Hey guys, sorry if i'm a bit behind the curve on this, but i've been reading on the technical disscussion thread and concerning the SI-1920HD; what would be the difference in going with this cam vs. the SI-3300M. I know that the 1920 is a 2/3" chip at 2 mpixel and the 3300 has a 1/2" chip at 3 mpixel. Is there a significant difference in resolution, pic quality, or light latitude? Wouldnt the 3300 @ 3mpixel be output a better image or does the larger chip on the 1920 make that big of a difference? Just wondering what would be the better cam to go with in putting together an HD camera setup. Thanks, Omar

Jon,
We sell the SI-1280F (IBIS-5A) in mono and color. You will get the best mono image with a mono sensor. The IBIS-5A is already not too sensitive - you toss over half the light with a Bayer filter. You also do not get an even spectral response at every pixel. For example, a pure red object will show up on 25% of the pixels of a color sensor - you lost spatial resolution. It will show up on all the pixels of a mono sensor.

Omar,
The SI-3300 has smaller pixels (3.2 microns vs 5.0 on the SI-1920HD) so it is less light sensitive. It also has a max clocking rate of about 60Mpixels/sec (24fps) where the SI-1920HD can do 150Mpix/sec. This allows faster readout so you can minimize rolling shutter artifacts, even at a frame rate of 24fps. As you said, the larger sensor gives you a shallower DOF. Both are low noise sensors, the 1920HD is better. The SI-3300 is viable for a camera, especially for the cost differential.

<<<-- Originally posted by Steve Nordhauser :
Wayne,
The IBIS-5A has some adjacent pixel smearing which ruins the color purity. -->>>

There is allways something ;) Are the high speed versions of the IBIS and Micron 1.3MP a viable cheaper camera to the Altasens? Say, could you make a camera for $1-2K out of them? Their architectures are far superior to the IBIS and micron (global shutter aren't they).

Now my question on the other thread, you mentioned a memory buffered GIGE camera, I believe, sometime ago, when is that coming? It will be interesting to see the tests when they start rolling in of the new cameras.

Well,

I love you guys who all tried, but as I predicted, it looks like it's all going to be way too late.

STRONG rumors suggest a Panasonic DVCProHD camera coming out for under 10 k some time this year. NAB will most likely unleash it. Records to mini-dv and also P2. The quality won't meet the "uncompressed" richness of what you guys have been trying for, but for cost-effectiveness, and rock-solid reliability, and user-friendliness, the trade off will be well worth it.

Sorry guys, I think they beat you to the punch.

. . . like I predicted.

Elvis has left the building.

Well you would be better going for the FX1 through component output. Now there is no gaurantee you won't be able to do this through the Pana, but maybe Juan will offer a RAW solution for it. But seriously, without these, even the IBIS and Micron RAW are still worh looking at. It is all a compromise between price, features, performance and convience, which you would like.

<<<-- Originally posted by Laurence Maher : Well,

I love you guys who all tried, but as I predicted, it looks like it's all going to be way too late.

STRONG rumors suggest a Panasonic DVCProHD camera coming out for under 10 k some time this year. NAB will most likely unleash it. Records to mini-dv and also P2. The quality won't meet the "uncompressed" richness of what you guys have been trying for, but for cost-effectiveness, and rock-solid reliability, and user-friendliness, the trade off will be well worth it.

Sorry guys, I think they beat you to the punch.
-->>>

NO! This will be just another DV Toy. No DOF, together with extrem compression. Just home made video pur. No, no. Let others play with those "kindergarten" things...

<<<-- Originally posted by Rai Orz : NO! This will be just another DV Toy. No DOF, together with extrem compression. Just home made video pur. No, no. Let others play with those "kindergarten" things... -->>>

Anything recording to 100mbs DVCProHD is not a DV toy. It's not uncompressed video, but you can get great professional images out of DVCProHD.

Indeed. The Varicam has been used on some very, very high end stuff and it uses DVCproHD. This isn't a lightweight format.

That said, it isn't an uncompressed camera. The market is absolutely still there.

Do you realy think the big companys bring out a realy good HD camera under 10 k ? Why? HDTV okay, but wait for the downsides (small sensors, compression artefacts, etc.)

Oh I'm not saying it will replace your camera. Far from it. I'm just saying that it's an interesting option.

A misunderstood: Shure DVCproHD is very, very high end. But i say, it will be not the best camera. We will see. Its like the statement "digital is better than analog". Just wait

Have they ever mentioned anything above 25mb/s stream? If they offer 50 or 100Mbs, that would be good, if they went for $9999, then maybe that is what they are planning, and many professionals would go for that.

For Rai:

No, its not the best. Forgetting about broadcasting for the moment, they know now that there is a definite market for people who can't afford the best and are willing to make a few sacrifices in image quality in order to get something better than they had.

And that's the market you're going after, Rai. The indie filmmakers, the MiniDV guys who are tired of MiniDV and want something better. And price is a major, major component to guys like this. You can't just say "It's worth the price," because they'll come back with "But I can't afford the price," and they'll go with a cheaper, almost as good cam. And if the difference between the Panasonic and the Drake is $5K or more, a lot of people will go with the Panasonic (regardless of the extras the Drake has) because of that $5K.

The race just got a lot more cutthroat.

For Wayne:

According to Jan, it will do 25mb/s, 50mb/s, and 100mb/s (though you need the pricey PS2 card to do 50 and 100).

Not bad, but price. I think the drake will be much more value for money to a cinetographer.

So the codec on the Pana, how is it for motion artifacts, will they have variable comrpession to handle this? What image format does that record to tape again?

Thanks

Wayne.

Still back to reality if it's 10K, we are instead looking at 5K systems (variouse projects here) with a computer that could be used for editing and HDD that are much much much cheaper than P2. Now if it was $5K (no P2 included) then yes there is a case. Realistically, I think we can go much lower than $5K for some projects.

I think every project will go into different markets.

<<<-- Originally posted by Wayne Morellini : Not bad, but price. I think the drake will be much more value for money to a cinetographer. -->>>

For a lot of the guys who are going to be looking at these cameras, I think price is going to outweigh value, especially when the value of the cheaper camera isn't that bad.

<<<-- Originally posted by Wayne Morellini : Still back to reality if it's 10K, we are instead looking at 5K systems (variouse projects here) with a computer that could be used for editing and HDD that are much much much cheaper than P2. -->>>

That would be great if we could get them, but none of the systems being built here are going to be sold for $5K, whether we want it to or not. They've said the Drake isn't. Obin's system costs about $5K in parts alone, so if you buy it from him, you're going to pay more than that, and he does need to make back all the development money he's been pouring into it.

Sure you can take his blueprints and build your own from $5K worth of parts, but then you still need software to make it run.

I think, right now, the $5K HD camera is a pipe dream unless there's someone out there willing to loose money in order to provide it.

That said, I'd buy Obin's camera for the price of the new Panasonic and feel pretty good about it.

Please, I want anybody here to search and post the price for a Beta SP camcorder.
Then if you can, post the price of a decent Video lens for it.
Just after that, let's start talking about pricing and if someone will buy or not an HD camera at 10,20 K.
Right?

<<<-- Originally posted by Joshua Starnes :I think, right now, the $5K HD camera is a pipe dream unless there's someone out there willing to loose money in order to provide it.

That said, I'd buy Obin's camera for the price of the new Panasonic and feel pretty good about it. -->>>

The cheapest price is going to be around $2K (incase you haven't read the threads). I would rather a convenional camera that spat out recordable 12bit 4:4:4 uncompressed images myself for the same price, but I'm not holding my breath.

Cameras are available from around $1K, though not the best ones, in future cheaper and better. Add system cost ontop of that ($500) software of $100, and batteries. The fact is that these systems are not terribly good in optical quality (except compared to say a HD10) so we don't bother about them, and prices will drop and other better systems will come out at this price. AAt the moment we are stuck between a rock and a hard place in terms of pricing to performance ratio, but still much better than many commercial cameras for filming.

I think my adaptor should be able to solve the blowout problem with the micron (want to use variable ND filer at least). But this blown out problem is tacky and unacceptable, something to be espected ona mobile phone cmos sensor, not on a $1K+ category camera, especiually considering hat they now do designs that eliminate it. But even after you pass all those there still is the rolling shutter problem, and I don't think anybody has found a full solution on the 1.3mp camera. But the reality is one new sensor will solve all thoise problems. My guess is there will be at least one this year.

Now out of curiosity, I notice 10 bit and MP sensors for mobile and cameras, and I wonder if any of these will outdo a 1.3MP Micron or Ibis chip? I have seen products for less $100 using chips with highspeed bus and better than normal Mobile phone and cheap digital performance. Once you have raw you are half way there. I am no talking about altasens beaters here, something more inbetween HD10 and FX1 quality, with raw and unclipped range still a bargain that fits in a sub $100 doller product. Now you see the whole system moves to $1K and beats a FX1.

The rolling shutter Death Sentence, this is also rediculouse, because the truth is it's where you can move the rolling shutter periode that matters. If it canbe moved into the inter frame period and accelerated (requiring a fast buffer to take he image) that it may become a non issue (stuff discussed many months ago). I don't know how successfull the experiments were, but people got enchanted by the Altasens that was due in a couple of months {:( and forgot about it. Now if we used a 48th second shutter on a 24fps (I think it is called a 180 degree shuter) image, can we move the whole rolling shuter effect into the non integrated periode? Is a mechanical shutter for 1/48th of a second a better way to achieve this?

So how successfull was removing rolling shuter effect expriments for he Micron and FF cameras, and how good is it going to be for the Altasens?

Every one will buy a HD camera, if it came under 10 k ?
Its remember me to the still digi cam run: "Your camera have only 2,3M pixel? Ohhh, my new camera have 4M pixel! And so on..."

There is a big market for those products. And a big profit, also if you see the following costs for P2... Lets people buy this. Why not.
We will go a different way.

But also Waynes dream is possible: A HD uncompressed system under 5K:
There a tons of senors for capturing in HD quality. But here in dvinfos alternative imaging methods, in the past we look for CMOS sensors, because CMOS produced mutch more "film look" like CCDs. But what sensor use the new panasonic? If you also look for CCDs, you will found mutch more with frame rates up 24fps and global shutter. Next step is found a camera head, or you build your own. This is the next point:
For example: A single IBIS5a Sensor cost round $ 180,-, ($ 100,- if you buy nn) Same thing on micron. 1.3M Pixel CCDs start at $45,-. High speed CMOS (500fps) cost round $1000,-. Altasens min $500,-. Camera heads manufactors calculate round 10 times of sensor cost for the rest. (IBIS camera heads $850-1800, High speed $5000 - 13000,-, Altasens ???? ask Steve N.) But what will you need, if you build your one camera(head)? A camera head to support ALL SENSORS? Not mutch. 60MHz clock is not realy high. But the biggest problem ist the PC Interface. Yes, Interface is the biggest part in a standart camera head. It must buffer pictures, convert to a 8Bit stream and the interface itself.
But why not without PC? Why not write direct to HDD?
There are MCs under $ 100,- with build in LINUX System, on Chip HDD controller for more than 4HDDs, I/O Ports, free programming and direct connectet to ALL Sensors (only change the Ports and sensor programming part). If you write direkt to HDD, you need no buffer, because hdds have build in buffers. All other things, like FPN correction etc. you can do in post.
With those system, you are ready for ALL sensors. You can build a ARRI D20 like camera or a cheap NTSC Video camera. All things are possible. And exact this is the way we will go.

Hurray!!!
at last :)

Thanks Rai

$45, maybe we should do this for a toy camera? So where are Micron 1300 cameras for $450?

The MC with all these features, can you give links, or names?

How much data rate would you need at the camera head:

3 chip 1080p, pixel shifted to 2160p, or 3240p 150mb/s+
Cheap Imax quality.

So 200MB/s at the camera head should do it (no high speed capture). For slow motion effect you will need 300MB/s, which future IDE drives should be able to do on a four drive system. But if you wanted top flight Imax technology on the cheap you would be looking at:

3 chip 2160p+, pixel shifted to 4320p, or 6480p 600mb/s+

I have one guy yesterday quoting me $5K for a 120fps FF camera.

Fillfactory / Cypress news from today:

>>>--February 16, 2005- Cypress Semiconductor Corp. (NYSE: CY) today announced that it has signed and closed a definitive agreement to acquire SMaL Camera Technologies, a designer of industry leading digital imaging solutions for a variety of business and consumer applications, such as ultra-slim digital still cameras, automotive vision systems, and cell phone cameras. Together with FillFactory, Cypress will now offer a complete portfolio of products with high margins, on the one hand, and the potential for extremely high volume, on the other--<<<

Background: Fillfactory create IBIS5 and other CMOS sensors. Last year Cypress has acquisition of FillFactory. Cypress build also some chips for digital video, so i think, they will push some things. Will see...

Yes, now we can see Microns butt kicked all around the stage ;) Hopefully they will acquire Foveon next.

This is the company with the sensors I have mentioned. Thin credit card sized cameras based on their 1.3Mp 27MB/s sensor platform sell for around $69 with USB, with comrpession engine (probably useless anyway), and autobright (advanced per pixel dual slope alternative). Thieir is also 2 and 3 Mp (70MB/s). I have been meaning to get around to investigating the sensors in detail (to measure quality and reduce rolling shutter effect etc). I was planing on looking at one of the cameras based on the platform tommorrow. These are bottom of the line sensors for botom of the line camera projects though.

Thanks Rai

Wayne.

@Wayne,
Last year, Microns MT9M413 CMOS sensor cost round $950,- (single)
It do max. 500fps but at 10 10 bit ports. So you have 100 digital out pins and you need 576 MB/Sec to record 1280x720 10bit 500fps.
Fillfactory LUPA1300 CMOS sensor cost the same. It do max. 450fps.
But high speed is not what i am looking for. So lets go back to the 24fps solution: Lets do a look to a big cinema company. The ARRI D20 (my personaly favorite) use only one CMOS senser, but with 2880x2620pixel. At 24fp 12bit, we need 224MB/sec. 4 high speed HDDs and you got it. But lets go down one step. 1920x1080, 12Bit, 24fps need only 75MB/sec. Thats easy, if you do not must press it down through a PC PCI bus (look at obins thread). You mentioned pixel shift. I see not a reason why you must do it before record. You blow only up your datas. You can do this, or other interpolations in post.

I'm not saying use all the pins/data-paths off the sensor or make a capture unit capable of 500fps. 300MB/s is only 50fps, just enough to slow the footage (for glances kisses etc). So nothing elaborate, though 120fps better. This is the sort of simple thing cinematographers will be interested in.

I am merely pointing to a realistic cheap top end for where sensors will go (notice it is still cheap future 1080p 3 chip) for your interchangeable sensor head, that your customers will be interested in at that time. Whatever ARRI does now doesn't mean they won't do something completely different in future and force people to buy a new camera to get it. When the technology gets sorted out, Arri will probably go to 32MP UHD, or three chip, in flash, if they thought they could market it. The truth is also that digital theatre will eventually lead to 50-100fps frame rates at resolutions upto 32MP UHD, as well. Your's is a commercial product, not like our hobby projects.

Pixel shift is pretty useless after record, as the accuracy (over post interpolation) is obtained by the misalignment of the original image creating real detail boundaries that can be extrapolated.

Thanks

Wayne.


For everybody, here is the full FF article in pdf:
http://www.fillfactory.com/htm/news/docs/SMaL_Press_Release.pdf

<<<-- Originally posted by Wayne Morellini : <<<-- The cheapest price is going to be around $2K (incase you haven't read the threads). I would rather a convenional camera that spat out recordable 12bit 4:4:4 uncompressed images myself for the same price, but I'm not holding my breath. -->>>

I've read the threads, which is why I know Obin's isn't going to be $2K. Now, you could be talking about buying the actual camera from the manufacturer, like SI, but then I still have to come up with a capture and conversion solution, or license someone elses. And that's not talking about lenses or anything yet. Either way, I'm not coming in under $2K. Probably, I think, not under $5K.

That is why you can get cameras for $1000 USB (I said it wasn't pretty), and you can figure the rest from there.

Obin's is only going to be one of many systems. A secondhand f1.2 35mm lens is $50 (or cheap CCTV lens) and make an adaptor. Rob's software (when finished) is supposed to be $100 (I think), and he Sumix system will hopefully have the software in their system, and for their existing $1000 camera. But well see what happens.

Of course the editing system might add some price, but that is not the camera system.

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
John, this is a great find. RAW camera mods for web cams. We now have high speed, descent latitude, HD webcams out there.

I'm sorry I missed this back then, I was getting so overloaded with links I still haven't read hundreds of them.

I have just been looking for hours (now morning) for a link to a Astro site listing all cameras and there performances, I thought posted by Ronald in the Technical thread, but I haven't been able to find it. I mention a list in this thread, but I can't find one in this thread, would anybody here know where it is?

Thanks

Wayne.

Wayne,
You mean something like this page?
http://homepage.ntlworld.com/molyned/web-cameras.htm

Yes, something similar to that but for all sorts of cameras, not just web cams. I tried googling for hours, and lots of useless hits. Do you know some good Astrology sites with a good links list, that you could refer me too, they might have a link?

What I am trying to do is find is sensors/cameras with the most sensitivity and latitude for a uncompressed camera experiment. The cameras have to be at least 426 pixels across, 640 or 720 pixels across preferably. Web cams are far to limited in range. I have found a nice imager, for satellite use, but the truth is that it is probably goign to cost a fortune in a camera, and might not be they best either.

I have been following the links on those pages and have come across some stuff and started another thread on the web-cam mods with the new links:

http://www.dvinfo.net/conf/showthread.php?p=455118#post455118

I did not find what I was looking for (lists of info on all cameras and sensors, not just web-cam or firewire), but I did find some helpful short lists (0.00015 lux cameras and a 200K -ev well capacity sensor) that might help.

Do you know some good Astrology sites with a good links list, that you could refer me too, they might have a link?
Not at this time, but if I found such a site I'll post here... or may be better in your new Web Cam/RAW mods for Astro, Digital Cinema camera use? (http://www.dvinfo.net/conf/showthread.php?p=455118#post455118) thread?

About the high sensitivity camera - DXB-9200EX (http://www.polarisusa.com/cgi-bin/view_product_detail.pl?listing_category_id=&product_category_id=1&template_id=9850) is a good one on paper and theoretically you can replace the sensor in a CCD webcam with this one and use the USB2(IEEE1394) connection already on webcam.
Two examples:
a Sony ICX414 onto the front end of a standard modified webcam (http://www.madpc.net/~greg/icx414.htm)
replacing the ICX098 CCD in the Phillips Toucam with a ICX424 (http://www.pmdo.com/wsc3.htm)

Be aware that the Philips SPC900NC (one of the most expensive webcams today) is a USB1.1 device! VERY BAD because the sensor is quite good - Philips SPC900NC (http://www.home.zonnet.nl/m.m.j.meijer/D_I_Y/spc900nc.htm)
Some e-shop sites claim that this webcam is USB2.0 and capable of 90fps,but when i see the screen shots from the drivers interface, I see 640x480@15fps is the maximum!

Thanks for the links and advice, I will have to leave it for today, as I am not feeling the best.

About those web ads advertising 90 fps, they might have confused a small sub windowing rate with the full frame rate.