Fliques Osman
Well-known member
Amen.
A-cam dII 'High Definition RAW Format Motion Picture Camera
- Thread starter seejay1031
- Start date
Amen.
Otis Grapsas
Carbonite Member
Some information that might help someone looking into one of these cameras or any new ccd camera.
1) Antismear performance. CCD electronic shutter antismear varies up to 60dBs from one CCD to another (60dB better means 1000x more light to produce the same smear problem). One ccd will only smear with head on car headlight, another camera will smear with anything, including a reflection of a small incadescent in a glass, normal incadescent home light fixtures, street lighting, light coming in from a window etc. This is something that must be shown in actual image samples. Many huge CCDs are inferior to consumer 1/6" 3CCD cameras in antismear performance, so aceptable performance must be demonstrated.
2) CCD performance is extremely sensitive to frame rate. Going from 25fps to 50 or 60fps can mean about 18dBs of difference in noise (and dynamic range) on the same CCD just because the clock is faster, especially at HD resolution. This can make a 11stop 25fps camera produce a 10stop 30fps result and a 6 or 7 stop 60fps result. This must be shown in actual images or specified with a s/n number. Most scientific imaging application designs go for the slowest frame rate that is acceptable, because the noise advantage is huge. From my own CCD designs, 25fps model vs 38 fps model = 8dB of difference in performance which is a lot in cinematography.
3) Regarding DNG, do a test and find how many megapixels (images of known size) your computer can debayer per minute using dslr bayer files. This will be a useful indication of workflow performance with a DNG video camera of a given resolution. Just devide with the data rate of the DG video camera and see if you can tolerate the performance.
4) Uncompressed editing is not a very big problem. Once the disk data rate is provided, it's very fast because there is no complex decoding required. The 1.5TB disks we have now start at 120mb/sec at their fastest point so that will not be an issue. A 5disk raid can edit full resolution online if that is required. It usually is not because the material is normally cut in smaller resolutions and its duration becomes much smaller before it reaches the online stage.
5) Verify the actual sensor/DNG file resolution.
If antismear and noise performance is acceptable this camera is an extremely good purchase.
1) Antismear performance. CCD electronic shutter antismear varies up to 60dBs from one CCD to another (60dB better means 1000x more light to produce the same smear problem). One ccd will only smear with head on car headlight, another camera will smear with anything, including a reflection of a small incadescent in a glass, normal incadescent home light fixtures, street lighting, light coming in from a window etc. This is something that must be shown in actual image samples. Many huge CCDs are inferior to consumer 1/6" 3CCD cameras in antismear performance, so aceptable performance must be demonstrated.
2) CCD performance is extremely sensitive to frame rate. Going from 25fps to 50 or 60fps can mean about 18dBs of difference in noise (and dynamic range) on the same CCD just because the clock is faster, especially at HD resolution. This can make a 11stop 25fps camera produce a 10stop 30fps result and a 6 or 7 stop 60fps result. This must be shown in actual images or specified with a s/n number. Most scientific imaging application designs go for the slowest frame rate that is acceptable, because the noise advantage is huge. From my own CCD designs, 25fps model vs 38 fps model = 8dB of difference in performance which is a lot in cinematography.
3) Regarding DNG, do a test and find how many megapixels (images of known size) your computer can debayer per minute using dslr bayer files. This will be a useful indication of workflow performance with a DNG video camera of a given resolution. Just devide with the data rate of the DG video camera and see if you can tolerate the performance.
4) Uncompressed editing is not a very big problem. Once the disk data rate is provided, it's very fast because there is no complex decoding required. The 1.5TB disks we have now start at 120mb/sec at their fastest point so that will not be an issue. A 5disk raid can edit full resolution online if that is required. It usually is not because the material is normally cut in smaller resolutions and its duration becomes much smaller before it reaches the online stage.
5) Verify the actual sensor/DNG file resolution.
If antismear and noise performance is acceptable this camera is an extremely good purchase.
Hunter Hampton
Carbonite Member
This seems to be the kodak sensor here (or at least it fits the specs and looks like the same ccd in the d2 mock-up):
"The KODAK KAI-02150 Image Sensor is a 1920(H) x 1080(V) resolution, 2/3" optical format, progressive scan interline CCD. A flexible readout architecture is used that enables the use of either 1, 2 or 4 outputs to achieve frame rates up to 64 fps. The vertical overflow drain structure provides antiblooming protection and enables electronic shuttering for precise exposure control. Other features include low dark current, negligible lag and low smear."
Dr. Otis, Do those specs look good?
Link: http://www.kodak.com/global/en/busi...erline/KAI-02150/overview.jhtml?pq-path=13614
"The KODAK KAI-02150 Image Sensor is a 1920(H) x 1080(V) resolution, 2/3" optical format, progressive scan interline CCD. A flexible readout architecture is used that enables the use of either 1, 2 or 4 outputs to achieve frame rates up to 64 fps. The vertical overflow drain structure provides antiblooming protection and enables electronic shuttering for precise exposure control. Other features include low dark current, negligible lag and low smear."
Dr. Otis, Do those specs look good?
Link: http://www.kodak.com/global/en/busi...erline/KAI-02150/overview.jhtml?pq-path=13614
Otis Grapsas
Carbonite Member
Oh, I'm familiar with that one. It is actually 2/3". A lot smaller than super16.
The sensor is as good as the larger 1" 1920x1080p and 1.2" 2048x2048.
Antismear is improved over the previous models (used by NoX and our camera prototype). Not as good as the Sony 2/3" we use or the 2/3" camcorders, but better than some 1/3", so it's not a limiting factor. The antismear improvement is the real advantage over the other Kodaks used by NoX on their camera and available as an option on our camera project.
Noise looks ok. As good as the larger Kodaks even with the smaller pixel size. It will need 2x the light to to get full saturation due to smaller pixel size though which negates the noise advantage. By comparison the 2/3" sony is about 2 stops better.
Dynamic range is about 10 stops. Not as good as Sony 2/3" I am using (11+) but still good. I can't be sure about noise before we see a full specsheet though with the clocking frequencies the specs represents.
The sensor uses multiple outputs so it could have split screen phenomena in some cases depending on temperature and implementation. HD100 uses a similar technique and the other Kodaks also use more than one output.
One problem I see is that they do not write/provide software, so they cannot implement their own gamma curves and overexposure protection schemes. The user will have to rely on third party raw software. As an example, the Sony cinealtas make the best of their custom gamma curves and no RAW bayer software comes close. It also becomes a problem on extreme dynamic range where geting the best from the sensor requires an accurate preview of the final dynamic result. Something the manufacturer can control and predict.
I have to say that the price of this camera is fantastic for a CCD implementation. They take advantage of simplicity. I believe in modularity though. Implementing new sensor options as they appear, more than one camera heads at lower cost, increasing audio capability, adding and using new storage options and interfaces as they appear etc. When I started the camera project only two sensors were available that could be used for cinema. Now we have this new Kodak sensor and I just received an email from a manufacturer on a global shutter super35mm CMOS that aproaches release. On a modular design you can add new sensor options easily, on a fixed design you need a new camera. These days, solutions get old very fast (storage, sensors, interfaces etc) and when we are talking 10k to 30k euro, some upgradability is nice to have, especially for low budget filmmaking, where the gigs of the month cannot pay for a new camera.
The sensor is as good as the larger 1" 1920x1080p and 1.2" 2048x2048.
Antismear is improved over the previous models (used by NoX and our camera prototype). Not as good as the Sony 2/3" we use or the 2/3" camcorders, but better than some 1/3", so it's not a limiting factor. The antismear improvement is the real advantage over the other Kodaks used by NoX on their camera and available as an option on our camera project.
Noise looks ok. As good as the larger Kodaks even with the smaller pixel size. It will need 2x the light to to get full saturation due to smaller pixel size though which negates the noise advantage. By comparison the 2/3" sony is about 2 stops better.
Dynamic range is about 10 stops. Not as good as Sony 2/3" I am using (11+) but still good. I can't be sure about noise before we see a full specsheet though with the clocking frequencies the specs represents.
The sensor uses multiple outputs so it could have split screen phenomena in some cases depending on temperature and implementation. HD100 uses a similar technique and the other Kodaks also use more than one output.
One problem I see is that they do not write/provide software, so they cannot implement their own gamma curves and overexposure protection schemes. The user will have to rely on third party raw software. As an example, the Sony cinealtas make the best of their custom gamma curves and no RAW bayer software comes close. It also becomes a problem on extreme dynamic range where geting the best from the sensor requires an accurate preview of the final dynamic result. Something the manufacturer can control and predict.
I have to say that the price of this camera is fantastic for a CCD implementation. They take advantage of simplicity. I believe in modularity though. Implementing new sensor options as they appear, more than one camera heads at lower cost, increasing audio capability, adding and using new storage options and interfaces as they appear etc. When I started the camera project only two sensors were available that could be used for cinema. Now we have this new Kodak sensor and I just received an email from a manufacturer on a global shutter super35mm CMOS that aproaches release. On a modular design you can add new sensor options easily, on a fixed design you need a new camera. These days, solutions get old very fast (storage, sensors, interfaces etc) and when we are talking 10k to 30k euro, some upgradability is nice to have, especially for low budget filmmaking, where the gigs of the month cannot pay for a new camera.
lars steenhoff
Active member
http://www.dalsa.com/markets/ccd_vs_cmos.asp
This link provides an interesting read about the differences between CCD and CMOS.
Both have their advantages depending on the application.
This link provides an interesting read about the differences between CCD and CMOS.
Both have their advantages depending on the application.
Hello
I have been looking at the DII camera with great interest since I have some 16mm cine-nikkors c mount lenses that surely can perform really well on it. I was wondering if the best option is to get the c-mount version of the DII or if I might be better off getting the ims version which, besided giving more freedom of choice for the lenses, is also more robust than the c-mount thread.
Andreas
I have been looking at the DII camera with great interest since I have some 16mm cine-nikkors c mount lenses that surely can perform really well on it. I was wondering if the best option is to get the c-mount version of the DII or if I might be better off getting the ims version which, besided giving more freedom of choice for the lenses, is also more robust than the c-mount thread.
Andreas
lars steenhoff
Active member
IMS seems the best option to me, if you look for using many different lens brands.
it will also be the most expensive and slightly bigger/ heavier than the c-mount.
Preliminary conclusion: if you use strictly c-mount lenses, the c-mount option saves some money and makes the whole camera a little more portable.
I don't know about robustness, very interesting questions ..
it will also be the most expensive and slightly bigger/ heavier than the c-mount.
Preliminary conclusion: if you use strictly c-mount lenses, the c-mount option saves some money and makes the whole camera a little more portable.
I don't know about robustness, very interesting questions ..
Last edited:
Hunter Hampton
Carbonite Member
I bet they will end-up just doing IMS only when it comes down to it, theres no reason to have all of the options for the camera body when you can have IMS and get anything you want.
Hunter Hampton
Carbonite Member
Otis-
Thanks for the comments. I wonder if that really is the sensor?
About image processing, I bet a dedicated Motion DNG "Lightroom" type program will come out with Lut support from adobe sometime, if not someone else will develop it, the technology is already there it just needs to be adapted.
Also, I bet in the next few years the motion DNG standard will become the standard for high-end electronic cinematography and replace compression all together, especially with how cheap storage is getting.
So after reading that cmos to ccd comparison, and finding out that all ccds spit out analog in voltage... is that one of the reasons that ccd cameras seem to have a bit more organic look to them? Especially in the noise/grain structure? (this is just what I have noticed)
Thanks for the comments. I wonder if that really is the sensor?
About image processing, I bet a dedicated Motion DNG "Lightroom" type program will come out with Lut support from adobe sometime, if not someone else will develop it, the technology is already there it just needs to be adapted.
Also, I bet in the next few years the motion DNG standard will become the standard for high-end electronic cinematography and replace compression all together, especially with how cheap storage is getting.
So after reading that cmos to ccd comparison, and finding out that all ccds spit out analog in voltage... is that one of the reasons that ccd cameras seem to have a bit more organic look to them? Especially in the noise/grain structure? (this is just what I have noticed)
I guess you are right Hunter. Probably ims would also be a sturdier mount compared to c-mount. It did happen to me, while focusing a lens with a stiff ring, that the lens actually unscrewed from the camera.
Does anybody have any idea of how expensive it is to get the various adapters for the individual lenses, so that all fit in the ims mount?
Andreas
Hunter Hampton
Carbonite Member
Most the adapters for the IMS system range from around $300-600. I think you can get a price list from P+S technik.
lars steenhoff
Active member
I have a few nikon lenses i'd love to try out on the a-cam dII.
Who knows how much of the image will be cropped on this sensor compared to 35 mm?
Who knows how much of the image will be cropped on this sensor compared to 35 mm?
Last edited:
Hunter Hampton
Carbonite Member
I was thinking about trying out ENG b4 mount lenses. Also using a nikon 300mm f2.8 sounds like fun for telephoto.
Otis Grapsas
Carbonite Member
The horizontal crop factor vs full frame still cameras is 36mm/10.6mm=3.4. A 50mm f1.4 is equivalent to 170mm f4.76 for imaging and retains f1.4 ight gathering of course.
augenblick
Well-known member
The diagonal of the Ikonoskop 2/3 inch sensor (10.6 x 6mm) is 12.18 mm.
A square plus B square equals C square
Peter
A square plus B square equals C square
Peter
I found a useful link, which I hope is going to clarify sensor size
http://www.dpreview.com/news/0210/02100402sensorsizes.asp
http://www.dpreview.com/news/0210/02100402sensorsizes.asp