in HD (like I presume in film, too) I take it for granted that a serious recording is sharp, and that the sort of glossy-just-not style is possible I believe, too, that is boring pretty soon, unless maybe when it is truly glossy, which requires something glossy to film of which there is little around in Holland, I must say.

Sound  is very important in a good video or film, so I used this setup for instance to test seperate sound recording from the excellent cam microphones, for a lot better stereo, and still a lot vetter sound: two AT2020´s with the Lexicon Omega´s excellent preamps and 24 bit AD converters are hard to beat:

Of course recording the sound away from the camera postion is good too. A random rest recording with steaminess, which even in 2k or about 2Megapixel film is hard to get easy:

No way that is HD video, Theo, you´re conning us for sure. Maybe a bit advanced picture with some of your effects or so. Eh, no. This is one of the first HD recordings I made after basic practice with the cam, and it hasn´t been processed at all, and in fact this is a frame from the PureVideo enhanced windows media player (PureVideo harware HD acceleration requires a player) straight screendumped from a partially moving picture (the plants move in the wind). And this isn´t a lucky frame or so (here´s another maybe better to prove it) it is simply what the cam can do, and the player too, although that requires more than a bit of work, including installing the purevideo driver, which isn´t free (but only $25 or so).

So the above moves !? Sure, and totally realistic, which looks great.

Dirac wavelet video coding

Well, it is an excuse to do block estimation and implicit feature extraction and multilevel estimations too and therefore could be worth to use instead of the ´motion field m*f*´ idea which I saw on the last big IBC tradeshow being presented at Microsoft as the main example processing software, being used for ´The Office´ (US version). That isn´t great. It s*s, although I´d be glad to be able to do deinterlacing the way I´d like it and to get some good local motion estimates for colour enhancements and other improvements on my HD video.

That "dirac" doesn´t work Theo, it´s just brittish propaganda, it is only sounding nice, they got some stuff from some nice californian inhabitants, and that should do something, but they don´t have a clue either! It doesn´t compile, and otherwise there are fundamental impossibilities to use it!

Arg. Well, I guess I didn´t give Unix (professional predecessor of Linux, which copied just about anything from it, which mostly was legally no problem) courses back in the late 80´s for nothing, and didn´t program at university for no reason.

It can be used, but honestly it does force the use of uncompressed files in various forms and is probably therefore educationally decent, and as a consequence eats disk space even for a few seconds of 1080HD film being fed through it frame by frame. I took a little piece of HD film I made which I transformed from mpeg-2TS (25Mb/s) to uncompressed 8 bit RGB TIFF frames using Cinelerra.

Those I streamed into a converter program from Imagamagic, which can be yum-ed from Fedora (8/64) to make a big red green blue with a byte per colour:

   convert t/ti00*.tif -adjoin ti.rgb

I presumed the shell or library wildcard expansion would do a numerically ascending list job, which appeared to be the case. Then I used the Dirac converter to 444YUV form, stating the picture size (I didn´t expand the size to 1920 wide, whcih I did afterward again, though probably the pixel interpretation through the encoder isn´t perfect this way) and number of frames (or fields):

   dirac_RGBtoYUV444 < ti.rgb > ti.yuv 1440 1080 64

Now the direc encoder, presented with a ´- local´ option, is started to generate output frames fromthe encoding process, which goes at about a frame per 3 seconds on the fast notebook I used (though it used only one core of the 2), and I think I might have added ´-interlaced´ which would correspond with the source (´I´) format but the encoder cannot be given options it lists itself when called with --help to tell it to use bottom frame first and encode either with or without interlaced result:

   dirac_encoder -HD1080I50 -cformat YUV444P -verbose -width 1440 -qf 9 -local ti.yuv tt

The verbose option shows a lot of information about the coding process, and of course there are many more options for which I used the defaults. The result needs to be converted back to a format I can read;

   dirac_YUV444toRGB < tt.localdec.yuv > ./tt.localdec.rgb 1440 1080 64

Now the imagemagic viewer can be used on a machine with sufficient memory:

   display -size 1440x1080 -depth 8 tt.localdec.rgb

Frame for frame the output can be viewed that way. I used another imagemagic command line tool to get seperate frames back from the encoder results, in this case as a set of very high quality jpeg images:

   convert -size 1440x1080 -depth 8 -quality 98 tt.localdec.rgb t/ttlocdec.jpg

Which I put back into a mpeg2 transport stream using ffmpeg to use my Purevideo player to show the result as a short movie:

   ffmpeg -i t/ttlocdec-%d.jpg -f mpegts -b 40000k -r 25 -aspect 16:9 ttlocdec.mpg

The following picture is a screen dump from this last step:

All these pictures are big, but otherwise the results look a bit deteriorated. Oh, yeah, I know about the magicall ping format: doesn´t make it much better for serious videos or pictures, possibly for screendumps, and at least these big images are compatible with most browsers on earth.

To make the above a bit more interesting, the frame is from a moving scene: the camera was making a fairly slow pan during the above fragment. It is clear the encoder at quality equals 9 isn´t perfect. The original with Purevideo is fairly straight ok, but the general impression after the wavelet coding is quite different and more realistic, and probably even though there are blurred edges, the realism increase through motion estimation and the broad, unedged average filtering of the wavelet is a fresh and welcome effect after so many ´sharpness effects´ and pixel pattern messes.

I´ve experimented years ago with the then new jpeg2000, probably because I had example code for a DSP (which I never used) and the multilevel analysis of this is probably interesting for the new digital cinema which cannot afford to run from a multiple striped disk doing big full frames at 4k/25fps times 32 bits times the number of colour channels, if not it is usefull to analyse frames with, and keep away from dct block tricks making every light effect larger than 16 pixels broken from the start. So are the not so incredible effects of the transcoding visible you might ask, yes they are. I view on a 1080 HD monitor screen on which the perception of the image at distance varying from close to about a meter of the 17 inch screen are profound enough to spell out the differences between lets say a good projector impression of good material, a good book picture (a well made one), and a badly procecced video, myriads of pixelmessed web pictures and  "The Getaway" from HDdvd (including supplied x2xn magnification), compressed video and a burst of HDR raw camera images.

System struggles and solutions

Well, I installed Fedora 8/64 on two machines, amoung which the server on which these pages are served to you, and I installed Fedora 8 (32 bits dual core)  on the Dell M90 I use to make this page at the moment, which is a great improvement because on the notebook I can now use the Lexicon Omega 24 bit AD/DA Usb converter on Linux with Rosegarden which gives a very good audio signal path, and of course with synthesizer programs and effects (LADSPA effects with the stereo Jack-based signal path of jack-rack and also the notebook now has a midi interface through it.

Also, the advantage of F8 is the use of dvgrab -f mpeg2 which means I can read (also portable) HD video from the camera without glitches or startup time issues (as it was). And the mount.cifs local network mounts to server window shares works easy and transparent, so big video files (I mean several gigabytes easily) can be used y just clicking onthem and putting the time slider anywhere and the files will be served and the video player respons instantly without problems, which for HD with the Purevideo hardware rendered isn´t bad. Even the extra 320 GB external disk is put on the server and can be accessed over the (imitated, it is Redhat Fedora 8) local network.

There is a struggle here though: dvgrab works on a 64 and a 32 dual core system, but not on the server. I even compiled it myself ( I add string.h defines everywhere, skipped some test if with casting definitions and later skipped another test which erred) but it consistently doesn´t work on the server machine, maybe the old german firewire card doesn´t work with the new protocol stacks, I don´t know.

Which brings me to the next subject, I started to work on a webcam which is called "The BirdCam" because it is making a picture of a birdhouse! It can be easily understood I wanted the firewire to the old video camera to get images, and I even plan to stream video, or make it downloadable as mpeg4 from the server, and preferably usable on a mobile phone like the Nokia E90 (the Communicator with big screen) which should be able to use mobile internet to get a DVD sized movie from the server. In fact I tested this is possible with the real good quality but years old consumer Sony digital video cam with major ´analog´ zoom lens  which instead of over (digital) firewire cable (al long one, and it does support ieee1394) is coupled to the server with a S-video cable (and sound connection) via a Pinnacle PCTV Deluxe digitizer USB box, which is yeeeaars old, and has broken already two supplys and which encodes into mpeg2 (or 1) in hardware for about 6 watts of extra power. Also, it can be adjusted to receive cable TV channels at very good quality for consumer norms at 6 megabit per seconds. The box has a good new major wall wart supply form Conrad now, which is rated to the required power and not so big, which doesn´t get hot (the previous one was not normally loaded too much when measured, but it had a little less power then the rated box consumption, probably peak, so it blew up when I switched it on the last time...) Since Fedora 6 I could use the usb box together with the Dune tools, which I compiled and then reasonably reliably connect over the use connection. Hours of TV watching over the local network are norm, not exception.

   Original size from hardware mpeg2 encoder played with mplayer

   Example test movie after mpeg-4 encoding played with Quicktime.

You can download the above test movie yourselve from here (  ) and I tested it on a Communicator, which indeed plays it wonderfully! It does probably require a mimetype upgrade on the server, because it downloads a bit cumbersome.

Recently I've added th mp4 mimetype for apache and now the mobile downloads neatly, no problem. Currently the can is live (but mayeb I'll change things again), and the last 10 seconds movie that has been made should be downloadable from here.

Based on the Dirac transcoded HD I tried out processing a resulting frame with the 8 bit Gimp, getting some of the combination of sharpening filtering and intensity variation things to work like likable pictures, even from long ago:

Or a (not just desaturated) black and white TV or movie effect:

the effectiveness of the effects also depends on the monitor, and of course I don´t have 4k materials to reduce to 2k and then pixelize in ´certain ways´. Wink wink.

So this calls for cinepaint at least, an I would prefer a Dirac (or schroedinger implementation which is the same codec spec) which would use more input and output bits, of course at the expense of bandwidth of the coded signal. I have cinepaint, and also cinelerra can work in 32 bit floating point numbers per colour! That´s great, but cinepaint doesn´t do everything gimp does, like the sharpening filter isn´t adjustable to be more subtle. I know there are also scientific reasons for many things, and that I shouldn´t aks people to send me a free photoshop with my wishes implemented, but then again, I am sure certain things are worth looking after. Like how oneis going to use images; on a small screen, a computer screen (of which resolution woud better increase with time) a TV (higher brightness possibly, and attention for motion) or a cinema grade system, high Q monitor or small or big beamer.

The current solution to have my interest: 4k 16bpc 24 or more f/s recording and at least even bigger (more than 106 inch diagonal) projection with lets say 4000 lumens and great and type-adjustable pixel colors and of course 4k resolution. And I keep waiting for a great job to pay 20k and the rent and ticket or so, too...

Well, and then the Lexicon USB AD/DA converter fell of a small table and again didn´t work good, which happed since I tried to pull the knobs of the fron real hard. I resoldered the potmeters and it worked again, but this time I start to suspect a break in the printed circuit board, though I did get it to work again. The F8 on the notebook with mplayer and the new dvgrab allowed me to do real time HD viewing from the cam, which I used to inspect the lexicon pcb´s, at least to look up chip numbers, of which this is a screen dump:

3D Is not a struggle at the fedora 8 notebook much because I installed a driver which is enough from NVidia (although I think it came over LIVNA repository) and I put on the very latest Gelato and the latest Blender, which works great.

The above is an older gelato rendering of a web found model, the image was onlyin 8 bits per colour but it looks kind of funny, and I processed it a bit. Clearly it isn't rendered perfectly.



Oh and finally, regardless of wether I find the copper lead crack or whatever it is with the lexicon, the pulseaudio matter is still a struggle, but I have 3 heavy machines with the pulseaudio deamon running, and some music software on, and jack and qjackctl and I experimented a bit with the pulse/jack library which is a struggle and the whole distributed audio/midi is still a total struggle, altough I did get the distributed server to work by using a paplay command with music from a file, that can be played from another machine with the 16 bit wav audio coming from another machine. The pulseaudio itself works, with varying stability, the latency isn´t very long it seems, but there are hickups, the pulseaudio live mix/volume controls aren´t smooth on all machines that is some hold back when moved, and I am not so sure the streams are as reliable and stable and therefore high quality as the jack or espeically engenuity with jack I had on FC6. And as noted, I didn´t yet get jackd to work with pulseaudio as a whole setup. Maybe I should restart pulse after starting jack, with the right startup script and then maybe with the right user ( su can do that probably ) it should be possible to get pulse synced by jack and it´s IO terminals visible in jack, and then hopefully the broadcast UDP stream conversion would work too. I have restarted pulse as another user I think and had loaded in jacl connectors, but that setup didn´t get stable. It would be nice to know when audio streams are sample converted, too, pulse can do that but I´m not sure it sais when that happens, which is not straight enough for my use. A clocked network interconnected system with one AD/DA would be OK for now, for my uses. Although a well coupled set of DA converters is cool for surround, but I don´t know if that would work.

Loudspeaker distortion!

Yeah! Well, uhm, no! Of course not! My spitzenreferenzklassesystem shouldn´t have that of course. Well sorry, every loudspeaker (even headphones) exhibits noticable distortion. Period. You could take an expensive studio headphone or an expensive HiFi system from the highest regions of the HiFiStereo maganize tests including exhorbitant prices (like a 18.000 euros JBL Home Cinema speaker) and still if one would take a good measurement microphone and for instance a decent spectrum analyzer, distortion in electronics sense is easily measurable, I´m sure. I mean I recall long ago already spectrum analysers would have a dynamic range of 160dB, which implies an accuracy of 8 decades or 0.000001 percent measurement accuracy, which also is virtually never present in audio amplification, or DA conversion. Honestly, it happens to be the case that my recent projects (2 and 3-fold multiamped amplification systems) make use of a preamplifier Operational Amplifier chip which at amplification 1 and 1 kHz and with everything working good (..) does come in the direction, see the specified harmonic distortion in the relevant portion of the datasheet which I quoted below (there´s even a free spice model on the TI site):

   From the BurrBrown / Texas Instruments Datasheet of the OPA 627 OpAmp

But that is not the power amplifier needed to drive the speaker, or the DA converter (even my fairly good reference being a Lexicon 24 bit converter I recall has about a few thousandsth of a percent distortion at a good frequency) because it is rare that these specifications are electronically speaking hard to measure.

Speakers are known to generally speaking easily distort a few percent, and then you ae really lucky. So speaking as an electronicist (which I am also by education) ´duhhu´ :) . The types of distortion of a speaker are different so it still makes sense to reduce all kinds of electronics and digital distortion magnitudes because even with high quality equipment, low distortion levels can still often easily be perceived, although transient response and other issues like wire resistance and grounding contribute, too.

That makes speakers still by far as they called in good audio books I read as a kid in the seventies the weakest link in the audio reproduction chain. Maybe room acoustics should be added, too. And therefore probably the most rewarding components to improve, just like spending half of a HiFi budget on speakers was supposed to reflect!

I use my own current speaker setup with pleasure and pride because they work very good, even at times extremely good, depending on what material I feed them, but even Guns ´n Roses from a (digital) cable company media box from the VH1 channel with various coding and (non-...) compressors working right can sound short of really great on the system I use the most, which is great when combined with a good beamer projecting faithfully on a large screen (106" diagonal, pro grade).

This of course could be totally subjective, and also based on coincidental results of audio materials with equipment, but I defy that suggestion because that´s not how I made and improved the system, and I did measure some things, and certainly used straight audio sources to test it (like a Yamaha S90 synthesizer producing straight sine waves over the whole audio range with it´s 20 bit DA converter). I used measurement microphones with good preamps and digital meter and a Audio Technica 2020 microphone and a very good Lexicon preamp and 24 bit AD converter to measure some things in the past, for instance to determine band limit frequencies for the 2 and 3 fold multi amping I use.

I would not acknowledge that I did extensive industrial grade dead room testing of my setup, and for the 10 power amplifiers I rely on factory distortion specifications from the chip datasheets, which are very good: 0.005% THD at 1kHz @ 50 Watts RMS, and they are DC coupled amps working like operational amplifiers with FET outputs, having wide frequency range and good slew rate. I built the 4 way + sub speaker system on the basis of a lot of know-how and component testing and they behave a lot like I´d expected them to, so that I think is a good sign, and I love to do the simplests tests of all: record or obtain neutral materials and play them back to check for High Fidelity: when it sounds really close to the original, without tricks, and similar volume, the whole chain works good. So recording and playing back voices, bird sounds, noises, bands and using neutrally recorded material like Chesky audio CD and other recording with good microphones and studio equipment I am at least satisfied that such test works pretty good on my system.

In the context of speaker distortion I visited the dutch AES lecture from Dr. Kippel from Germany recently which was about the measurement equipment Kippel Gmbh produces and an analysis of speaker distortion which concerns a 2d order approximation of various effects, of which he clearly examplified inductance variation of the voice coil through excursion and the maximum power  which can be applied to a speaker, especially at low frequencies during a distortion correction using the aforementioned speaker model.

Honestly, it all sounded like (uhm) total crap (I mean for lets say a serious HiFi enthousiast), but that I am sure that was because of the demo speakers being PA speakers and not intended as studio reference or so. The estimated and exagerated distortion demo was very clearly audible, and it was as I thought: 500 watss (as claimed) on two fairly small PA speakers with I guess 2 8 inchers or so per side just isn´t going to give you smooth and deep working bass, those poor cones are too small even if you´d use a kilowatt on them to move enough air! So Tracy Chapman even when distortion was analysed and compensated somehow still sounded like coming from resonating too small enclosures and only with hyped and not low extending bass.

But the correction effect was very clear, and the equipment could nicely measure cone excursion and impedance change as a consequence and even small speakers were demonstrated which were measured to have very slow return to centre characteristics and that was live-measured. I am sure it would be cool to hook up my speaker setup to a good measurement unit, but I am not sure what would happen with my closed but big pressure encloseres with damping and the 4 speakers from 12 inch through dome in them. Could be exiting to try, though I think I´d also like a good (lets say a 20k$ HP) spectrum or wave analyser with calibrated measurement pressure sensor and a good working dead room. What a challenge. But they would I am SURE be able to amplify a signal up to 10 Kilo Watts with the abovementioned distortion if you hand them sufficient funding (lets say millions) and they feel like it. Yeah. Great.

Anyhow the lecture I´m sure was very good at myth-busting (I´d have prefered to have been in San Francisco to see it!) and of course a lot of distortion of various kinds can be (automatically or organised) found in all kinds of speakers and to measure and even partially corrected: that is a good thing.

download: mpeg4 video  ( 14 sec., 3.4 MB )
The superwide (1:2.2 or so) format is chosen such that a Nokia Communicator mobile phone with wide screen can exactly play this video, and also the codec is correct for that phone, be sure to try if you have one, this does NOT look like a cheesy ´phone movie´ (it was made with a HD camera) and the internet can let you download this quality whereever you are and the player is able to let the whole screen be used for movie playback. Beats CNN-mobile!

Of course like inthe screendump above a normal not too old PC can play the file with Quicktime or mplayer without problems, too.

The original (16:9) HD recording is about 56 megabytes (if somebody wants it it is here).

What would the Kippel equipment cost...?

Redone Lecture Videos

Coding video and audio one always runs the risk the receivers can't decode the most perfect or just possible version of compressed material the maker came up with..

The example of Images applied to Music.

Well uhm, so can I afford to make such comparisons? Sure. I´m more than enough editfied, talented and skilled in either, but it is the big picture here I´m writing about because that is where appearently everything new age is a big nasty taboo.

Lets say I cannot compare great 70´s posters, the Classical Paintings, Dirty Harry (let´s say) and the whole of Rock, Jazz and the better Pop (50s, 60s, 70s and 80s) in ANY WAY AT ALL with the new age sh*. Sorry, no can do.

So comparisons of the nonsense, the sin if you like and the Good Things are out of the question, no matter how hard the rigged and uninteresting hit parade tries "we have got to be taken serious...". It´s mamma´s money only so who cares.

So it´s all about the money? I think money, power, fake religions, patterns of the wrong kind (idolatry) from various sources and of various implementation, and it starting to eat on many people that that is showing and that a gouvernment of fear cannot stop the realisation of such facts.

But for people in trouble or who must maybe even convert from adverse and damagind views it may be important at times to consider Where It All Comes From, and how certain worlds of sin and terror hold together, and to get out: How Not.

House and rap ain´t cool, never have been. Hip hop is fascism, which is not particularly interesting and certainly emotionally straining. Rock was and is cool, and still is interesting after decades while the New Age House was crumbling before it started and only attracts real attention from mamma´s boys and deluded whores when things are more normal. And Pedofiles, possibly, because they get patterns to play with. And of course all the evil (as opposed to ´inclined to evil´) people and their organisation forms have plotted for centuries to make as many people as possible take part in the crime of the 20th century, to take over the good music with devil (literally) inspired sh* which can be used to kill all that is good and nice and worth while in the name of the beast (computers or Europe, which ever would win I guess) or because of the great whore from Rome with the more smalltime whores mothered under her waiting for her punishment pimp, the beast.

Sounds stupid ? would to me, but also dramatic enough to have attention for, so when I started to be convinced of biblical truths long ago I sure paid attention to the big picture.

But Theo, the christians know all this, and manage the forgiveness just fine! Motherf*. Sorry, I need that word, but I don´t want these pages to be banned for undecent language, that don´t serve my purpose. That is the sin of christians, like

   Some of them that work forces (perform miracles) are the same that burn crosses.


   Yo, it´s just another bombtrack

and all the other great lyrics from "Rage against the machine" (The beasts from the sea or the land which come up and go to perdition)

Holy roller lingo? No. Biblical language. So if you call me stupid or wrong, especially as christian, than many people claim to believe in the same, and as christian: I challenge you to put me straight, seriously.

And I don´t really care, I don´t need the new age stuff, I really don´t need anything they got. So they want the scraps and the garbage form the real art and electronics, well, who cares. Their problem, I don´t care. But the delusion of a lot of people especially when they don´t live all to bad or just want normal things bothers me, and I think I don´t want to meet or see the implementations of the sh*. I don´t find it a pretty sight and a hot one hundred filled with fascism and offence and scoffing makes me sad.

Europe and Information Technology you said? Eh well, as opposed to Mary and het little Lamb or so. I think that is what I just wrote, isn´t it. Well, you may not say that! Oh, really, in the free world?! Of course, imagine you´d say things that offend Our Conservative (read ´satanist´) leaders! So what? I did all the time already, won´t make much difference, won´t it! But then if we agree we would suffer with you! Oh, is that is, am I suffering, I though I just didn´t have money and everybody is blackmailed all the time, sorry I understood wrong. So: I am suffering and you guys are not up to it? Or, I´m better and y´all may not admit the good things aren´t of the secret sneaky self ascribed motherf* leaders ?

Hmm, lets get to some more meaningfull discussions usong content, such like why are my image considerations comparable with the just brought forward considerations and what is parallel to this in audio or music?

First of all the comparison is that both images and audio are probably about art, and about things that many people like and influenced by. That of course is a main target of what have we: wannabee world leaders, the catholic hierarchy (dominicans and jesuits alike) and all the religions following, the antichrist (I guess Hillary C. should watch damien II) aka Nimrod (the big Negro ´man´ supposed to help Cybile, the goddes of worse evils than I like to write about) and all other big time organised badness threading the grass of normal people.

Second: both subjects should live in freedom and both are terrorized into submission and a purpose which is subdued to serving the alledgedly to be evil rulers.

Third, there is a great desire, probably from the antichrist-likes and the computermotherf*s (I think the word at least sais some truthfull things about the actualities, I hear it was popular when the nazis were fought, too) to include the subject of sampling and messing with digital representations in both audio, samples and music alike, and pictures, both moving and still.

The above is a screen dump from a notebook making sounds, with good quality thusfar from the lexicon 24 bit converter, and there is a fast usb midi connection driving 2 DX7 simulators and an analog synth simulation, with both sounds having their own effect rack to process the sound with LADSPA effects connected over the JACK sounddeamon. That is a lot of stuff! Sure, and it seems to work stable for hours and certainly this didn't drain all the processor powers so that I added rosegarden recording and made some small test recordings.

The next is a combination of softsynth sounds with LADSPA effects to do amplifier and speaker simulation, and reverb and tapped delay IIRC with some mixing

    synthcombo  mp3   wav  (1 min 8)

The following is a test in the context of the Listen To Keyboards site, which compares top synthesizers/keyboards in terms of some of the classic sounds on them.

I did a bit of one of the pieces of music myself on a Yamaha S90 and compared the recorded wav (uncompressed CD quality audio) and the mpeg compressed result.

   pianosound    mp3   wav  (33 sec)

As I expected, even fairly high bitrate mp3 isn't anywhere near the sensation of the wav file. Even on a good mobile phone, and especially on my big multiamped monitor prototypes with milage.

So any good records as a teen ager ? Sure.

These LPs I bought (or got as present) (and had) myself from around 80 to about 85, LP´s mind you, not recordings of, or inherited old albums or CD´s or music cassettes. What a covers, compared with almost anything new (are printers not so good anymore because they´re digital ?! Uuueeehm, these are digitisations, aren´t they....):

Jimmy Swaggart "Live from Nashville"

   (Graham Kendrick, "Songs of Fellowship": Let God Arise.)

   Amy grant, this one had some absolutely great music, even jazz and some synthesizers.

  Kurt Kaiser. A guess buy which worked out good, good piano playing with at least one piece which should be considered a classic piano playing with I seem to recall bass and drums with a sort of lyrical jazz like turn in it, probaly it was the song Pass it On, first on second side.

Need I say more, I bought this  one together with a friend (to record it on tape with good quality), on if the great works of their classical rock, next to A Night At the Opera.
   A lesser known but equally good Kraftwerk synthesizer album. A great influence of mine in the 80s, when I made my own music electronics.

  I think this must have been after the 70s. Great work of art with the most challenging musical instruments: synthesizers (and computer equipment to go with it).

  Probably the first LP I got for my birthday, at my request. The guy who gave got a good copy too, I´m sure.

  Probably an underestimated band, they still make music, mainly or probably only with synthesizers (and drum computers and sequencers and studio equipment). And the human voice. I loved the music on this album, on a good system with CD player it sounds great, too, probably mps don´t. Most well known "I just can´t get enough", which iIrc was with emulators (sample machines). I liked the analog stuff the best at the time, musical, spacious, interesting, challenging (and not only rythmically).
   The best LP album I had myself?! Possibly, depending on mood. I had made a good quality record player and (self made) preamp, and this could funk my 2x70 Watts driven (self made amp) speakers (self made) and room, and mind. Great music.

God's view ?

(Old american TV voice "You´ll find a pair of gloves, put them on.")