3 channel 36 bit 1280 X 720 low $ camera - Viper?

[Wayne Morellini]

I made an error with the potential camera designs, I forgot to mention sound specs. There are three options integfrated, prosumer external USB/Firewire boxes, and internal sound card. Now because most of our designs will not have a spare PCI slot, we will have to settle for an external box and or integrated audio.

I am not a sound professional so take some of the following with a pinch of salt until you have a pro opinion (as I am winging most of it from memory):

Issues: audio chipsets that have seperate DAC (digital to Anlogue converters for sound output) and ADC (anlogue to Digital convertors for sound input) work best. As if the DAC/ADC is integrated with the main processing chip significant thermal niose will build up and corrupt the signal (especially from the ADC amplifier). Designs that integrate these chips should be avioded. From memory amplifiers should allways be on seperate chips to the DAC/ADC for the same reason.

DAC and ADC (and some seperate sub-components and amplifiers) vary in performance from one another, and any can corrupt the sound chain. Some people replace individual components to improve performance, and you can find discussions in forums on this, but for the rest of us good reviews will have to suxfice. One such mod was to put bluetack across the crystal to stop vibration induced niose being transmitted through it (I do have the knowledge, and have not researched this). Even though some manufactures of cards, or chips, are better than others, it varies from card to card, asnd chip to chip. I don't like Sigtel (I think that was the name) chips some people prefer the high end Cyrstal Semiconductor, AKM DACS and a couple of others (though cheap versions can have less quality). When reading the reviews I noticed that they performance and part price follow each other. The price varies from dollers to hundreds of dollers for dacs, and I guess the smae is true for ADC's.

Third, recording performance is often bad compared to sound reproduction. Looking at the reviews it will be noticed that recorded sound is often a lot worse than what the DACS will produce with significant channel cross talk.

Fourth, advertised performance figures are often far higher than the tested performance, particularly recording. You will also see cards advertised with a frequency (96Khz) only to see the signal drop completely off long before 96Khz, or at 44.1Khz (you might be able to send them a 96Khz track but they can't playit properly).

Fith sound mixing. Certain cards have mixing and processing units that work only at certain frequencies and bit depths. So this will result in all frequencies above that to be down converted and played at the lower frequencies. Other things that have happened, is that only the digital out can use the higher frequency, or only if it is played straight (bypassing the processing/mixing) without any effects. The result of down mixing problems (the full and half versions of the VIA Envy have 36bit hardware mixing) produces quantanisation errors. A lot of consumer market sound advertising is tainted by marketing hype, so read good reveiws from good sources.

Sixth, Signal to Niose ratio is desceptive. Manufacturers can use a gate to artifically boost the SN figure. But how this works is that the gate is applied when there is no sound signal coming through and the SN is measured with no signal, that can make a 50db SN card look like 120db SN card. I think the cards make abrupt volume increases/pops whe the sound signal comes out of quiet periods.

Professional cards cost a lot more (until now) and can handle multitrack recording (for surround purposes), but significantly, you should be able to get better quality recording (get multiple reviews first).

Now when you go and look at sound card reviews pay special attention to comparisons between Creative Audigy Cards and cards based on the VIA Envy chipsets from professional sound companies. A number of these cards are coming out in external versions, the problem is to find a good performance one that has mulit=track input. With a USB2 or firewire version you then by pass the need for an PCI slot.

While many motherboards have sound, only a few of them are very descent due to the low quality integrated dacs/adc (and lack of mixing hardware they use). So check out good reveiws done with Rightmark Audio Analyser (or equivalent) if you are going the motherboard route. VIA bought out a leading audio chip manufacturer a few years ago, and the VIA Vinyl six/eight trac audio DAC and Envy chipset are some of the better ones out (I have not been tracking it for a while so there maybe some better, and I have not checked out recording performance on inegrated). The six track maybe combined with another one for 8 channels, so check that to, as it maybe poor. But motherboard audio is at best descent stereo recording. If you want good mul;titrack recording you have to go for a card.

I have recently come across a new card claiming to, and actually delivering really close to, 120db performance at a cheap price (previously 110db was the ebst from a much more expensive card). It has it's own limitations, that a professional opinion should be sought on, and I guess maybe the first of a number of cards from different manufacturers using this new AKM convertor.

http://www.extremetech.com/print_art...=123687,00.asp

Just in Digit-life review:
http://www.digit-life.com/articles2/...820/index.html

The dynamic range of 16bit CD is around 96db, 20bit is around 120db, and 24bit is around 144db. CD is 44Khz, 48Khz for DVD, upto 96Khz for multi channel DVD adio and upto 192Khz for stereo DVD audio. CD is supposed to max out the human hearing range, but tests indicate (and I have heard it) higher frequencies produce a better tighter sound. I would suggest that, ideally, a card capable of a 96db+ range/SN, and 48Khz+ singal would be ideal for good sound. Below this and 72db+ (12-bit+) 48khz for descent sound. A limited number of cards come up to the 96-106db range, and few cards above that, in playback, but at the same time most of them fail/struggle to get to 96db recording.

I have gone and sought 3rd (max CD quality playback) level sound hardware (compared to the 4th, 5th, or 6th level sound from Motherboardfs) and I can tell you there is a big satisfying difference between descent and good sound. Good sound makes a better movie, bad sound makes a bad movie (ever see a DV camera with surround 16-bit sound, no, they don't want you to have it). In a cinema, or on a good home theatre, descent sound will take the edge off your sound and make it sound harsh.

Most of this is from the perspective of hometheatre than from sound recording but I have picked up a bit.

Links:

www.3dsoundsurge.com
News, and a nice forum with much consumer/prosumer discuussion. You can see links to professional forums being mentioned there.

www.digit-life.com
The people that did the rightmark audio analyser, used to do nice reviews I could whole heartedly agree with then suddenly the reviewer dissapeared and eventually a new reveiwer has come in.

www.rightmark.org
Opensource Rightmark Audio Analyser for testing audio.

http://www.via.com.tw/en/digitalbril.../via_audio.jsp
Some of the Via product.

http://www.extremetech.com/print_art...=130106,00.asp

New High definition audio from Intel is descent looking (haven't read the reveiw but looked at the figures for playback (still not any indication of recording goto digit-life for that), figures don't tell the full story as each brand of Dac has its own style of sound, so you best to check the ADC/DAC sound, some will be warmer, some strongrer/clearer in highs, mids, or bass ).

From 3D soundsurge and digit-life you can get most of the links you need. Sound card companies may also list links to revieiws from the product page.

Wow, look what I just dug up some good links on sound issues at extremetech :

http://www.extremetech.com/resource_...,s=1074,00.asp

[Wayne Morellini]

I've come accross a number of things in the last week.

In new scientist, 19th June 2004, "Give it Some Gas" p26, they are talking about microengines that burn fuel (propane, m/ethanol) to produce electricity. They sound simular to a couple of designs I wanted to develope myself, so I think they are worth watching next year. Fuel has around 40 times more density than Li-ion battery, the alternative would be some future capacitor that stored more than the battery but may blow your head off if it blew up in your ENG camera, or a fuel cell that has it's own problems.

Commercialisation of micro fuel cells are due over the next couple of years, engines may take that long to come out and catch up. So shortly they might be viable for these cameras. In a interesting side note. In the past, in New Scientist, I saw a car by some UK college that used conventional parts to run very competively on compressed air. If anybody out there is experienced in this area, could we do the same as a power source for this camera?

A couple of the engine designs are simular to some I was hoping to work on myself. Some time ago a freind of mine was trying to put together a Bourke engine (the most efficient engine ever designed, would even run off the lubricant oil being fed the cyclinder). I was musing with a way to eliminate the scotch yoke and turn the piston through it's stroke to produce an electric current, and I pretty sure he then just suggested that the straight up and down motion (less scotch yoke) could be used to produce electricity (details left out), so I named it the Watson engine. Another, my own, sounds a little simular to the turbine version they are talking about there. Which ever way, they definetly are onto something.

http://www.brighthand.com/article/To...Tiny_Fuel_Cell

[Laurence Maher]

What kind of fuel?

[Wayne Morellini]

<<<-- Originally posted by Wayne Morellini :
that burn fuel (propane, m/ethanol) to produce electricity. -->>>

Laurence well probably see you around, I'm taking a bit of a break because of health problems at the moment.

Thanks

Wayne.

[Adrian White]

Haven't posted for a while as I have been following the contributions of the more technically knowledgeable people on this forum. I have come up with a possible low cost hd workflow solution. (feel free to respond and pick holes as any feedback is very welcome.)

Hardware:
Camera link camera (silicon imaging or imperx) 1920*1080
frame grabber
Streampix recorder software.
Suitable spec pc with either raid or number of external hard drives.
Use 16mm c mount lenses bolex or sneider (or get c to f mount adaptor and use nikon 35mm?)

1. Use streampix software to record 8bit images from camera and save to hard disk/raid. Save in a file format that is compatible with both streampix and Vegas.
2. Import video files into Vegas Video 5. Here are a list of file formats acceptable, when i last checked, streampix uses at least 3-4 of these:

AC-3 Dolby Digital AC-3**
AIF Macintosh® AIFF
AVI Microsoft® Video for Windows®
BMP Windows® Bitmap
GIF CompuServe Graphics Interchange Format (stills and animated)
JPG Joint Picture Experts Group (JPEG)
MOV Apple® QuickTime® Movie
MP3 MPEG-1 Layer 3 (Audio)
MPG MPEG-1 or MPEG-2 Video
OGG Ogg Vorbis
PCA Perfect Clarity Audio™
PNG Portable Network Graphics
PSD Adobe® Photoshop®
RM RealNetworks® RealAudio® 9
RM RealNetworks RealVideo® 9
SWF Macromedia Flash
TGA Targa™ File Format
TIF Tagged Image File Format
W64 Sony Pictures Digital Wave 64™
WAV Microsoft Wave®
WMA Microsoft Windows Media® 9 (Audio)
WMV Microsoft Windows Media 9 (Video)
Still Image Sequences (Script)

Apparently Vegas 5 is resolution independant and can manage frame sizes up to 2048*2048! 1920*1080 is supported with 23.976,24,25,30 fps options.

I dowloaded the trail version toady and will be web hunting for small clip samples to try.

Further info can be found at "sony pictures digital"
Am I missing something here or is this workable, any responce welcome.

P.S. If this is workable, I also found out about a linescan dalsa camerlink camera currently in beta version that has a frame size of 2048*2048 at 30fps but can be reprogrammed to do 24p, would this be practical?

[Laurence Maher]

I think imperx has global shutter, doesn't it? Might not be too good. Not sure about SI.

[Laurence Maher]

Hey wayne, ya, . . . duh . . . you already posted the fuel type. Not eating my wheaties. Hope your health is good.

[Jason Rodriguez]

Both Ikegami, JVC, and Kinetta are going to be using the Altasens 3560 in their cameras. How come they don't have any problems with rolling shutter artifacting?

In fact I saw the 3560 and some other cameras built on the 3530 at NAB, and again, there were no problems with rolling shutters. I know there must be solution here because these are manufacturers using the same chips, and aren't having any dificulties with slanted lines, etc., because these types of artifacts wouldn't be acceptable to their production-bases markets.

So again, if this the way the chip is built, and these are the manufacturers successfully using it, then there must be solution somewhere (and it's not in loosing the high frame-rates of the chip, because Kinetta is going to use the 3560, and they are going up to 60fps without dropping frames).

[Anhar Miah]

Thats last Century (literally),

Thats old tech, if you really want some thing more advanced you should look up about trapping into the engery from the vacum of free space, just google MEG (or Motionless Electromagnetic Generator)

Its FREE unlimted amounts of Energy to power anything really, (still mostly theory though, which *some* working prototype





=--===----=--=--=-=-=====-==-
----=--=-=-=---=-=-=-==-=-====
-=-=-==-----==-=-=-=-=-=-=-=-=

Anhar Miah

[Obin Olson]

easy, run your caemra at 50-60mhz

[Laurence Maher]

Hmmmm

Interesting Jason. Maybe we can call and ask about the chip as a paranoid "possible customer" or something, and mentioned you're worried about that possibility, and have them explain how they got around it?

[Matthew Miller]

Altasens claims on their website that they use a patented technology called "tapered reset", but I'm not sure that has anything to do with eliminating rolling-shutter artifacts. It says it allows for "lower noise and lower image lag than competing alternatives"
The chip can also be timed with a mechanical shutter or electronic gating. And there is mention of something called line-mixing, but that might be for sub-sampling to lower resolutions.

Does anyone know if you can buy these chips individually and how much they run for? Specifically, I want to know about the 2560 since I want to work with a 1280x720 image.

[Steve Nordhauser]

The tapered reset is noise reduction only. There is noise associated with the reset pulse to every line. By removing the high frequenicies (slower edges) you get less noise. It is an unbelievably quiet chip.

The mixing lets you get size reductions that are not 2:1, 4:1, 8:1 - the standard subsamples. They did this specifically to get to 720p - which is not a perfect multiple of 1080i or 1080p. The 1080p Altasens can run at 720p.

I don't know about them but most chip makers of complex parts don't like to sell single chips. They figure the cost to them to get a single camera running (tech support) and the cost to get an OEM going with one chip is about the same.

[Juan M. M. Fiebelkorn]

Looking for FPGA in Google I found this:

"We've implemented the new JPEG2000 Standard
on a Virtex 1000 and required about 2/3 of its resources"

Hope it is of any use....

here is a link to a Jpeg compressor on a FPGA.It has huffman and the rest.Don't know if it could be modified to use some parts of it or.....

http://www.ece.cmu.edu/~ee545/f98/swingers/index.html

Here more really interesting links..

HUFFMAN ENCODER

http://vlsi1.engr.utk.edu/~mswiatko/ee552/proj/pres.htm

WAVELET TRANSFORM BASED ADAPTIVE IMAGE COMPRESSION ON FPGA.
(VHDL source code)

http://www.ittc.ukans.edu/projects/A...rin_thesis.pdf (PDF)

REALTIME IMAGE ROTATION AND RESIZING, IMPLEMENTATIONS.

http://www.xilinx.com/products/logic...ion_resize.pdf (PDF)

ESTIMATING FPGA REQUIREMENTS FOR DSP APPLICATIONS.

http://www.hunteng.co.uk/info/fpga-size.htm

REALTIME IMAGE PROCESSING WITH FPGA (source code and diagrams)

http://www.ece.ncsu.edu/imaging/Arch...eoCard/Report/


LATEST NEWS!!!!!:
I've just did a comparison test between HuffyuvRGB and Morgan J2K lossless compression 4:4:4.
The compresion ratios for a relative clean source are:
Huffyuv(RGB) : 2.3:1
J2K LOSSLESS (4:4:4) : 4:1

J2k took twice the time to compress.To me this looks amazing.
Are these ratios possible?

[Obin Olson]

how is the Altsense camera doing Steve? any news?