Theo Verelst Local Diary Page 37


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

  Mon Dec 25  2:03,  2006

Strange christmas night in a strange town, with a strange setup (a notebook and a 21 inch screen connected with it and powerfull applications running on more than 1 computer), but kind of fun, with some TV (after the owner of the new ´media box´ from the cable company was not using the remote) in a strange way: decoded from the digital cable channels (almost all for free for a few months to try them out) the analog output goes to a pinnacle real time hardware encoder box, which squeezed 5Mb/s over a Usb-2 link to a (fedora core 4/64) fairly fast linux server, which runs a script to make that stream available over a local network (with consumer grade router containing the adsl modem which serves this page to the world to read) where a notebook runs a complementary script which feeds ´mplayer´ which runs on the secondary screen in good looking full screen mode.

There are worse things than ´strange´...

I've been thinking about Preparatory Film making work, and came up with

* course
I at a good point in time took a serious couse recording video. Not how to switch a handycam on, or a cozy talk with ramblings about the subject, but a decent semi pro course to make a script, a stortbook, and implement the setting up of the lighting, the camera setup and shooting, and the editing on a semi-pro edit table and of course beaming and making the sound. I after the course regularly used one of the big cameras of the Delft University Cultural centre  to make good recordings. The end of the course was a project to make a short movie, for which I also composed and played a soundtrack, linked over smpte with a midi sequencer on a computer. I guess it was a very good idea to do the (fairly long) course, I never regretted having learned doing decent white-balances the accepted course of events with a storybook and good attention for the camera motions to be decent and functional. If your videos make people seasick, you don't know wether it is possible to change those blue outside lights when filming and you are sure you could make a Spielberg equivalent from the top of your head, then either don't bother comparing, or do a good course!

* lighting
Over the years I've played with various lighting conditions, from the extensive stage lights (which I fixed) at highschool to more reent trying out of various halogen lights for low budget video shooting. Of course a good set of spotlights and some diffusors would be good.

* players
Probably one of the more interesting recordings I did in this respect was from a stage play in Delft, though of course in general it is hard to find good actors and to motivate and 'get' them to work good on film.

* coding
In this time almost everything turns digtial at some point, or is digital from the start, so the format or standard, and especially also the compression format is important and a source of many troubles and unclarities and 'bugs'.  Some people seem to like deficiencies to play with, I like to think working HD on the internet.

* test recordings
I've recently started to record in High Definition format, which means a whole lot more pixels then you can see on a normal TV or even a computer screen 1920 x 1080 ! That's great because working with so accurate video resembles film to an extend, and looks real good when done right, and is a interesting challenge. Of course, like with 'real' film, more accuracy requires more and more accurate work. Just 'swinging the cam around' is NEVER good enough for anything but home video, but in HD even with motion estimation as postprocess hand shot footage is most like a total waste except fot the fun of it. Unusable for the most part. So luckily I now use a very cheap but good tripod and film like qualites are starting to come along.

Some test footage, which I might use for a film later on I shot at Artis (a Zoo) in Amsterdam, like this shot from a aquarium:



Clearly you need a big screen to view this screendump from windows media player, which shows the size of the images which come by at 25 per second...




The water was blue and there are all kinds of breaking and wave effects which show up a lot better if you could see the image move.
The fun is of course that this is like making 25 2 megapixel photos per second.


Above a test of an edited piece where the cam moves and the scene is clearly vague, but there is a title imposed by Cinelerra, a Linux editing program.




Image quality is a bit poor in this picture, but there is now more clearity in the background, while the fish is moving.



During playback with sufficient framerate, the above looks fairly good, if I recall correctly I did adaptive filtering to improve colours, except I don't have full motion estimation software (or Hardware ...........) and only a real good deinterlacer in the NVidia Purevideo viewer which is part of windows media player, not of the processing software I can use, so the moving fish in the still above looks interlaced.

Another test is this plant setup, here photographed with a (decent) 5 Megapixel digital camera:



The following image is an averaging of a number of HD frames but with 32 bits floating per colour (a whopping 96 or with RGBA 128 bits per pixel!), and probably some other processing:




Here I've added an amount of an equalized version (in High Dynamic Range, probably 1 bits per colour) to the above, to give another visual effect:



Nice to play with.

Is christianity out forever ?


Well, honestly what I found during my life, in sound doctrine and the supposed word of God (and God the Holy Spirit) reminds me of my highschool lessons greek and latin and about the general literature doctrine of the classical hero: be dies doing good. I clearly recall I felt anguish about trying that thought and half raped in my sense of righteousness, because I thought that idea was wrong: a saving her, who does good and no wrong, and who indeed saves should not be dead, but alive. btw, I only did latin and greek for at most a year, I dropped those courses because my beta course list (the heaviest possible: physics, math (I and II), economics, chemistry, english, dutch) was good enough, and I found it absolutel for dorks to on average score more than a 9 (out of ten) and then not all the time, please, and certainly I saw no need to torture myself with (the idea...) and extra course for my exam. Sjeez, as if I had no hobbies or friends.

Anyhow, christians by and large ended up as almost the opposite of heros (anti-christs ?), and are even proud of it, and I see that not change during my lifetime. Of course some were both real and hero and truly sent by God and Christ and worthwhile and worthy of honour. But today ? You tell me who that would be.

My dutch exam work at highschool (this was like in the 80´s) was about humor. I´d only go back in history (like my work on hebrew translation) when there is a reason, this guy is not:






Open Source in the army: hi ha!

Quote:
"Delta3D is looking for a senior software engineer to work on the leading military open source game engine"

Yeah yeah the army too is in open source, well, good. A game engine? "Lets put all our transputers to work on this cool war-by-wire internet Quake"?! Well, you can always say ´no´! Like people in the army did recently when Kerry came to town, which I read on netscape. One, two...

Or maybe we could start an olympic math game hierarchy and that the best open source program may win: http://web.usna.navy.mil/~wdj/opensource_math.html.

I guess mandelbrot would win the marketing prize for now.


Star Trek on the higher definition BBC Sat

I had to record that on disc in space travelled-bbc encoded 576 widescreen and progressive format of course, presuming all the bits travelled without change over the digisat protocol which may have used retransmitted packages to get it so, but normally it looks ok or ot disturbes or is gone, nothing much in between, except that the audio bits are more secure, well, more certain to arrive without the decoder havng problems with it. It's about 4 megabits per second, whcih is lower than most TV signal encoders I use produce, with 256 kbps mpeg2 sound at 48kHz which is just enough to work with though I'd prefer 384, that works most of the time. But the video is pretty good for most material quite superior to normal TV  or cable service boxes in fact, and with probably progressive techniques goes in the the direction of HD TV, which I cannot receive at the moment, though the beep has a test actual HD channel, which in fact I can receivem but not record or decode because of liitations of the computer receiving the signal, even though it is a fast and advanced dual channel receiver, which can receive two digisat signals simultaneously, and even show two at once. Tante pis like the french say.

So I loaded the whole (!) '79 star trek the movie in cinerella from that format recording (about 4 gig I suppose), which takes some minutes to create the little pictues for on th timeline but then its good fun and very educational to watch the frames as they come by and bing able to perfectly freeze frame and analize picture for picture and the motions and ditherings and mini-motions between frames. Th best thing is to crank up NVidia's purevideo (which works only with windows media player, I'd love to have that on Linux, or open source software usable, to circumvent Vista altogether, possibly going xen or so, and it would even be better to have acces to the hardware HD-encoder which appears to be lurking in the better cards...



Anyhow, BBC material played full screen on a (near or more than HD) screen with purevideo is great for TV viewing. Very clearly both in the film, and in the encoding, tricks and  film methods are noticable, like progressive frmes, averaging noise and sampling artifacts, and unless I'm very mistaking intelligent dithering to create sufficiently many derived formats.


A screendump from a Fedora Core 6 / 64 bits Linux workstation (AMD Athlon 64, 3.3GHz) probably you are viewing this page from the machine with this screen:



Meanwhile, at the 3D front, I worked more than a bit with Nvidia Gelato, which is a 3D graphics language, which I also used to render HD size frames (1920 times 1080 pixels, bigger than most computer screens) from 3D models. For instance the image below I made from a 3D model (a long textfile with numbers describing geometries) I found on the internet. I rendered it with different lighting with Gelato, made the background transparent, and added a picture in the background from a real (I think a Nasa telescope) star picture :




Haha, starwars! These really are graphics, not pictures (except for the background) Here I added artificial noise (fog) tothe same model rendered from another angle:



Here is a rendering, again with gelato, with another existing star background:

(click for full HD size)

The rendering time for these is not sceonds but maybe 12-20 minutes, but it is possible to put frames like on a row and make movie with it, and also the frames can be made in HDR and processed as such quite a bit, also with Cinelerra (the movie edit program with a lot of effects which also work per frame on HDR data),. which is then high quality HD movie. Great.

Honestly, when working in at least 16 bits per pixel the format is good enough for serous cinema movies, which is a major challenge, and of course hard to make good material for. I render the results in mpeg 2 or mpeg 4 which is more limited but still a 2K frame resolution is higher than cerain 'official' movies, which leave a lot to the inter or extrapolation of the projector.

For clarity: I really made the renderings above from an existing 3D model, I mean I didn't use autocad or so to make the three dimensional model, but I took such a model and made the rendering myself, which is cool to do and rewarding when the results are so good as possible with the hardware accelerated, professional gelato.

Other great shows:



The Commodores drummer, a picture I found on the internet.

The ethernet mixer

Many people in (music) computerland are pretty confused in certain important aspects, like what is and is not important to achieve good audio quality.

This list I'd jotted down here for myself to determine the number of samples of high quality volume controls I have received from Texas Instruments:
 
pga4311u 3x + 2 on adapter board
pga2311pa 3x, 1x on arm7020 board, 1x ?
pga2310pa 2x
pga4311ua, 5x

which is enough for a bit of serious multi channel mixing. I want to use them intelligently with a lot of additional hardware, a FPGA to drive the volume chips, a linux mini-board to memoriz settings, and communicate with ethernet and the fpga and a display. This is a test setup with those parts:



    A small working Linux system connected up with displays, RTC and programmable logic

My setup, as can be read on various pages, is aimed at preferably compromiseless straight, High Fidelity meaning as little distortion as possible and low noise, well behaved with transients and of course good frequency range and transfer straightness and low intermodulation products, good channel seperation, speaker spread, sufficient power reserve, and no boom effects instead of solid tight and powerfull deep bass.

That costs effort, space, weight, ´presence´ of equipment (big speakers and amps), many parts, expensive electronic filtering, supplies, quality parts and buildup, speakers and wires, good setup (ear-height) and acoustics control (damping, placement), and of course very good sound sources after technical and objective specifications, not hype and buz or exterior design. Tropical rainwood and ivory are not going to make your speakers or amp sound better, neither are golden feet for a cd player or an extra mixer of certain design in your signal path.

Apart from clear, objective and also hard-to-measure sound quality there is of course the luxury of a sound system, especially for home use. Though it has long already been trend to make lets say spartan high end audio equipment for both objective technical reasons on one hand, and on the other to make machines look slick. Luxury is probably more considered for consumer use stereo equipment, like remote control, lots of controls to play with, and all kinds of handy shapes and attractive casings.

My main equipment is not searching for compromise as in size, amount of work and parts, part quality, complexity and weight, though I most certainly haven´t accessed large sums of money for parts and tools and research, except for my own intellectual capital. A 3 way active seperation filter amp setup weighing probably way over 10 kg composed of 2 19" rack units of 2 high with 6 high quality amp units and a excellent set of pre-amp and filter chips aren´t $199 in total, but probably worth less in parts the way I worked than what an audiofile would spend on one amp alone.

Two four way (4 speaker units) main speakers of 40x40x70 cm aren´t exactly small or average in number of speakers, and than an extra sub-woofer cabinet of 50x80x30 housing a 15 inch subwoofer speaker unit isn´t small or average either.

Driving that system with a very good DA converter from a wav source sounds excellent, in the sense of high fidelity, but also in terms of sound power, because amps and speakers allow the whole audio spectrum to be reproduced at significant sound pressure levels without distorting easily.

Many recording are made with lesser systems in mind, where various types of distortion and limitations are compensated to some degree in the processsed recordings which end up on CD or other sound carrier. Those recordings sounds still good on this system, but making 24 bit very good quality recordings without any processing and playing them back very well honours the classical HiFi thought: playback sound the same as recording, and this works for a microphone connected to the system as well: the sound is uncolored and natural and has wide dynamic range.

Those are no easy objectives and require carefull engineering of the whole audio path and using high quality (fidelity) ready-made parts, like the old rule of thumb about audio systems: the whole chain is as weak as its weakest link (actually less than that), so all parts must be good. Usually the weakest link is are the speakers, but all parts have their own weak points.

Mixing consoles are no exception, their nature isn´t very demanding for electronicists, but as soon as a mixer is added to a good HiFi system, differences can be heard in sound, just like when using different pre-amplifiers. In general any addition to the chain where the sound signal passes through can be heard to degradate sound quality in more or lesser degree and mixers must be quite expensive, or quite simple, to keep distortion to a minimum. Lets say the SSL mixer  in main studios probably can be traded for a nice house and a boat to go with it. Really simple mixers, like I made one with passive parts only, can have good quality, but more than a few channels, tone and effect controls and cheap parts can have quite a negative effect on the sound when compared with expensive HiFi sets.

It is even worse with remote control  units controlling the volume: out of the question for a high end system, absolutely disastreous.

So I want a quality mixer, remote controllable, computer controllable even (still worse), without going digital (no AD DA conversion) which is as transparent as possible and preferably not limited to a small number of channels and even surround-capable, and as independent programmable machine.

Thats quite a prototype! Yep, and I´ve not made it all yet, so it is exciting to aim so high and again have a real ego challence: will it actually wotk in some reasonable amount of time, and be up to the challenge of meeting my high audio quality listening standards ?

Well, luckily, I´ve got a working mini-linux system (self compiled kernel 2.4) with working ethernet interface and webbrowser and serial connectionsm, and some time ago (around local diary page 20) I´ve done long experiments with some Texas Instruments digitally controllable, analog signal path programmable  gain amplifier chips, which gave me satisfactory results, as corresponds with the factory supplied electronical data about all main audio properties.

And I have a metal 1 unit high 19 inch rack unit ready to be used lying around with parts like plugs and the system like the picture above. I can make that work, really, THAT doesn´t make me sweat, it will be good to have a computer (with MIDI too) controllable mixer with surround (4 channel) capability and very high quality. I hope...