4:4:4 10bit single CMOS HD project

[Eliot Mack]

It would be nice to split the thread into 3 areas:

-Current development work of Obin, Rob(s), Steve N, etc. Discussion of Micron and Altasens based cameras, software development, storage, processing, and editing.

-FPGA development: FGPA boards and public domain software algorithms to run on them. This looks like the next obvious step in the camera's evolution.

-The Long View: Things that require significant optical engineering ability or are a ways away: 3 chip versions, 6 MP cameras, parallel processing arrays, etc.

This way, there is a place for everything, and any post that lands in the wrong thread can be readily relocated. I originally thought I could keep the Wiki up to date but the info volume is quite large. Moderators, could this be done?

Thanks,

Eliot

[Obin Olson]

ok thread started for FPGA talk!

a quicktime idea:

http://cameralink.greatnow.com/fpga.mov

[Obin Olson]

I just made a new thread and uploaded a quicktime file for you guys to take a look at :=}

http://www.dvinfo.net/conf/showthrea...threadid=28773

[Obin Olson]

Steve just what framerate can you get at 3.2mega pixels with a 32bit grabber and also with a 64bit grabber? could you get 24fps from it?? at full resolution?

[Juan M. M. Fiebelkorn]

Really nice.Exactly what I'm talking about in concept...

[Obin Olson]

that is the $50,000 quote I got - for that concept

then I started thinking why use FPGA instead of MicroATX or ITX PC stuff? and it seems that micro PC stuff would only need a good programmer and not all the hardware design that the $50,000 would go for...hmmm so I am now going down the road of LINUX micro PC with a software IU. My Canon 10D has a DOS UI why not somthing like that for a video camera? TIVO is a standard PC with Linux on it - we can do the same for a video camera

Even Xbox has windows on it

I just need to find a good code writer at a good price :=}

[Eliot Mack]

Hello,

I just created a thread called 'Blue Sky Concepts for Digital Cinema Camera'. Its purpose is to house many of the useful potential future areas (3 CCD imaging blocks, 6 MP chips, parallel computation arrays) being researched by people on the forum that do not relate specifically to the current CMOS camera/software development underway. Obin has created a thread for FPGA work; combined with this and the original thread it should organize the posts well.

Thanks,

Eliot

[Juan M. M. Fiebelkorn]

EZW is supposed to be a part of JPEG2000 standard.

Some implementation:

http://pesona.mmu.edu.my/~msng/EZW.html

Original Matlab:

http://perso.wanadoo.fr/polyvalens/clemens/ezw/ezw.html

Wavelet image constructor Kit

http://www.geoffdavis.net/dartmouth/...t/wavelet.html


Jpeg2000 supports 16 bit and down, LOSSLESS and LOSSY, any combination possible in the same stream.Metadata is included.So you could store info about color balance etc, focus and anything else.

FPGA, microcomputer....

SUSAKU PROJECT

http://www.atmark-techno.com/en/product/suzaku.html

http://suzaku.sourceforge.net/index....kuHardwareInfo




OpenCores already have cores for:

8x8 fully pipelined parallel DCT. Provides a DCT result every clock cycle.
QNR. Quantization & Rounding Unit.
Run-Length-Encoder.
Huffman Encoder / Decoder.

http://www.opencores.org/


Site about Mini-ITX and Nano-ITX motherboards and systems

http://www.mini-itx.com


A Xilinx Spartan3 FPGA 1 million gates costs $15


And here is a developmente board

http://www.nuhorizons.com/products/x...ent-board.html

Another place for cheap development boards.

http://www.fpga4fun.com/board.html



FREE IPs

SDRAM Controller:

http://www.cmosexod.com/sdram.html



Open Core compression system

http://www.opencores.org/projects.cg...stems/overview


Diagrams of the Elphel camera:

http://www.elphel.com/3fhlo/index.html


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/ logi...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/


Implementation of DWT (VHDL source code)

http://kondor.etf.bg.ac.yu/~dejaniv/...t/DWT_VHDL.htm

Jpeg2000 acceleration using GPU

http://www.cse.cuhk.edu.hk/~ttwong/d...pu/dwtgpu.html
(open source, you need at least a GeForce FX)

HERE IS A LONG LIST OF FPGA BOARD WITH OR WITHOUT PCI INTERFACE


http://www.fpga-faq.com/FPGA_Boards.shtml

[Les Dit]

The absolute coolest FPGA dev system I have seen is Altium's Nexar.
They allow you to use pre synthisized blocks that they have pre defined functions. Some of these blocks are complete CPU's with memory. You can use interactive debuggers in C with JTAG support on these FPGa processors ! Right now they only support 8 bit CPU's but are adding 16 and 32 bit processors.
I swear they have designers that were found in a crashed UFO!!!
Their stuff is really that cool. Device independent too.
You don't have to know HDL. It all integrates with Protel, a high end PCB program. Transparently.
For example, it can reassign pins on the FPGA design to make the PCB routing better.
It can simulate the entire design with the C code running in it at the same time.

I believe the dev system is $8K.

See the webcast of the system intro at
http://www.altium.com/nexar/

[Rob Lohman]

Obin, Juan, Wayne (when you join), Les and ALL others who
want to post links.

PLEASE PLEASE PLEASE make the links hyperlinked. I've edited
all the posts above to make links work and with the high volume
of posts these threads create I have lots of work to just do that.

It is *really* simple to do, just do:

[ url ] link [ /url ]

Without any of the spaces. This will create a working link.

For more information read this link, please.

Thank you very much!

[Rob Lohman]

Please do NOT split off any more threads then the two just
created. We already have 3 threads running on this subject
and now 5. I agree that things need compacting etc., but the
best way is probably either through an article or Wiki. Then
the long running threads can be closed and new ones can
be created instead of added.

I am going on vacation for two weeks this friday and this is a
very busy week for me and the other wranglers and admins on
this board. So any chances will probably not happen till august.
I will bring these points up to the higher powers.

Thank you for the ideas, suggestions and support!

[Wayne Morellini]

Too late, sorry, I just did while I was replying, see below. If you don't like them delete all the new threads, but I think the ones I have started divide it up nice and tightly. Once people get cameras left right and center, all development discussion is going to be swamped by camera setup issues, so we will need them anyway. By the way welcome back, I haven't heard back from you yet on the email, Rob is thinking about it.

Thanks

Wayne.

Originally posted by Jason Rodriguez : BTW,

FYI, if you're using a 2/3" chip, you're going to get around the same DOF as a 16mm frame.

Now unlike video lenses which have to deal with prism optics, 16mm film lenses open up way more, like down to f1.3, and at that f-stop, you're getting around the same DOF as a f2.8-f4 split in 35mm, which is not a lot of DOF at all-that can give you very shallow DOF like you're wanting without the need for a 35mm ground-glass optical-type effect.

For instance, there's a reason that people pay $115K for a set of Zeiss digiprimes. They can basically open up to F1.6 (the maximum f-stop for B4-mount (2/3 inch Sony mount) lenses), and be sharp as a tack whereas many other lenses fall apart at that F-stop on video cameras (again, there's a prism block limiting the maximum aperture of a video camera). With a single-sensor design there is no such problem, hence you can use optics that are very fast, and as a result get that nice out-of-focus DOF you've been wanting, without the need for a ground-glass 35mm converter.

You know, I learn a lot of things here. this would explain the low f-stop for 3chip consumer models.

Originally posted by Steve Nordhauser : The 3.2Mpix Micron is *not* a replacement for the Altasens. It is just a different product that might fit a certain set of needs. Here is where the Micron data lives:
http://www.micron.com/products/imaging/products/datasheets.html

Wayne on Gigabit:
Most of our cameras will end up on gigabit over time. This is the raw sensor data being transmitted. When using an Intel Pro1000 interface card (<$50) and our custom drivers, expect 800+Mbps (that is bits) with any pixel over 8 bits taking 16 bits. We just tested 1920x1080x24fps@10 bits (796Mbps) and got continuous transfer and display with the SI-3300.

As you are breaking from traditional cameralink capture boards, why don't you goto the next logical step and pack the pixels, then 30fps would become available? From the way your comments read it seems that you are implying that only a add in card can do 800Mps, which means not an integrated port :(

Originally posted by Juan M. M. Fiebelkorn : Guys, what we are planning with FPGA is the same and more tha what we can do with DSP :)
I don't quite understand why everybody thinks that FPGA are expensive, that their development and programming is expensive and that they are power hungry......it seems so many lies have been told to poor consumers :).

I VOTE for a new Thread, only technical!!
And with access only to registered users (no need to tell you why ;) )

@Wayne, sorry guy but sometimes you dream to much, like a brainstorming of one guy!!
Don't expect the Via machines to be a DSP laboratory.
What is at hand now is more than enough.I say this cause the easiest way now would be to use one of that C3 machines with a simple $100 FPGA developmente board (PCI interface), program some simple routines on the FPGA (to compress?) and then recording to disk.
This way we use normal C code, combined with Handel-C or SystemC to offload from CPU intensive tasks.We have the IDE interfaces of the motherboard, we use Linux, we program a graphical interface to control the Camera System, we use a LCD touchscreen of 7" diagonal (directly conected to Motherboard's VGA) and we also add some De-Bayer simple algo to show the image realtime on the touchscreen.
This will drain only around 15-20 Watts.

I doubt it, you see the difference between cheap DSP and cheap FPGA could be significant, but we only need limitted compression power for lossless (though I don't know about high compression ratios) somebody quoted figures before so some of this stuff I quote comes close to, or exceeds those figures. Sure FPGA might have lower power requirments today, but low powered compared to the hardwired alternative to a low end FPGA, I doubt it. We would be needing multiple parrallel low end FPGA's to get the performance, and that can get complex. I post FPGA stuff here to humour people, not because I think it is the most practical, lower powered, affordable solution. The things would be more time consuming than using the existing routines, from the C code, it would have to emulate. What people are failing to see is that cheap mass produced (low cost) DSP in motherboard is worth more than FPGA on PCI card. One, it is probably gauranteed to be lower powered, if not with more performance. I think the sample PCI FPGA is a good idea, if we can get a high powered model cheap enough. Another problem is that we have only one slot for Cameralink, and that will take up case room, I could fit two cards into the case, if a suitable riser is available, but this all requires space and power compared to GigE. Does your power quote include Hard Drive power cosumption. With out GigE you might aswell forget a handheld camcorder version of the cases. Brainstorming, you guys have to look at the picture from start to end and all the factors (and how to get the cheapest widely accepted product) if I can do that great, don't knock it. The FPGA/compression chip discussion is a bit premature, people are excited and bored, until somebody starts doing it, I might be enthusiatic about low pricing myself, but those examples are based on best practice mass market implementations. Which means, unless somebody like, VIA in MB, or Sumix who announced comrpession first, or SI, puts it in hardware for us will probably land up costing more than cameralink capture. I have been around this hardware development stuff before, and it aren't cheap compared to programming, not even clearspeed apparently (maybe next year) which I was aiming to approach a manufacturer to implement for us. Most of my ideas are better performers that somebody else might implement, instead of us, cheap. Here's a challenge, go and find the cost of a FPGA card/chip (chip needs another 100-200% to be incorporated) that matches the Gflop performance figures posted a few weeks ago. If you can find it cheap enough good, I'll be happy, if it only adds $10-$30 to the finale price, like somethings I am pursuing, even better. You could also go back in the three threads and see who first suggested compression in the camera head, and a number of ideas Sumix is using.


The idea for a new thread, I have allready given the homemade camera thread over to discussion. In this thread, multiple things are happening and sometimes the discussion will go one way and sometimes another, and after Obin sorts out his camera, and people sort out their compression parts, there will be little traffic. I allready post much of the camera spec and parts stuff to the Viper thread rather than here. Maybe we need threads for "Home Made HD Cinema Cameras - General Discussion" (or use my thread), "Home Made HD Cinema Cameras - Problems and Performance" (for discussion on people's individual camera problems and performance), and "Home Made HD Cinema Cameras - Technical Discussion" thread (for technical discussion and suggestions not covered in the other two). I vote against closed threads, that is why we have multiple recipients in emails. Splitting technical up into threads on FPGA, programming etc at the moment will lead to minute volumes, and having this is one thread leads to lively idea exchanges, as each can make contributions to other parts of the project. You don't see Rob's doing much programming discussion, they do it by email/phone.

I just created them, and put a link to the wiki:

Home Made HD Cinema Cameras - General Discussion
http://www.dvinfo.net/conf/showthrea...threadid=28779

Home Made HD Cinema Cameras - Problems and Performance
http://www.dvinfo.net/conf/showthrea...threadid=28780

Home Made HD Cinema Cameras - Technical Discussion
http://www.dvinfo.net/conf/showthrea...threadid=28781

Have fun!

[Juan M. M. Fiebelkorn]

@ Wayne:
If I don't need glasses here it says:

http://www.via.com.tw/en/Products/eden_n.jsp

7 watts for 1GHZ. And if I can read well enough on the Maxtor Diamond 200 GB I have now on my left hand it says 1.63 watts.
So I can't see why it would consume far more than 20 watts....

I'm not fighting for the first prize.This is not a competition.
I'm just giving technical data about what is available with little money.I'm not saying that what you say is wrong, so I can't understand why anything I post is commented by you as "I doubt it", etc,etc.

You insist with the Clearspeed solution, OK, but don't say FPGA solution would be limited because of its cost, when a single Clearspeed chip costs 1,000 and a Jpeg2000 off-the-shelf chip costs 40.

Some more info.Generally Lossless compression requires more computation power than Lossy.So if we want or need Lossless we need more power.If you doubt about it,see an example:
When you compress or your machine, e.g. with a DV codec and with Huffyuv or Sheervideo, which one runs faster?

About the pricing of FPGA development boards and chips, it has been posted a lot of times.Boards go from 100 bucks to 3,500 (may be more I guess) and chips from 12 to 500 bucks.
If you need the URLs, I will post them at your request.

About the mini or nano-itx, I realized that it would be the cheapest and easiest solution at this moment.But seeing that an Eden processor is not a Pentium4, I guess that we would need a little more horsepower to be able to record anything on Disks, RTL.
These motherboards come with GEthernet, so no need for a framegrabber.The problem is that, if we don't use high compression, We won't be able to record on anything but a RAID 0, with its added problems.
I know also, I could use a mini-itx witha P4, but then I won't be able to run it on batteries, at least not for reasonable period of time because a P4 consumes around 100 watts.Not to mention it needs a Cooler, with a fan, which also needs electricity and produces noise.
Just my two cents.

[Wayne Morellini]

I don't really intend to, until somebody is near ready to start work on the FPGA design, which if it is Rob, will be after he finishes programming, which might be before Sumix brings their compressed camera out. And apart from not being very familiar with FPGA, you don't really need me anyway. Just cross post all my fpga links here, and those two UK magazines, I mentioned doing the FPGA tutorial series.

Just one piont, is that the thread name excludes alternative solutions. If you want to discuss alternatives, or post fpga stuff, feel free to post it to:

Home Made HD Cinema Cameras - Technical Discussion
http://www.dvinfo.net/conf/showthrea...782#post198782

By the way, I have been url hyperlinking for a long while, I notice people seem to get other people confused with me. But how do you hide the address and give it a display string like in Html anyway.

Have fun, I wish you the best.


Thanks

Wayne.

[Wayne Morellini]

I suggest you re-read your post and at least remove the "at" @ symbol. Now I post good realistic stuff to get you guys thinking, that doesn't allways work. The only "dreaming" I'm doing is dreaming of low cost realities.

The amount of watts for that hard drive is suprising, I'll give you that. The cpu, main board, and one drive would put you in that range. I am not certain about the extra power for the capture card, camera and fpga, but maybe it is still in range. Somebody recently suggested 200Watts power consumption, I thought that strange, but haven't got around tor researching it.

I am not "insisting" on the clearspeed. As has been posted before, Steve got a quote, and they are being to over zellouse in pricing at the moment, so I dropped it, maybe after they get a mass market deals it will drop in price. maybe next year. they also plan to do a PCI board.

I just thought of another problem with using PCI for FPGA and Cameralink, they both will require a lot of data transfer, and should satuate the bus on anything more than 720 Bayer. You still also have the problem of finding a part that is fast enough, more expensive, or using multiple chips, that pushes up costs compared to a comrpession chip, as you also have to factor in the extra costs in the board design, component handeling. All design/manufacturing issues. I have pushed FPGA before, but the solution has economical restrictions. Again, have you located a cheap FPGA with enough power, preferebly on a PCI 64-bit/66Mhz/PCI E card? When you say $40 of the shelf, is that single sample or the 100-10K quantities we should be talking about.

Lossless compression requiring more power, I've seen some excellent figures compared to mpeg4 etc (probably because of the extra processing to find image compromises), which lossless and which lossy compression routines are you comparing?

I suggest you read the three threads, and see who actually first suggested the itx and eden, camera head compression and most of the other things you just posted.

Originally posted by Juan M. M. Fiebelkorn :
I'm not fighting for the first prize.This is not a competition.
I'm just giving technical data about what is available with little money.I'm not saying that what you say is wrong, so I can't understand why anything I post is commented by you as "I doubt it", etc,etc.

Good, that sounds like what I have been doing all along. I'll restate my aim. Research and promote a new low cost alternative standard in video and cinema for the free benefit of the comminty. Working towards maximising it's user freindlyness, and low cost mass market components, to maximise it's sales and success in the community (by upto 100 fold), which would also help lower the cost. Seeking out the best, longterm, lowcost solutions. If it was priced low enough, reprogrammable clearspeed, or Processor In Memory, would be best in the camera head, on the motherboard that would be integrated co-proccessors (DSP's and graphic cards, and Processor In Module PC memory Sticks). And as part of this to fit it into acceptably sized casing, that users will accept. I think about these other peoples' requirements not just my own, if other people don't, then why are they bothering.

Now if you want to discuss this fpga further there is the fpga thread.