View Full Version : Alternatives to SI and Sumix


Pages : [1] 2

John Nagle
November 20th, 2004, 12:28 AM
Hello everybody,

Everyone seems to have settled on SI or Sumix as the only box cameras that could be used to build a HD system.
From reading all the threads I have noticed that some people raise the question of other possibilities but get lost in the hustle so I thought it might be useful to start a thread where people can discuss and post links to other possible solutions.

Juan M. M. Fiebelkorn
November 20th, 2004, 07:59 PM
Good Idea.Anyway people is interested on SI and Sumix because they have given us their support, just that.

John Nagle
November 20th, 2004, 10:48 PM
Hi Juan,

I know what you mean I emailed Steve at SI and found him to be very helpful. I just thought I would start this thread so if anyone wishes to discuss alternatives they can.
I have seen it come up a few times in the other threads but those are really more about getting working systems together. I don’t know if this will be of any use but I am not a programmer and I felt a need to try to contribute something.

John Nagle
November 20th, 2004, 11:22 PM
I know the monochrome version was posted before but Pixelink have a new colour version out next month

http://www.pixelink.com/products_info.asp?id=43

maybe an option ?

Laurence Maher
November 22nd, 2004, 09:19 PM
Wow,

Says good resolution and variable frame rate. Global shutter which is good. Do you know about the Mhz, or transfer rate, ? I assume you can do full resolution all the way up to 105 fps. (Awesome). C-mount is cool. Does it record to a PC? Mac? Does it matter as long as it is firewire? Mentions Camera control without software. Is it on-camera? Is it progressive or interlaced framing?

Thanks for the post. Very curious!

Valeriu Campan
November 23rd, 2004, 01:39 AM
There is a frame rate calculator here:
http://66.148.40.196/uploads/pl-a74x_frame_rate_cal.xls.
Steve N mentioned in a post that the kit is based on a IBIS-5A camera. The SDK seems to be for Windows only so far.

Rob Lohman
November 23rd, 2004, 07:54 AM
Weren't there issues with most "machine vision" camera's that
would render them useless for cinematography? Like smear or
noise or something like that? I think I remember someone talking
about that.

John Nagle
November 23rd, 2004, 01:57 PM
I was looking at the Xcap web site and I found this camera and board package. I know some people here have had problems with Xcap but would this be a package that might be worth looking at?

http://www.epixinc.com/products/sv9m001.htm

The price seems good if the package is up to scratch.

Rob Lohman
November 24th, 2004, 05:26 AM
Xcap does what it says, but it is hard to use for cinematography.
That's why some people are writing their own "capture" software
so get it more in line with how we want to use a camera.

Jason Rodriguez
November 24th, 2004, 07:48 AM
Also Xcap isn't very efficient at writing to a Hard-drive, like maybe 40-50MB/s max. So for 1080p, you'll need to write to RAM, which isn't cheap.

John Nagle
November 24th, 2004, 03:05 PM
Thank you guys,

About how many seconds of footage could I realistically record with 1 gig of ram?

Rob Lohman
November 25th, 2004, 02:58 AM
That totally depends on:

1. the resolution of the frames coming in

2. the bitdepth of the images (8, 10 or 12 bits)

3. the framerate

4. the output format (uncompressed bayer, packed bayer, uncompressed RGB etc.)


But to give you some ideas:

1280 x 720 x 8 bits @ 24 fps in bayer = 21.09 MB/s (48 sec. for 1 GB)
1280 x 720 x 12 bits @ 24 fps in bayer = 42.19 MB/s (24 sec. for 1 GB)
1280 x 720 x 12 bits @ 24 fps in packed bayer = 31.64 MB/s (32 sec. for 1 GB)

1920 x 1080 x 12 bits @ 24 fps in bayer = 94.92 MB/s (10 sec. for 1 GB)
1920 x 1080 x 12 bits @ 24 fps in packed bayer = 71.19 MB/s (14 sec. for 1 GB)

1920 x 1080 x 12 bits @ 48 fps in packed bayer = 142.38 MB/s (7 sec. for 1 GB)
1920 x 1080 x 12 bits @ 48 fps in RGB = 569.53 MB/s (1 sec. for 1 GB)

Wayne Morellini
November 28th, 2004, 09:00 AM
John, people just don't care, thats their funeral. I have been in discussions with people from other companies that would love to be involved, but if nobodies ionterested, wel (sorry bit ticked off today).

I have listed this in my technical thread (and the 10 bit threads), and you will find links to lists of many many cameras and systems there just waiting for somebody to wade through them all and acess them for suitability. I don't know if you are likely to find suitable replacements for the SI and Sumix cameras, and the Altasens is far better. I have suspicions on two. But I would suggest hunt down the mobile phone and digital camera sensor makers and see if any of their products squeeze accross the 720p line (actual throughput/bytes per pixel (depends on the send format) /1280*720/desired readout fraction of a second=1) then check if that is below the actual max frame rate.

Now you will need very fast readout (that is what that desired readout fraction of a second was about), or you will need global shutter.

John Nagle
November 28th, 2004, 12:26 PM
Wayne,

I know on your other thread you were talking about the possibility of really cheap alternatives, I don’t know if you will be interested in this but I will post it anyway.

I know this is not what we are after but it might give some ideas.

http://www.astrosurf.com/astrobond/ebrawe.htm

Régine Weinberg
November 30th, 2004, 03:44 AM
Great work
that is a nice page
and the Linux stuff
on a french page
I can not believe it
WOW

Laurence Maher
December 1st, 2004, 11:42 PM
Tell me what you think about this box camera built "specifically for digital cinema/tv" . . .

http://www.isgchips.com/Templates/t_quadhdtv.htm

Looks incredible, but I imagine it costs a lot?

John Nagle
December 2nd, 2004, 12:18 AM
Ronald,

I hope you can make some use of it and if you think any of the ideas there might translate to the projects here please post.

Laurence,

That is a serious camera, very big on the WOW! factor only trouble is it’s also very big on the price factor as well $85k as far as I know.

Larry Liang
December 22nd, 2004, 06:07 AM
The price for the bare 4-HD sensor is around 50K $. Olympus has developed one type of HD camera based on such sensor.

For alternative choice of camera head, pls. check www.elphel.com

Elphel's camera use the micron's 1.3 Mega and 3 Mega sensor. It can output Motion-Jpeg stream with a speed of 22 fps@1280*1024, 31fps@1280*720.

With Elphel's camera, it is much easier for storage, because the video has been compressed before output, but hard to display: you need additional CPU capacity of decoding the JPEG picture. Generally, To display the video with full speed and resolution, a 3G P4.CPU is needed.

Conor Ryan
December 22nd, 2004, 10:13 PM
I saw this camera, CCD, 2048x2048, cameralink. according to my (suspect) calculations it will do 60fps at 2048x1080. I could be wildly wrong here, so take it with a pinch of salt. Neither have I seen pictures from this camera.

on the upside, f-mount, and a 4/3" chip.
on the downside, costs a little under $6000.

http://www.cohu-cameras.com/products/7900.htm

Juan M. M. Fiebelkorn
December 23rd, 2004, 07:12 AM
How on earth are you able to get 60 fps at 2048x1080 from a sensor wich runs at 8 fps at 2048x2048 ???

Larry Liang
December 23rd, 2004, 06:52 PM
Generally, the CCD sensor's data rate is lower than that of CMOS sensor.
if the fps is 8 at 2048 by 2048, it will less than 16 at 2048 by 1080
I have find some CCD sensor with very high data rate, but such sensor is only monochrome type, seems only designed for the 3-CCD camera.

http://www.semicon.panasonic.co.jp/ds/eng/SMD00014AEC.pdf

Conor Ryan
December 23rd, 2004, 07:24 PM
well, i said i could be wildly wrong. if you mail them they can send you out their calculation formula.

Larry Liang
December 23rd, 2004, 07:37 PM
The limitation for fps is the maximum frequency of the sensor's clock. Normally, with partial scanning, higher fps can be achieved, but the total pixel is output in certain time is almost constant ( only a littler lower for small window).

Valeriu Campan
December 24th, 2004, 08:52 PM
Merry Christmas all!

The new "Motion" series cameras from the RedLake is quite interesting: http://redlake.com/high_speed/index.html.
High speed (650fps!!!) at 1280x1024 with 12 micron size pixels. Sensor size is 1", larger than s16mm and closer to 35mm movie frame size.
The cameras have internal memory (4G) where data can be stored before dowloading to laptop computer (PC and/or Mac!!!) via USB cable; no frame grabbers and the host computer is not a high spec machine.

Larry Liang
December 24th, 2004, 09:47 PM
Merry X'mas!

The RedLake should ( I guess) the Micron's MT9M413 sensor. Such sensor has a sensitive area of 15.4mm by 12.3, smaller the 1/4 of 35mm film.
http://download.micron.com/pdf/datasheets/imaging/mt9m413c36stc.pdf

4 GB memory can store about 100 second video in 1280*720*30fps*10bit.

Valeriu Campan
December 24th, 2004, 10:51 PM
From the RedLake's brochures I understand that you can transfer the data to the computer as well, by-passing the internal memory. I must read the manual closer to see if that's possible.

Valeriu Campan
December 29th, 2004, 02:22 AM
A recent re-visit to a camera site came up with this:
http://www.visiblesolutions.com/phantomv9.html
A bit scary: global shutter, 1000fps at full sensor (1600x1200), 600 ISO/color, Nikon lenses, optional 12G on board memory will deliver acording to my calulations ~ 4.4 mins@24fps (6400 frames at full resoultion - same as the 400 foot magazine on a 35mm camera), optional continuous output to a external storage device via Gbit ethernet, firewire and SDI output (PAL, NTSC)...
Am I dreaming? Correct me if I'm wrong!
Should find out the price of this toy... Hope is not $$$$$

Laurence Maher
December 29th, 2004, 03:09 AM
Uh . . . wow . . . this thing looks tastey.

Someone find out the price please.

Larry Liang
December 29th, 2004, 05:04 AM
I guess, 10K$

John Nagle
December 29th, 2004, 10:50 AM
I think the Phantom cameras have been mentioned before and I think they are a lot more that $10k.

I hope I am wrong and have emailed them for a price list I will post when I get a reply.

Juan M. M. Fiebelkorn
December 29th, 2004, 07:57 PM
The Phantom costs around $ 95,000.
I got a quotation for it.

Valeriu Campan
December 29th, 2004, 08:01 PM
Woooops! Maybe a 95% discount will help.

Rob Lohman
January 3rd, 2005, 03:45 AM
Keep in mind that such gear is targated to the PROFESSIONAL
world. 95K is pretty cheap there, especially for such a system!

It isn't targeted for the con-/prosumer market!

Filip Kovcin
January 3rd, 2005, 07:43 PM
<<<-- Originally posted by Juan M. M. Fiebelkorn : The Phantom costs around $ 95,000.
I got a quotation for it. -->>>

can you tell us WHICH phantom camera is quoted?
i can see that there are more cameras...

and 95K? on their site i also saw 95K but that was just "speed limit"sign.

take a look:
(scroll a bit down)

http://www.visiblesolutions.com/phantomv5.html

he, he is this a concidence? ;)

filip

Juan M. M. Fiebelkorn
January 3rd, 2005, 10:10 PM
If you don't like it, it is easy.Send them a mail and ask for a quotation.

Filip Kovcin
January 4th, 2005, 04:01 AM
no, no...

please do not treat my previous mail so seriously.

first part of my mail was, i should say "normal" - i just asked about other models - since you said that certain model was for 95K. nothing else. i thought you could know more then us. since you asked them for the prices etc.

second part of my mail was more joke like -

i just saw coincidence between your mentioned price and sign on their site where also 95K is mentioned, but in different context.
(it says: speed limit 95k per second)

sorry if you find that somehow attacking. this was NOT my intention at all.

thank you,

filip

Juan M. M. Fiebelkorn
January 4th, 2005, 02:11 PM
ohh.no problem! :)
I thought you were saying the price was wrong.Maybe now with all these new high speed cameras coming out they get cheaper.
The model was the last one.Don't remember if it was Phantom 7 or Phantom 9.

Wayne Morellini
January 4th, 2005, 02:45 PM
Good effort, maybe some prices on these cameras people are ralking about.

The whole sale price of the IBIS5a sensor is around 100 eruo, so if you look around you might find somebody that will sell a camera head for $500. There are many lists to look up these things. The cameralink standards organisation has links to companies. Capture board companies have lists with even more companies, then there is the astro list above etc. But for X hunderd of dollers more you can get the high speed version of the FillF and Micron sensors which optically are much better (as far as I know). Even though these sensors can go several hundered FPS global shutter (I think) you don't need use it and make a expensive capture system. A relatively cheap camera system could be made using one of these sensors at a low frame rate instead. As far as I remember these sensors costed several hundered dollers, but that is old information, the ibis5a has been dropping in price quiet a bit, maybe these are to. So it is possible that we could be looking at a low speed camera head for $1500 or less. Then again I could be totally wrong on pricing and it is still more.

The truth is for cheap quality cmos cameras there are very few chioces, apart from Fillf and Micron. I would suggest www.smalcamera.com. Even if they don't have what your after they could make it (including the sensor).

Wayne.

John Nagle
January 4th, 2005, 02:58 PM
Wayne,

Just wanted to check that link for smallcamera.com, I followed it and there is only a site under construction

Valeriu Campan
January 4th, 2005, 10:19 PM
I was looking at some sample frames from the Pixelink PL-A742 camera:
http://66.148.40.196/uploads/PL-A742-E6ms-G0db-S125-GMoff.bmp
I noticed a VERY strong "chromatic aberration" artifact. I was told by a Pixelink rep that that is a "normal" CMOS/lens artifact and the chart was shot with a cheap lens. AFAIK, a better lens will allow this kind of artifacts to be even more noticeable, especially with high contrast subjects. Any thoughts on this?
Looking at the chart there is no way to use this on any HD or even SD system.

Obin Olson
January 4th, 2005, 10:55 PM
Wow that is a new PixeLink camera..I would try that one out..I bet it's got the Micron chip..not to good but not as bad as IbiS5

John Nagle
January 4th, 2005, 11:05 PM
Hi guys,
I looked at this Pixelink model a while back; it uses an IBIS5A sensor so all the regular problems would apply. That said, the Drake uses an IBIS5A so depending on the A/D converters it might still be an option.

Valeriu Campan
January 5th, 2005, 12:06 AM
The pixelink camera has an IBIS5 chip with 6.7micron pixels. I will speak tomorrow with a tech from a local distributor about it.
The Micron chips are also very interesting, though they are much larger (12micron pixels) with a diagonal size of 19.67mm (between s16 and 35mm gate)... Sweet, very sweet:
http://www.micron.com/products/imaging/products/MT9M413.html
I wonder if there are any camera heads made around this chip (I think that Basler A-500 series).
I still would like to know your thoughts about the chromatic aberration displayed by the Pixelink.

Juan M. M. Fiebelkorn
January 5th, 2005, 06:26 AM
I guess that chromatic aberrations you see are in part because of really bad optics.The rest is caused by a crappy debayering algorithm

John Nagle
January 5th, 2005, 12:02 PM
There has been a lot of talk about the IBIS5 on the forum, the problems with the sensor seem to revolve around needing to much light and getting 24 fps with decent exposure times so I would be inclined to agree with Juan about the optics and debayering.

Valeriu Campan
January 5th, 2005, 05:40 PM
Thank you for the reply. This is part of the message I received fromPixelink:

"..._The lens used for these images was an inexpensive C-mount lens that does not have any correction for_chromatic aberrations. With a high resolving power, the Bayer artifacts become more noticeable._ As the lens resolving power is reduced (or the focus moved away from the ideal) the Bayer artifacts go away..."

Imagine what will happen if you shoot with Zeiss Digiprimes or Cooke S4!
I will talk to a tech from the local distributor today and try to pick his brain.

Wayne Morellini
January 5th, 2005, 10:56 PM
<<<-- Originally posted by John Nagle : Wayne,

Just wanted to check that link for smallcamera.com, I followed it and there is only a site under construction -->>>

Sorry about that John, it is actually http://www.smalcamera.com/ with one L.

I'm just chucking this in here, I think with a bit of co-op work they might come to the table, but still you can check the links to the sample images etc in my technical thread. I don't think it is the most brilliant sensor, but excellent as a cheap alternative to Altasens/Ibis5a.

Now the Ibis5a is supposed to use a special pad around it's sensor to collect stray light that misses the sensor, without using a microlense (which redices the maxiumnum aperature that canbe used). I think it was also to recude niose, but still people says it needs lots of light, strange marketing technology. I wonder what it actually does. Out of curiousity are the images dull in color, or just washed out in lower light levels?

Re-edit: Just had a look at some of the sensor specs again, rolling shutter, but enough data rate to run higher shutter (but calculations for region of interest regioning max shutter not made clear). Again this is a good sensor for a cheap (even doco special) camera, not Altasens quality. There is hidden (somewhere) their 1920 wide security camera (which would need two on a split prism), but seriously they have there own sensor designs and produce there own cameras, they would have a lot of designs they could make hidden from the public.

Thanks.

Lester Mclewis
February 1st, 2005, 06:54 AM
Check Out the Basler 622f camera, Who knows it might be the one used in the Drake(Just a Guess)

Steve Nordhauser
February 1st, 2005, 09:21 AM
Lester,
The Basler is an IBIS-5A again. You can tell from: 6.7 micron pitch, CMOS, global shutter, 1.3Mpix, 24fps.

Wayne,
The IBIS-5A has some adjacent pixel smearing which ruins the color purity.

Jon Schroth
February 1st, 2005, 04:34 PM
Anybody know how the mono version of the IBIS-5A performs?

It seems that most of the complaints about this chip are related to color.

Also, in general, I am curious how a bayer filtered color image that has been converted to grayscale differs visually from one captured in monochrome.