Thanks Steve

I've got a question. Increasing MHZ, will that mean that integration time will be reduced despite the FPS used (leading to lots of lighting)?

The rolling shutter shutter problem. I think the problem is that it is in the wrong direction compared to the film shutter (causing the slant on horizontal movement). If we rotated a 4:3 camera 90 degrees and used the right anamaorphic convertor lense to covert to either 16:9 or 2.35:1, we would get a shutter effect simular to film and more freindly to horizontal movement.

I know the screen size in pixels is no longer right, and there is a capture speed hit because all horizontal lines (now vertical) have to be entirely read, but good as a specialist solution.

Thanks

Wayne.

I forgot to ask, for the 20-30% cpu utilisation, what processor and MB were they using?

I'm impressed by the figures you gave, they are everything I hoped for.

Thanks

Wayne.

GigE CPU utilization:
2.8GHz P4 with HT. You don't need too much memory (512MB??)

Thanks, my question on increasing SI-3300 Mhz, is it correct?

Wayne,
The SI-3300 has been tested to about 60MHz (1920x1080 @24fps). It *might* go to 75MHz (30fps) max. Keep in mind that you can get more color smear as you increase the clock rate beyond what the analog components can handle (shift register, amps, gain stage, A/D).

You could rotate the camera. The frame rates aren't quite symetrical, but you can still do a window. For example on the 3300, 1280x720 can run at 49fps but 720x1280 runs at 42fps (60MHz).

Steve, when will you have a product that is Gigabit? like the 3300?

Gigabit ethernet:
We already have two box solutions for the SI-1280F, SI-3170, SI-1300, SI-3300 (beta). The group has a 1920HD but hasn't started with it yet. A single box solutin is about 12 weeks away.

Thats good, look forward to hearing about it then.

I was asking about where ever it was right to say that increasing Mhz reduces integration time, whatever fps is used. But it doesn't matter any longer, thanks for your help anyway.

For everybody that wants to know more about sensor specs I have discovered this page:

http://www.ccd.com/ccdu.html

Silicon Imaging also has some basic tutorial pages on their site.

GPU programming.

In case anybody is still interested in learning more about GPU programming I found this article at gamasutra:

http://www.gamasutra.com/features/20...mccool_pfv.htm

The reference to the toolkit are:

Steve,

I'm still trying to figure out this rolling shutter with the higher-clocked Mhz issue.

If I'm understanding it correctly, you can run the chip at a higher Mhz, (like its max) to reduce rolling shutter artifacts, but use as long an integration time as you want for variable frame-rates and actual shutter speeds? In other words, can you run the chip at the Mhz that would be for 60fps, but only shoot at 24fps with a 1/48th second shutter (that shutter is slower than the 60fps the chip is running at, hence I'm wondering if this is possible)? Or does the chip need to be run at a Mhz that will enable a multiple of the frame-rate, and long enough to accomidate the shutter speed (so no faster than 48fps for a 1/48th of a second shutter speed).

I'm wondering if the problem becomes once you clock the chip too high, you can't get the shutter speed down (but you have reduced rolling artifacts), and you get what I call the "Saving Private Ryan" effect of a really jerky, shuttery camera image from too fast a shutter that doesn't allow enough motion blur (albeit with reduced rolling shutter artifacting).

Hi Steve,

Another question. I saw this in the brochure:What does it mean that only 3x can be used for color images? Does that mean that the 720p version will only be in monochrome-I thought I'd heard otherwise. Either that, or I had thought we wouldn't have to resort to windowing to get 720p, thus reducing the effective optical size of the sensor.

Thanks.

If you can get the equivalent of 48 fps, you should be able to skip every other frame to get the equivalent of 24 fps with 1/48 sec exposure.

If you could get 96 fps (as you could with the AltaSens @ 720p), then I wonder if you could "average" two -- or four -- frames together ... ? This would reduce the rolling shutter artifact and (I think) give the right amount of motion blur as well.
I was wondering the same thing thing -- as I recall, Steve said that AltaSens had a 720 mode that used binning and used the entire chip in a "higher sensitivity" mode; and was still a Bayer pattern, albeit a somewhat odd one. At least, that's what I think he said.

Yah, I'm a little confused myself since I know that JVC is using these chips in their new 3-chip "broadcast quality" HDV camera, but the specs on that camera are 1280x720-so you can use the Atasens 3560 at 720p, but I'm not sure what the size of those chips were (if it's windowing, then the effective optical size becomes a 1/2" rather than 2/3").

What I recall is that it does row skipping/binning to get 720p at the full (2/3") size of the chip.

Jason on Subsampling:
You are correct about the binning mode - it is only useful for monochrome sensors - I was wrong about the reduction. I'm not sure but I think you might be able to create a new Bayer interpolator for the 2/3 format (getting you to 1280x720 subsampling so at 2/3"). Row 0 is RGGRRGG. The next row is GBBGGBB. Then a row is skipped so the next is blue again, and then red.

Jason and Rob on rolling shutter:
There is usually a programmable horizontal and vertical blanking time. On the SI-1300, that can be set to a full frame time. This means that it is possible to set the clock for 60MHz, extend the vertical blanking to get the readout+blanking time to be equal to 1/24th sec. You should be able to expose during the blanking time (this is just a non-readout time). Since the readout happens very fast, you get minimal RS artifacts. Of course you also get a 60Mpix/sec bus load which will be too fast for 32 bit machines (2 bytes per pixel unpacked) but you could drop to 50MHz and set it up so a frame is moved to memory and then to RAID during the blanking time.

Steve, thanks for answering that here.

The 1920HD PDF that Steve posted earlier has the examples of the 2/3 process.

The process should be suitable for 720p conversion using the standard bayer filter.

720p bayer pattern after skipping:

GR-RG-GR
BG-GB-BG
---------
BG-GB-BG
GR-RG-GR

- Is the skipped pixels in the bayer pattern.

The combining process would result in the pattern seen in the PDF.

I am wondering,we have motherboards that can take between 4 and 8 GB of memory which is around two minutes of bayer 1080 footage. Maybe we can save money by using Rob's idea of saving out during pause aswell. We could then use slower cheaper higher capacity single/double consumer drives (you will need lots of them for a movie).


Thanks

Wayne.

Another interesting article on using GPU programming for audio processing.

http://www.tomshardware.com/hardnews...02_135943.html

http://www.bionicfx.com/

Re-edit:

I seem to have found an article that says that the AGP backchannel is 266MB/s peak, more than enough to save out a comrpessed SHD stream (but the card/chipset architecher would have to be built to take advantage of this, ie. it could even be as slow as 32MB/s in some real life cards). Still PCI-E should be better.

http://graphics.tomshardware.com/gra...0/pcie-09.html

http://developer.nvidia.com/object/General_FAQ.html

On high 3D card power requirements. We don't nessacarily need to use the latest and greatest GPU, even a low powered integrated one might offer enough performance assistance.

Thanks

Wayne.

Here is something more interesting than relevant to us (only has HD component in). High DEF (3D I think) PVR with full Intervideo suit.

http://www.via.com.tw/en/Digital%20L...niC-Club3D.jsp

http://www.s3graphics.com/pressrel/2004_06_03.html

http://www.club-3d.com/en/club-3D.html

New Sony HDV.
 

Wooh Hoo, Wooh Hoo (thats sarcasm, I'm not that big of a Oprah fan). The new Sony HDV camera is finally here (on time not liek all those people saying it would be next year). Now, maybe, they will ALLOW, JVC and Sharp to release their new HD cameras. Got to read the specs, but I doubt it will allow raw output.

http://www.camcorderinfo.com/content...-Camcorder.htm

Quiet an improvement over the pd170, but In comparison to our project, the camera looks hobbled, the article says 960*1080 pixel chips, but has 3 lux performance.

No HDV camera will ever output RAW. That's not in the HDV spec,
nor in the DV spec. Not to mention how they would store it anway.

Personally I don't see how an MPEG2 stream is going to be the
future of HD(V). But maybe I haven't been converted yet <g>

Sorry I have taken so long, I was going to re-edit that but the machine crashed in the middle of it.

It was a bit of a joke, but since we have started this project we have seen an unconfirmed report of a $2K 3CCD JVC HD camera, now this cheaper price of the Sony to $3700, which is just enough to discourage people from a project like ours, it is tempting to think we are getting too them ;) If we are successfull then they will need to include raw connections in future to keep people away from us.

I'm just including this as a comparision to our cameras.

Thought so Wayne, just wanted to make sure <g>

guys I hate to burst your bubble that has been floating about for some time but that JVC camera is $20,000 not $2,000

Finally got around to phoning JVC america today (no response by email in past), and got some guy on the phone, who thought his computer screen was an oracle or something, no mention of the camera, couldn't even ask him who I could speak to about it, or send him a link to his site that proved the camera existed. I had to explain, to him, that I might aswell just leave it until it was released, as it was probably preliminary information and the camera was most likely delayed (but went in one ear and out the other). Spent a few hours googling aswell, and nobody else has a price listed, even the shop companies with the camera listed.

So nobody has gotten a confirmation on the price in the video systems article, except for my post about asking a local Australian agent, who thought the $20K I suggested was more likely (but still knew nothing).

Somebody started a new thread about a 1024*768 CCD Sony Box camera with capture software, and firewire, for around $1.5K, thats getting pretty close. So maybe the big guys are catching on, and trying to stramp out all the effort we have invested in these cameras. As long as they aviode a HD resoltion 3 chip camera, Sony can feel reasonably confident in talking a pro customer into spending a lot more on a HD spec camera, or buy the inferior box camera.

Hey everyone,

Been out of the loop for a while. Hey wayne, can you give me the 1 paragraph low-down on how the cameras are coming? What about the Steve Nordhauser cameras? Summix? And seems you're getting strange other rumors now, huh?

Been spending all my time editing for a customer.

Thanks!

My new mac is working beautifully. Here's an HD capture card that just came out from aurora.

http://www.dv.com/reviews/reviews_it...cleId=38200001

Might help me when I get the camera, if it's SDI.

Pretty much the same as last time. The micron 1080 camera is out, a bit dark compared to the 1280. The 1920HD is supposed to be close (maybe October). Other people are silently working on other things, like sharks (not a negative, sharks dive and come up from underneath unseens). With the JVC, we can't really say anything until we get a price directly from JVC, or one of the cameras resellers. But, except for 3chip SDI compliance, don't expect it to be as good as the camerlink solutions. Have a look at the other thread, I have taken a bit of a break, as I've been out of it recently.

Thanks

Wayne.

Community Technical Update.
 

Re-edit: Forgot to mention these:

Camcorder info: Goes to IBC show and talks, a little, about the pro version of the Sony HDV camera for 6KEuro, and JVC still showing off it's camera behind closed doors.

http://www.camcorderinfo.com/content...how-Report.htm

http://www.camcorderinfo.com/content...-Camcorder.htm


Ultra Wide Band USB/Firewire is coming with, Motorola spin off expecting to raise it to 1 Gb/s within 12 months.

http://www.theregister.com/2004/06/0...e_uwb_roadmap/
http://www.theregister.com/2004/02/1...up_dumps_ieee/


There is a low powered CE-ATA interface coming out for hard disks for handhelds.

http://www.tomshardware.com/hardnews...909_162037.htm


New 3GB/s SATA-IO development

http://www.tomshardware.com/hardnews...09_161726.html
http://www.sata-io.org/


Finger print recognition, though not really needed for our cameras, it still can insure that nobody can accidentally access it when your not looking, or get it working if they steal it.

Microsoft is the latest to introduce new hardware:

http://www.tomshardware.com/hardnews...08_223459.html

Give me details on the micron 1080 camera (specs please)

Have just been to the IBC conference daily news site, this conference will most likely have most of the HD news for the next couple of months.

I have been informed of a new box camera to be released at this conference, but haven't seen it yet.

1920*720 50 HD, anyone, , they are allready talking about next gen HD, how many formats will we get?
http://hugecgi.com/cgi-bin/ibc_daily...=22712&issue=5

Avid's DNxHD
http://hugecgi.com/cgi-bin/ibc_daily...=22559&issue=5

D-Cinema talk
http://hugecgi.com/cgi-bin/ibc_daily...=22605&issue=5

Planned 180 cinemas accross Europe:
http://hugecgi.com/cgi-bin/ibc_daily...=22714&issue=5

Our competition (apart from that HDV Sony ;) Arri digital 35mm handheld (??).
http://hugecgi.com/cgi-bin/ibc_daily...=22701&issue=5


Thanks

Wayne.

<<<-- Originally posted by Laurence Maher : Give me details on the micron 1080 camera (specs please) -->>>

It will be at the SI site as 3170 (or was that 3300), micron should also have some sensor info. Alltogether the Altasens is much better, but also more expensive.

I think I'm going to say something here more as a challenge, not against SI's excellent work, but to Micron. I would like these sensors for still, machine or security cameras, but for broadcast it is what we should expect from a good prosumer single chip HDV.

The article on the Arri above, has them saying they have finally solved the reduced S/N problem of cmos compared to CCD, that means that existing cmos cameras must not be upto it. I have noted the SN specs are not as good as I hoped in times past, and yes the Altasens S/N is a minium I would like to see, not a maxium. You can't reach down into the darkness, at gain, without visible niose, without it. Now I "know" that there are more than one person here who would love to take me up on that statement, but I'm sticking by it. So if anybody wants to challenge this, go and film a room lit by a 5 lux light source compared to a manually adjusted PD170 (by no means a low light winner compared to pro eng cameras). Where would you use it, ala natural, doco's, and lite lighting, I have, looks nice.

All in all, if micron priced the sensors we are using, so that cameras (on their own) were available from $200, I personally would go for it in a sec. But if you want to save a little money and quality is high enough, I would go for them at the current price, but if quality is prime go for the Altasens.

Hey Wayne,

The SN's on the Micron chips are awful-only around 46-48db. Forget the dynamic range of 68db that they advertise, to get that dynamic range you're pumping the image full of noise. SD cameras that are CCD-based and interlaced are using pixel binning/line-mixing to get better S/N ratios, that's why your PD-150's look so good in the dark. Progressive cameras can't play those tricks, so for instance the Sony's F900 HD camera only has a S/N ratio in progressive mode of 54db, while the Altasens has a typical S/N of 60db+ for it's progressive images. So Altasens has also "solved" the S/N problems of CMOS like ARRI claims, and IMHO, the first mass-producing company to "solve" this problem was Canon with their D30 CMOS chip.

Dual Slope
 

For people new, and old, to the project, I thought I would get around to archiving some posts on my favourite subject, dual slope (also bits on Global shutter and image performance):

Seems like a good idea (except for rolling shutter problem on Micron). Please note, for people who don't realise, this only expands range upwards, and it will sacrifice some lower light ability. I think human vision needs more range in bright scenes than low light, so I think it is a good compromise. We calculate a piont to split the frame rate, maybe just below where the pixels will max out then let the next frame have the rest of the rate. But the timing of this split should be minimal to not let the frame change too much (but then movement blurs anyway, so how much of a problem is this) now the problem is this fast frame will spike bus data a lot. Can we change the linear performance to something a bit better this way. I can't get my head around it at the moment, what do you guys think?

Yes I have heard that Altasens has some sort of fix.

That was that Auto gain feature, I was thinking about.

I admit it, I was basing my statement on them crowing they were the best compared to all other cmos sensors (meaning everybody else must be behind CCD's), and what I've used before. Are any of the top HD CCD cameras better?

I would like something at least as good as the old JVC GY-DV500, better than the PD150, but much less than modern ENG. The spec sheet talks about 60db or more SN and 68db dyn Ran, (and f1.4 prism, good). That's what I would call minimal for our purposes (and ironically around the Altasens specs). I know we are talking Apples and Oranges as far as pixel size is concerned, but that has allways been my production goal.

Yes I remember noticing the Dynamic range/SN micron thing before (should have remembered that one).

Our eyesight's resolution, framerate and dynamic range (bits per pixel) drops in the dark, I'm prepared to live with half of 1080 binning, or something like that, for low light, the audience's visual centres will be expecting it anyway. PD-150 is a poor example anyway, I think pro ENG CCD designs are just more refined in S/N and range (not to mention big glass and big chips).

Regards,

Thanks

Wayne.

the whole point of this project is lowcost filming..not DOC work! if you are going to shoot a "real" project then you better have a nice light set! and if you want to shoot a DOC then buy a SOny HDV cam..it will be GREAT for DOC work...I wish people would lay off the lowlight issue..it's a non issue

This project has allways BEEN a general work camera project, not a privatist venture. Your camera maybe a cinema camera but that little market is not what the general, original, Steve I, Sumix (pluss all cameras including SI), project was all about (but it was part of it, the cream on the cake that the idies would get off the larger marketed work camera). The only reason I put "cinema camera" in the title of this thread was so people would not get confused. As such it includes the project discussed in the Steve I thread and as a side track (as many people where from that thread) the SI only based project. The fancy thread title gives a fancy title reflecting the maxium extent work camera systems will go (i.e, we wont be aiming to be scientifically filming sunspots with Brad Pitt clearly seen as a reference in the foreground ;).

I hope it's not too difficult to realise, that being so general it can both include cameras that are good for cinema, and cameras that are good for ala natural doco. And to realise that ala natural cameras (the medium compressed detail on HDV being a burden, and the camera less suitable in natural scenes) under this project, are also easier to setup lighting for cinema filming, and probably will produce better results with the weaker lighting.

Remember the rest of the potential customer base that can help lower the price for all of us. We are starting with cinema cameras, but that is not where it should end.

Have fun.

Well, following your thoughts Wayne , I realized a couple of things.
I always thought it was a project for Cinema making.
If it is thought for making "normal' Docs it should follow a different approach.
For normal Docs, everything is more or less at hand by now.
We don't need a large area sensor (to get decent DOF).
Just get the SI 3300 at 30p, the Epix FG, the software from the Robs, A pentium M mini-itx or 5.25 board and ready.It would be enough to get a very good image for a minimum price.
After testing some images Steve N. took I'm beginning to think that in fact it is much better than what has been said before..
(if you give me your e-mail I can send you the image I got after de-mosaic)

If sensitivity is an important issue, just use Medium format lenses instead of SLR (it's been tested and yes, you get increased sensitivity if used with a relay lens adapter).Medium format lenses aren't so expensive when bought second hand.

SI 3300 is 10 bits linear so getting 8 bit log out of it would be enough.An 8 bit uncompressed ( I mean uncompressed and mathematically lossless compressed are the same thing to me) has a lot of room for CC anyway. (I've been correcting DV for several years with decent results, so....).

8 bit+lossless will let you use just one disk (and at 1920x1080!!!)as if it were a tape cartridge and would be really practical for DOCs....
No need to tell you that you could also get a couple of audio channels at 16 bit 44 Khz (also lossless compressed if you like).
That kind of system shouldn't cost more than 3,500~4,000 for the whole thing...


Even the new Sony HDV camera would be a great choice for documentary work at around the same price...

The whole idea is that we have many many cameras on the market NOW that anyone can buy and make a DOC. DOC Directing/Producing/Shooting is nothing more then a well told extended news story like you watch on TV every day..the whole point of this camera is Cinema because the standard resolution of DOC cameras is not enough to project on the big screen...ALL THE NETWORKS use SD cameras for DOCS all day long that is a non-issue for them! GET REAL WAYNE.

I can shoot a DOC anyday I want with the Panasonic 24p camera I have or the JVC 500 or the Sony vx1000 hell I could even go to BestBuy and buy the $600 3ccd minidv camera and shoot a DOC that ANY network would air at this moment in time.

how much have you been paid this year Wayne from your SD compressed 4:1:1 camera? Mine makes me a living everyr year..and nobody ever asks for anything better..not untill HD is the norm and even then they will accept HDV if that is a format that the big boys push on us!

So no I would not say this is about DOC projects..not to say that it will not work..you can resize for NTSC but what is the point? none....unless you want to create lots of extra work and render times

Jaun, please don't get offended if the following is too simple and detailed, I am writing it so newcomers to this project can understand the issues we are talking about.
-------------------

re-edit: I should also put, that part of the project is specifically taylored to Cinema, like what is happening at the other thread, and any individual can do what he likes with his camera, though we should not forget everybody else. I give a helping hand to a better future, you don't see me dictating like some other people.

Yes, I was there near the beginning so I read everything that has happened (but scanned a months worth too). I appreciate your arguments, but I want a doco camera that can capture the scene with such visual clarity as to be comfortable to the viewer. Big sensor chip and light gathering power, allow you to naturally follow people into normal lighting situations without the annoying spot "(d)effect" all the time, letting the veiwer veiw something that looks like what they expect to see in those situations (though a bit moody and artistic), rather than an episode of "Cops". Though I would use a spot in the dark at night, there are many instances in the daytime where alot of cameras need a spot if somebody walks through a dim room, that a good camera will just jog through.

We had a doco on here the other night, a guy who walked accross Australia totally unassisted living off the land (wonder how he recharged the camera). I think he had a PD150/170 or VX2000/VX2100. The thing looked like a kindegaden crayon art class, the shiny leaves on the scenery blew out (many leaves are birght in the dry regions with whitish undersides). (I know he needed a ND or to adjust the iris, but most of the scene was OK, it was just the range, the footage of him being attacked by wild dingos at night was acceptable). It looked horrible but still watchable. Now I don't expect the Sony HDV to be anybetter. So a bigger 2/3" chip on the Altalsens would have been able to render that scene much better.

So in both cases we need the range and sensitivity to match the footage that could be shoot on a JVCGY-DV500 ENG (let alone a 5000, or new JVC HDV). Many handheld doco's don't look as good.

Now if I was filming a cheap, low low budget cinema production, at night. Steet/sidewalk lighting affords some pretty good moody scenes around here, a camera that can capture that would be good. But it is a production preference, others might want to light everything up like a light house for a cheerier lighting scheeme (mood again).

The one thing I warn people about is not to get stuck with equipment that is only good for shootting in one lighting style (like the older high stop loss 35mm adaptors, where everything rolls off into black fast). Whatever style it is, it might be trendy for the moment, but you'll want range enough to do something different eventually.
--------------------

The medium format idea, yes I have been advocating that for over a year. I am still waiting to find a good MF prime at a good price, and am planning to get a SLR zoom with wide enough circle to use as a medium format zoom.

I've been advocating different levels of camera (handheld, Eng, Cinema) for a while, from cheap to $5K altalsens systems. Thats why, even though we are starting with lossless at the moment, I allways advocate lossy compression afterwards, from Cineform quality, through DVCpro50 quality down to HDV quality, to suit every occasion and use. So I can come significantly cheaper than you suggested for single chip 720p system, but not with altalsens. I am waiting to see what happens with the compressed cameras (if I can wait), and then use cheaper ITX VIA system to save the compressed stream to single or dual disks, in a smaller package. I would love to use a Camera with standard interface instead of a framegrabber too, but the price is not cheap enough to warrant the datarate restrictions compared to a framegrabber.

Much love

Wayne.

"The thing looked like a kindegaden crayon art class, the shiny leaves on the scenery blew out"

this shows your non understanding of the subject at hand, your talking about HOTSPOTS and blowouts not lowlight!

what side are you on Wayne?