Theo Verelst Local Diary Page 39

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

  Fri Aug 10 14:42,  2007

That´s when I started this page with some reports of my activities and personal thoughts for sort of public consumption.

This is not a blog! It´s called a diary page because that used to be what it is, maybe it is more like a report of my R&D activities (mostly Open Source and currently non-commercial).

So what am I, a unpaid programmer, like the people/person from Cinelerra (a open source and free video editing program for Linux, also suitable for HD) ? Well, unpaid is true, but that big project is not by me, although I´ve compiled that program myself that is hardly the same as having programmed it ...

I´ve also made hardware even open source (programmable logic, schematics, see previous pages and synth page) and the communication with mr Stallman a number of pages ago, and I guess both the hardware and the software aim to raise awareness about better things in life and surely to promote myself to an extend, or correct wrongs and wrong opinions, and of course for fun too.

I got this fortune from the server where this page is on:

Fortune from Mon Aug 20 13:29:43 CEST 2007:
Mathematicians are like Frenchmen:  whatever you say to them they translate
into their own language and forthwith it is something entirely different.
-- Johann Wolfgang von Goethe
(At the moment the fortune script is not in place because I'm in the transition from FC6 to FC7).
Various pictures on this page can be clicked on for a larger/onprocessed version.

From Chic to AC/DC

How come that this:

at times has been replaced by this:

Because I´ve become attracted to the principle and practice of getting the good music of many sorts that I have available on CDs (and on discs) in a better state by decompressing songs.

I start with a (reasonably legal) CD copy which ´ripped´ to a wav file (for personal use) on disc, which I´ve done for about 1500 songs and then the harder part starts, where I utilize Cinelerra to process the audio from the CD track in 32 bit floating point (throughout) accuracy, though of course the original is 16 bits stereo.

The step to arrive a great audio for my main audio system (4 way per side + 15" sub, three way multiamped with very high quality analog components driven by a 24 bit Lexicon DA converter on a Fedora 7 64 bit Athlon 3.3GHz Linux system) I process a song I like with respectively a three band parametric eq, a graph based compresser acting as a decompressor, and another 3b eq, and some additional effects, like gain controls and vu meters, all available Cinelerra, which originates from the audio processing realm, and does no simultaneous video processing.

The idea is to raise the leven of the sub-low in the CD recording to hopefully the original level by a large and shape controlled low-lift, and possibly the same or the converse for a part of the mid range , then with the right attack and decay settings decompress the material, again hopefully to something resembling the recording before mastering compression, and then another equalizer lifts the low and a bit the highs because often these suffered in the mixing to get more headroom, and possibly a correction in the mid for fellow-compressed mids or compensated mids.

In short: lift sub-low --> decompress --> further equalisation (aimed at neutralness of the result).

Does this work? Yes, often it works great! Absolutely great even, all that popular CD compressedness out of the way, the sublow back where it belongs, less compressed voices, restored high levels for older recordings makes for very good listening on my ain audio system, and also the mid-sized bi-amped monitors I made sound great this way.

But the for the big system the benefits are the greatest: live experience is completely possible after carefully trying to compensate with the right timing and Eq and curve settings.

Mind you: I compensate for the recording/mastering/CD pressing process, NOT for my speakers, that is my monitors are fairly good and straight in their response, I don´t create  bass lifts or compression effects to flatter my audio amplification setup (which normally wouldn´t work much, and frankly there SO much power and low present in it, that the big systems REALLY doesn´t require such work), but I try to get at higher fidelity by trying to decompress the audio back to lets say before the mastering which is done also to make the music work on smaller systems and the radio, and were perceived louness cannot be traded in fotr lets say 12dB more headroom to make the lows come out right.

Hoping that isn´t a problem I´ll put up an example here (if there is a problem: mail me, and I´ll remove them) , which is only a short piece of the beginning of the well known hit Le Freak (by Chic) , which in fact is decompressed even in two seperate frequency bands, the low and the mid ranges, like below:

here are short examples of the result, in 16 ir 24 bit wav files (the processing is in 32 bits floating point IEEE format, which has 23 bit mantissa)

    decompressed Le Freak intro:  16 bit wav  (9 MByte)
                                                      24 bit wav (13 MByte)

The graphs in the parametric EQ windows are probably not completely correct, though the controls appear to work as I anticipated. The graphs in the compressor windows indicate decompression, and the curves and the timing parameters are result of trying and listening and tuning. untill the result sounds natural. I do have good referance material which tests great on the system like cheski audiofile (unprocessed) CD recordings of various kinds, self made recording with neutral or neutralized microphones, and than the setup can sound quite extremely neutral, so it is likely the music as processed like above indeed has become more neutral and not hyped for a certain audio system setup.

And as I mentioned: the results are marvelous, the only way getting closer to a real AC/DC Black in Black experience is a real concert (or lifting a severe deliberate volume limititation that was build in the current system and/or setting it up where 100s of watts can blast without causing trouble...).

Also note that this decompression idea works best on real good systems, probably all the better on systems with actual big sized woofers so that they can reproduce low frequencies naturally at sufficient volume to be heard, which is often not the case on certain types of modernistic equipment. In general: the more HiFi in good tradition the system is, the better it will sound. Even on a not so huge but igh quality system, you might hear the difference between the 16 and the 24 bit version of the above files though. And the 16 bit version sure sound alive and clear and solid on a notebook even.

What music have I tried, well this is list of cinelerra project (xml) files which I have saved after the songs ´worked´:

abacadabra1b.xml             hasta.xml                remedy.xml
abacadabra1.xml heyjoehendrix.xml rollbeethoven.xml
acdcifyouwant2.xml honeyhoney2.xml roxane.xml
acdcifyouwant3.xml honeyhoney.xml sade3.xml
acdcwalk2.xml ideofunk.xml seeyoulater.xml
acdcwalk.xml itsbynnereel.xml seeyouwhen.xml
backhome.xml iwasmade.xml sforhtedevil.xml
barbed1.xml jacomedley10t.xml shaft.xml
barleestr2.xml knowingmeknowingyou.xml simanandgamerica.xml
beach2s2.xml lateintheevening.xml simonandg2.xml
beach2s.xml lefreak4tr.xml simonandg.xml
bigfan.xml lefreak.xml sofine2b.xml
billhaleykansas.xml libmpeg3-1.7 sofine2c.xml
billhaleyrazzle.xml littlebluebh.xml sofine2d.xml
billhaleyrockaround.xml livelyup2.xml sofine2e.xml
bloodywellright.xml livelyup.xml sofine2f.xml
boogywonderland2.xml longsummerhendrix.xml sofine2g.xml
boogywonderland3.xml miller_teentown.xml sofine2h.xml
boogywonderland4_2.xml mix1.xml sofine2sade2.xml
boogywonderland4_3.xml mix2000-1.0.1 sofine2sade.xml
boogywonderland4.xml mix2005-1.2 sofine2.xml
boogywonderland.xml movharbour1.xml sofine.xml
bornonthebayou.xml movharbour2.xml sternplay.xml
brickhouse.xml movharbour3.xml stillcrazy.xml
bridgeovertroubledwater.xml movharbour4.xml surround1.xml
car3.xml movharbour5.xml sweethitchhiker.xml
chant.xml movharbour6.xml tipatinas.xml
chicforbiddenlover.xml movharbour7.xml train2.xml
cinelerra-2.1 movharbour8.xml trenchtownrock.xml
commodoreswonderland.xml mrsrobinson.xml water3.xml
dudu1b.xml natu1.xml weide1b.xml
eagles1.xml nightshift.xml weide1c.xml
eenden1.xml oasis2.xml weide1.xml
eenden2.xml oasis3.xml wholelottashakin.xml
equinoxe2.xml oasis.xml wouldyouplhendrix.xml
equinoxe4.xml oldfriends.xml xmovie-1.9.13
feelinggroovy.xml outtatown.xml yellowj2b.xml
firehose-0.6.0 playthatf.xml yellowj2cloudness.xml
georgia.xml pocketcalcu.xml yellowj2c.xml
goodtimes.xml quicktime4linux-2.2 yellowj2.xml
harb6b.xml rebelsarewe.xml yellowj.xml
harb6.xml remedyb.xml
All files which remind one of a pop or jazz song are decompression projects, all of which listed have worked satisfactory, the others are HD video editing projects.

On smaller speakers, even on a (good) notebook some of these results sound good too, but of course the bass end will not be so great, and remember this is not an optimisation method for getting a good response from ´certain basereflex´ systems, it is even quite possible that not so straight and boomy systems respond poorly to the same signal, because the assumption of these results is a straight system, not a bass reflex system, and also not the big woofer disco setup where the amps are slow and the tweeters few, or the well known not too big PA systems where power is present but not always nice and not full range-straight.

For people who don´t really know what I´m talking about: why does this work? Well, when listening to a record, a CD or even a TV program´s audio, on the production side of the audio chain a lot happens with the signals, and usually this includes what is called compression which in essence means the softer passages are made louder and the loud peaks in the music or sound are made softer by a certain amount. Many people will know cassette recorder or video camera recording where its is often clearly audible the virtual volume knob during recording adjusts itself automatically during louder and softer noises being recorded.

The professional compressors do something similar, with accurately adjustable timing controls, and a settable amount of compression, and with various volume sensitivity curves. The reason is for instance to make a voice more intelligeble, to make a piece of music sound louder without the meters going in the red, and often to create a sound of a certain nature which also works good to produce a certain type of bassiness on small speakers.

For the HiFi thought compression isn´t good: on a good hifi system the music or other recordings are not going to sound more natural in most cases it is better to strife for a audio signal path which is as straight as possible: a good exact microphone, good quality mixing and recording equipment with straight frequency response and little distortion, and of course the very good properties of the digital audio CD (compared to the rumbling and fluttering and distorting pickup), a good CD player with straight and well design digital to analog converter, a  neutral amplifier and capable and neutral speakers (used to be good to cost half your audio budget).

The result:  sitting decent in front of the speakers (set to ear height in an equally legged triangle with the listener, and having damped annoying room resonances) the symphony orchestra will sound almost like you´re there in the real concert hall, the rock band will sound like try are performing in front of you, etc.

Sellers of records or CD´s or DVD films or so will often not want their product to sound soft (this cd is broken or not up to current recording standards...) or with large dynamic range (asks much of the superficial listener) so compression is used to make the soft parts louder and the louder parts also a bit more intruding by making the volume peaks lower and the rest louder. In various cases, the compression is part of instrument amplification, like with electrical guitars.

The individual instrument compression is hard to compensate, and not at all necessarily likable, it is the overall compression I´m talking about. Assuming the compressor has a knowable volume response curve, and has certain known types of volume response time contstants it is to some extend possible to take a CD recording, get the data of the stereo tracks in a computer, and reverse the effect of the compressor. As a result the softer parts of the track will sound softer, and the peaks are to some extend restored to their original level, and moreover the sounds will in general sound more natural, but then it has to be said you need a good hifi system to really appreciate that idea.

The low frequencies are often present up to only a point and relatively soft, the bit lower still and so called ´sub low´ frequencies important in concerts are more and more soft on the records. However, simply turning the bass control on an amplifier up will not work satisfactory, because the way that control works doesn´t match the low response during recording, and especially because the effect of compression can only be undone by uncompression, if at all, which is not the same as a bass lift. Finally, the whole music is often influenced by the compressor in the (master) studio responding more to big volume bass sounds than most other parts of the spectrum combined, so to decompress right, the low range must be restored to how it sounded after  compression, and not be as relatively soft as on most pressings, which means especially the sub-low (very low frequencies lets say 50 Hz and lower) must be considerably lifted.

Don´t thy this at home? Well these computations are no simple and easy plugins, cinelerra does 32 bit floating point processing of quite a complexity. And the listening? Well, try the chic example, probably you´ll like it better, expecially on headphones when your stereo isn´t really top material. A small Senheiser I tried sounds magnificent this way, but uhm, well, uh, the system I try it on is like ahem really extreme, so to get the same is not likely to work on normally available audio systems, the 24 bit Lexicon and my own pre amp with the world´s best audio preamplifier chip it are probably not beaten by most anything. And my three way amp setup with excellent analog filters and the size and quality of the speakers are very far from average, and the power, ahem, uuuuhhh, well there´s rougly a kilowatt continous rated power toroid transformers in the whole setup. Good things sound better really loud....

Anything else, Theo? Well, (hihi) I also did a 6 seperate tracks accurate decompression project with the world famous music of the french synthesizer virtuoso Jean Michel Jarre in three frequency bands, where the stereo image and electronic sounds become really alive and very 3 dimensional, remember, the monitoring system is seriously quite extreme compared with most hifi.

Stills from film

It sounds appealing to have film material which is actually made of 24 great stills per second, within reason (excluding motion unless there is much light), well in these days of the High Definition materials that can to some extend be true, such as I prove below: all the pictures in this section are made from HD video (which I call film because I shoot the ´footage´as in learning to film, no to do fancy video shooting, which I already am good enough at).

These pictures are frames taken from a screen which shows 1080i (not p) mpeg-2TS (25 megabit/sec) works I made and reduced in size and more or less processed.

The race track at Zandvoort, I think there should even be F1 at times.

    From the same spot in the dunes, the view on factories at dawn.

   The same dunes at dawn, this is also a frame from a HD recording I made

   From the top of the dune ridge the view west, where the sun just set.

I used a (analog at the time) sony handycam years ago do take a number of stills from, see here
( ) , which is big time inproved on by the pictures shown here (taken from recently made (´prosumer´) HD).

   A duck in the Hague

   A seagull in Scheveningen (near the Hague)

   A fake boat on a sort of carpet

    A less downscaled frame from a scene where I tried out a driving (moving) camera on a big solid object, which looks great in 1080

   The tripod peeping out an open car roof

    The port of scheveningen

Little midis

In my everyday language ´a midi´ would be a file of usually between 10 and 100 kilobytes size which contains a song in .mid representation, which is a file format standard where MIDI synthesizer note and control information is stored on a timeline, so that a synthesizer or keyboard or organ with a midi input socket which is driven by those notes and events by means of a sequencer will play the song which was recorded like on a piano roll, except with more accuracy and possibilities. Nowadays it is more normal to play a midi file with a program on a computer which then must play virtual instruments implemented by software.

The sport I'm presenting here is to get some good midi, for instance from a known hit, and render it with a program or set of programsor a keyboard or synthesizer with midi playing capability, and make the best sounding result with a programor machine. There are of course programs which can play midi files by just clicking on them, like windows mediaplayer, which gives an impression of the song and can nowadays sound ok, but whole lot more is possible with the same midi normally. For instance Timidity on Linux is a dedicated midi player with builtin sounds (Free and Open Source) which sounds a whole lot better then that.

More challengingis to take a sequencer (software or even hardware which can output the midi stream real time to tone generators) and drive a set of sound plugins or hardware sound units (keyboard, synthesizer) with the midi notes and mix the used tone sources and apply effects and make it all sound good. Here are examples from two midi's, one transscribing the famous 80's hit Vienna, the other playing all the notes from ´Wasmasjien´ a hit (in Holland at least halfway the 80s) from quite some time ago from the I thought French but probably dutch/surinam band Traffassi (See previous diary page for the midi). I got bot the Midi files from the web, not too hard to find, I don´t remember who made them, but I think they´re pretty good.

viennafrommidi [.mp3]    'Vienna' from Ultravox rendered with a Yamaha S90 synthesizer (the previous version).  No additional processing if I recall correctly, but I did adapt the initial mix from the multitrack-like S90 setup quite a bit, also the effects, and I had to manually set the instruments to match the intended GM sounds, which were present in the S90  but didn´t automatically get chosen right by the midi Program Change messages however, and probably I tried some alternatives here and there. Don´t forget monitoring makes a big difference, too, like in this case I used the big sized (tri-amped, 4 way + 15" sub) audio system I made, and the medium sized bi-amped monitors to listen to the mix. There is more than enough natural (sub) low present, but of course the medium sized monitors which correspond better to home systems can´t reproduce all of that, and could be ´helped´ just like bass reflex bozes could by compression, but that doesn´t happen: there is some builtin compression in the S90 sounds, but no additional equalisation or compression or mastering multicompression or so are used, which generally is a much better idea and sounds more relaxed and natural (and compression is a drag to do nice in digital domain).

WasmasjienTrafassitim_256 [.mp3]    Traffassi's 'Wasmasjien'  (Washing machine in a caribian like accent) rendered by the open source program Timidity, and after using 24 bit audio turned into a 256 kbs mpeg by ffmpeg on Redhat Fedora Core 6/64bit, which can install timidity by  using the handy 'yum' command easily.

wasmasjien1 [.mp3]    Wasmasjien with Rosegarden mostly using a big (64 MB) General Midi sound canvas, and rosegarden recorded singing of two lines (by me), the silences are when rosegarden´s plugins take too much processor power. The sounds are not perfectly chosen, there are some FM sounds which at least sound a bit interesting, and the whole is nice but it was an early try.

wasmasjienmainlys90_256 [.mp3]   Wasmasjien using mostly S90 in sequencer (multitimbral) mode and some rosegarden sounds, recorded mainly in one take using a Lexicon Omega AD converter. I´ve included a short vocal track of me singing, which was recorded using an Audio Technica 2020 microphone (Americas most popular cheap professional microphone at the moment) using the Lexicon preamp and AD converter, and processed with rosegarden and LADSPA effects.

wasmasjienroses90revnocomp_256 [.mp3]    Wasmasjien with fluidsynth filled with Free samples (a number of Fluidsynth instances), Hexter (Dx-7 simulation), various LADSPA and S90 effects and S90 sax and steeldrum sounds, and a few added notes by me at the end. In this version I did some mixing, and I used subtracks to offload a number a submixed tracks from the main mix CPU usage (reading a 2x32 bit floating point (!) stereo audio track takes less CPU then rendering a number of midi tracks, and the disc is fast). I didn´t use mixer automation (didn´t try) I just mixed like ´live´ mixing while playing back this version. More than a few effects were used, and sometimes seperately recorded. Project duration was half a day IIRC, not much or no compression was used during mixing or track recording.

I guess these files which are mostly produced by me are free to use, except commercially, I don´t know how that would be with the midi file makers (which is hard work) and with the original artists (even harder work...) !

I´ve once asked Yamaha what they thought about sampling their sampled instruments, and the answer was more or less ´long live free experimentation´. Sampling some pop/rock/fubk classics for even two seconds or so I strongly recall was stricktlyforbidden by law for some ´you know what kind´ of wannabe artists. I guess I should´ve paid (IIRC about 25 dollar) for playing CD at my public audio system demonstration. Probably I can be forgiven in the same line of when I long ago called the tax office (at the time still in the Hague) and was informed I didn´t need to pay taxes over my (extensive) vacation job´s income, because there are lower limits anyhow.

For the latter audi example I wouldn´t say "don´t try this at home", because why wouldn´t you, but especially the last example on a big, quality sound system isn´t easy to get done very well (remember these are mpeg sound files, not too compressed in size, but certainly not like the wav (PCM) file I´ve presented in the previous page).

So is that the end, the synthesis of great sound and the use of very good samples? No, I long ago already found it very interesting to play with various forms of synthesis combined with added effects. Ingenuity and the linux music software allows that, although it has a few quirks and bugs. It is absolutely great though to be able to make a sound processing graph with complicated and quality effects and multiple sound sources and mixes which actually works in real time (I guess about 10 ms delay with respect to the (USB) midi keyboard).

The package to my knowledge is not generally available, but is free and Open Source for people to download and compile themselves. Actually its for advanced Unix (praise to their work) / Linux people to do that, more likely, because it requires uncommon libraries and a as yet not frequently used development version of Jack (the audio connection program) which needs to be compiled first. This can give conflicts in the installation because the jack library is used by many programs, and the new version 0.107 isn't compatible with the mainstream 0.103 . I've moved from Redhat Fedora Core 6 to Fedora 7 because I had all kinds of packages updated by Yum and the dependency tree got stuck because I had two camps of libC number dependencies and they couldn't be resolved. I did get Ingen to work at FC6 with a lot of patchwork and using two libdir settings but it wasn't easy. On FC7 it was easier also because there were more builts of required packages with development files available via standard yum.

Still, probably only for experts I think, probably I also changed a few lines of code to get it compiled but I don't recall exactly. But the result is marvellous: a interactive (live-changable) graph with audio connections between all LADSPA blocks and the system IO channels available to jack. The current version has problems though: midi connection with synth blocks don't work good, and when an Ingenuity subblock is instantiated in Patchage (the main graph with the system streams) and the inputs are connected to the instances´ pins it is required to reconnect the input pins inside the patchage connection graph but then: it works!

Complicated graphs are now possible using jack (but a new, not publically used yet version) and all the LADSPA and other plugins and synthesizer programs, also with stereo signal paths (just connect L and R), which is not possible with the effects in Rosegarden for every routing, and is quite powerfull, since the whole set of usably plugins is growing and quite usable and nice, including reverbs (gnu and plate), delays, phasor, various types of compressors/limiters (though no less than 10mS attack setting thus far), special effects, vocoding I think (didn´t try), multiband parametric equalizers, and distortion effects.

This is an example setup (click for real size screendump of 1920x1200):

I used a seperate Qjackctl (it isn´t easy to make one working with the requires jackd protocol version) to connect midi links, a tryout of the above sound in mp3 form:

ingentry1 [.mp3] Live played electrical piano made with 3 layered (16 voice polyphonic) hexter Dx simulation sounds, combined with a strong Zynsubaddfx synth sound which has its own reverb. The DX sounds are panned in a small row in the center, and are given two types of reverb (gnuverb and plate).

Very pretty though I didn't try to use it subtle.

The IQ issue

Some time ago while browsing the web for some purpose I forgot, I stumbled upon an online IQ test, the classical one, which without being al too serious or concentrated rated me at 130, sort of a geometrical mathematicist. Well well. Funny of course, and probably only an indication but it made me think of the following:

Take a population of n people with Intelligence Quotients: Q_i (0<i<n) , then

thesis 1:
   the IQ of the population as a whole thinking about some problem is seldom lower the floor(Q_i)
thesis 2:
   the average IQ due to miscommunication and strive is probably less then average(Q_i)
thesis 3:
   (my favorite!) the maximum average practically measured IQ of the whole population is likely to be quite less than ceil(Q_i) and more likely to be around average(Q_i)

Isn´t the kiss-kick-kick theorem layout better then kick-kick-kick ? Anyhow, Of course people could shrug and think ´whatever´ about such ideas, but in lets say families where intelligence is everything or where kids are supposed to be little hitlers or Einsteins, I suppose the ideas would do good in this time period where quantum physics´ determinism are again challenged but now by the darkly-inspired new age ´think´ers.

If you can come up with better thoughts or measured results, don´t hesitate to email me... Of course these are thesises, not acclaimed proven theories, for the advancement of democracy, not to bash managers with or so.

The inspiration for this thesis? Well, I suppose once again the insipid New Age and the evil inspired forces which implement leadership in it. Luckily the 70s  are returning, and soon a lot of the cr*p people are into even in Europe will have dimished and hopefully a lot. In the good parts of the 70s, people would not motherf* their own and other peoples lives to pices and the idea that life is nothing but a knowable and ratherinferior propgram would not have to fought by me by such thesises as above, which would then probably be considered fun. Currently it is meant both corrective and fun. If the shoe fits they say...

More pictures

I tried a tablet out, a flat machine where you can use a special pen on top to draw or write, or as a mouse controller, where I am trying to get the effect of charcole, which isn´t really working:

I´ve done some years of drawing at art academy in Rotterdam on saturdays (nude models and abstract) and I must say the real thing is uncomparably better although general drawing works with the tablet, and there interesting retouching facilities in the software that came with it, which certainly works nicer with the pressure sensitivity of the pen turned on. The specified accurcay is pretty good, but the line following accuracy in practice is somehow, maybe by the driver software, limited in ways a bit hard to describe.

Another accurate graphics game to get pictures is 3D rendering.

This image is from a 24 Megapixel (no misprint) image I made using Gelato, to render a 3D model for a printout, the model is from ´´, and free for use:

The original I made of 6kx4k pixels had 16 bit HDR accuracy per colour, but was eerie dark, for this page I couldn´t easily start up cinepaint (which runs on Fedora Core 5) to change the lighting in HDR, so I used 8 bit GIMP instead, resulting in the banding of the colours.

This picture (but then at full scale) I printed as A4, which looks very cool:

It´s from a HD recording in the dunes which looks pretty, just like this picture (which was NOT taken by me, but is from the wikipedia entry on San Francisco):

I processed it a bit and scaled it.


was a term from radio amatures (maybe pros too), where for instance a card would be sent to indicate reception of a certain transmitter. I new that as a kid from a friend who had an uncle, enfin whatever. Later I would associate the term with the famous DX-7 Yamaha synthesizer. Now it is a program from Ibm, which unlike the imago of a bureaucratic commercial molochs is an Open Source and Free program for Visual Programming. I think it has previously been a commercial package now made available for free (as in free speech), which resembles two software packages I worked with (also professionally) called Khoros, which was limited licenced open spource, and AVS a commercial package which I heard a few years ago was still in use at least in a supercomputer center in Great Brittain.

All of these packages feature a blockwise approach, like my BWise program, an all three do mainly image processing related jobs, though I remember Khoros for instance could run general shell scripts in blocks, and AVS had for the time (91 or so) a pretty decent 3D viewer for on a unix workstation. I used HP720´s at the time, which were about the fastest in the world a the time, probably about 10 (ten) times faster than the PCs of the era, and they would do software rendering of fairly complicated (up to tens of thousands of polygons) 3D models in interaction time.

The idea is that some where data is read in the network of blocks, at the left or at the top, and the data follows the graph structure of the blocks to get processed until it appears at the output. Every block does something specific like reading an image, applying a colouring, taking a cross section from a 3D dataset, etc., and the whole (directed) graph can be easily changed with all kinds of block available in libraries of processing routines with various types of data being passed between the blocks like images, colour lists, 3D models and 3D datasets, regular and irregular.

I`ve installed the free Dx on my latest Fedora install which is fedora 7 / 64bit on a Athlon 64 3.3Ghz , where miraculously the open source driver works fairly good on the cheap Nvidia 6200 card so I can use good hardware acceleration to view 3D models interactively, meaning for instance moving or rotating a complicated graphics representation works with direct response to the mouse.

Luckily the 3D viewer can also render to file, which I did to get a few hundred frames from one of the demonstration networks, and I tried to get a little movie from these frames with a newly compiled (from recent sources from the replacement of cvs) ffmpeg:

 ffmpeg -i /dev/shm/1/im.%d.tiff -b 6000 A/t.mp4 
which has made this (5 megabyte) mpeg4 movie

  mpeg 4 movie

which can for instance be played with Quicktime movie player.

A viewing room

I've installed more or less a viewing (if not screening) room, with not expensive but certainly very good quality film (well, film..) viewing and even editing (using Open Source Cinelerra on (free) Fedora 7 64bit for instance) possibilities.

The projector is a tested as very neutral and capable machine on various sites andindeed performs as such, it's a Mitsubishi HD1000, which is 720 HD capable (1280x720 actual pixels), of course this is not as much as the camera I'm using regularly, which is 1080i HD, but a projector with DLP (little mirrors on a piezo operated chip) at that resolution is quite expensive still. I suppose the difference between older video resolution beamers when presented with the right materials is considerable though.


Above the projector, a main use, and an image from the standard cable company media box directly connected to the projector.


I've printed some images from HD and computer graphics sources on a totally cheap but really quite excellent recent HP photosmart printer on A4 brilliant paper (not so cheap), which is good comparison for showing the 'film' (1080i HD video)  they came from on the projector.

The screen is a quite not small (106 inch diagonal, 2.4 m wide) pro grade screen with nothing extra but a good reflective 'sheet' (which as can be seen below in the picture has a curl hanging as it improvised is) with on purpose a ca. 1.5 reflection factor, more than normal white, and less then (probably expensive anyway) 2.0 or so, and pleasant and well performing angular radiation characteristics, namely more in the direction of the audience and less to walls and ceiling, and of course the maximum 1500 lumens of the beamer are even amplified by a factor 1.5.

Directly by the sides of the screen at ear height my major (self built but I suppose that is clear for most regular readers) speaker system with 3 fold (!) seperately amplified, electronically seperated, and using a in total 5 (!) way speaker system, when counting the big 15 inch sub on the left which reproduces frequencies under 40 herz...

Te tower in the middle is capable of under normal circumstances easily putting out 2x350Watts of RMS continuous power (!) but is limited by a factor of two or more at the moment by the preamplifier, and is in this setting NEVER used even near this power rating, only on a fair it has been tried out up to those sound levels (last year when it had just been built). The 3 stereo amplifiers present in the lower two enclosures of the tower are of excellent quality in the sense of neutralness, uncoloring of the sound, and are driven by the pre-amplifier and 3 band filtering present in the upper unit, which also contains my self built sound synthesizer with DSP, but mainly acts as the preamp and seperation filter with the best quality  audio opamps in the world (7 of them, some complimentory (free) samples some ordered from Newark), if we may believe Burr Brown/Texas Instruments.

The beamer is behind and above the viewer(s) close to the ceiling in upside down position on a stable enough improvised tower, and also suspended from three small ropes from the ceiling in case something goes very wrong.


The source for the beamer can be: an older (but good) video cam with s-video cable, the media box or the output of a normal consumer VCR to watch live TV, a HDMI to DVI cable with 5 meters DVI extension to connect to a media computer or the shown Dell M90 notebook (from last year) which has a Nvidia Quadro graphics card built in and also has Purevideo hardware High Definition Video decoding capability and a screen which can show perfect actual 1080 HD (more than the beamer can).

In the picture above the rear surround speakers are visible wich are the small 2 way monitors I've built years ago in this case, but I've had the medium sized bi-amped monitors there as well, at ear height, though this setup performs well, also at normal volumes, because the small but quality amplifier shown below has a big supply transformer.

Surround has in this case to come from a DVD being played on the media computer or 'live from space' from BCC digital videoand surround decoding, and is for the rear speakers DA converted by microsoft high definition  (what a term) DA converters, which isn't studio quality, but the better ones compared to many soundcards. The main channel is fed through a TOS (digital) audio link to a Lexicon studio grade DA converter of 24 bits, and is with the main audio system total studio quality, even of high standard.

Above the screen and on the sides black cloth has been used to make the radiation of bright scenes (which at 1500 lumens with a 1.5 reflective screen light up the whole room) not reflect back on the screen via the white ceiling or walls and imitation light wood main speakers.

The main (self designed and built) amplifiers and active filters with a poster used at last years audio/music fair behind it. This year the fair has been cancelled because there weren't sufficiently many visitors.

The improvised setup with the Lexicon USB AD/DA converter and all kinds of cables, for instance to connect a synthesizer up as well. The converter is connected to two computers, and can be driven from the server on the other side of the screen (which serves a bit as a sound shield as well) running Fedora 7 and for instance Cinelerra with the X server running on the notebook to create the expanded music I desribed froma large disk connected to the server, which over local 100baseT ethernet can also be accesed as network share to show HD video interactively.


The little but excellent rear speakers amplifier (because the medium sized biamped monitors are in use elsewhere).