New cinema camera coming very soon: SPARTA
- Thread starter Ooze3d
- Start date
DavidNJ
Carbonite Member
Question: has anyone priced the Dalsa head? http://www.dalsa.com/mv/products/cameradetail.aspx?partNumber=PT-41-04M60
F-Stops... We knew that was going to be an important question sooner or later.
Many new cameras announce their specs by giving the basic info from the sensor documentation. Some say they offer 13-14 F-Stops and once they get released, independent tests discover those 13 stops are actually 7-8. Our sensor gives 13 stops according to its specs. We're announcing 9 stops because, according to our initial tests, we think that'll be closer to the real specs of the finished camera. We also hope we're wrong and the camera ends up having 11 stops. Besides that, CCDs are usually more efficient in terms of detail and sensitivity compared to CMOS.
Many new cameras announce their specs by giving the basic info from the sensor documentation. Some say they offer 13-14 F-Stops and once they get released, independent tests discover those 13 stops are actually 7-8. Our sensor gives 13 stops according to its specs. We're announcing 9 stops because, according to our initial tests, we think that'll be closer to the real specs of the finished camera. We also hope we're wrong and the camera ends up having 11 stops. Besides that, CCDs are usually more efficient in terms of detail and sensitivity compared to CMOS.
Otis Grapsas
Carbonite Member
That's not very precise. Kodak do mention 64dB which is good in theory. Dividing 20,000 saturation electrons / 12 noise electrons is 1666 which is about 64dB but there are other sources of noise in a real implementation. Are the electronics in any camera perfect without electromagnetic interference? How hot is your chip in a low power passive cooling situation and how does this impact noise performance?
13stops is an insanely high figure of course and I know no chip that will get it without extreme implementation measures which will never happen in these designs (such as very slow CCD readout and extreme active cooling combined with a extremely low noise CCD in the first place). Kodak never mentioned 13stops, and even the manufacturers of the chips that can do it, never mention it because it's not possible in typical normally cooled high frame rate CCD designs.
A stops measurement using noise floor means nothing really. What is the dynamic range of film using this method? Even the measurement using commercial apps does not mean much because it depends on the environment (shutter speed, IR filtering, temperature, lighting temperature, color correction decisions etc). What is important is the visible dynamic range in a given situation and a photographer will easily notice the good and the bad cameras. This starts with a sensor but mostly depends on the implementation, the highlight protection and the shadow handling through gamma curves. This Kodak is certainly capable of impressive dynamic range compared to sensors with similar or smaller pixels, and that includes many full frame sensors and dSLRs.
In terms of noise, an uncompressed camera can have many times higher noise without it being distructing because there are no compression artifacts or blocking. Film scans are like that also and their noise levels are very high. The compresion codecs themselves can be measured for their impact on low level signal quality.
When you mention 8 stops in the pdf, you are making the chip look bad when it's performing excellently compared to chips that claim 11 stops and typically surpasses them by a long margin, especially in low light. The users are rarely engineers, and those who try to be usually misinform the public without even knowing it. If they would stick to subjective dynamic range comparisons, using cameras side by side and exposing for highlights, it would be better. They wouldn't have perceptions about numbers, but a meaningful actual comparison in real world scenes. It's all a photographer needs.
Here is a pdf taking the scientific approach to camera performance. Good read, it shows a lot about the implementation, but it doesn't show much about how it will perform in photography, after the video camera processing is added. Especially when the cameras we want to compare witll get different processing being from different manufacturers all having developed their own intellectual property, not to mention aesthetics:
http://www.baslerweb.com/downloads/19725/Basler_scA1600-14fm_EMVA_Standard_1288.pdf
EDIT: Typos, link added.
13stops is an insanely high figure of course and I know no chip that will get it without extreme implementation measures which will never happen in these designs (such as very slow CCD readout and extreme active cooling combined with a extremely low noise CCD in the first place). Kodak never mentioned 13stops, and even the manufacturers of the chips that can do it, never mention it because it's not possible in typical normally cooled high frame rate CCD designs.
A stops measurement using noise floor means nothing really. What is the dynamic range of film using this method? Even the measurement using commercial apps does not mean much because it depends on the environment (shutter speed, IR filtering, temperature, lighting temperature, color correction decisions etc). What is important is the visible dynamic range in a given situation and a photographer will easily notice the good and the bad cameras. This starts with a sensor but mostly depends on the implementation, the highlight protection and the shadow handling through gamma curves. This Kodak is certainly capable of impressive dynamic range compared to sensors with similar or smaller pixels, and that includes many full frame sensors and dSLRs.
In terms of noise, an uncompressed camera can have many times higher noise without it being distructing because there are no compression artifacts or blocking. Film scans are like that also and their noise levels are very high. The compresion codecs themselves can be measured for their impact on low level signal quality.
When you mention 8 stops in the pdf, you are making the chip look bad when it's performing excellently compared to chips that claim 11 stops and typically surpasses them by a long margin, especially in low light. The users are rarely engineers, and those who try to be usually misinform the public without even knowing it. If they would stick to subjective dynamic range comparisons, using cameras side by side and exposing for highlights, it would be better. They wouldn't have perceptions about numbers, but a meaningful actual comparison in real world scenes. It's all a photographer needs.
Here is a pdf taking the scientific approach to camera performance. Good read, it shows a lot about the implementation, but it doesn't show much about how it will perform in photography, after the video camera processing is added. Especially when the cameras we want to compare witll get different processing being from different manufacturers all having developed their own intellectual property, not to mention aesthetics:
http://www.baslerweb.com/downloads/19725/Basler_scA1600-14fm_EMVA_Standard_1288.pdf
EDIT: Typos, link added.
Last edited:
Chamber005
Carbonite Member
LOL -- Just read through this thread. One of the best ever. Totally enjoyed with my morning cup of tea. And isn't it weird, the difference between a free market and whatever's going on in Europe? They seriously force manufacturers to raise their prices?! That's insane. If I want to sell a camera for cost because I'm hoping that you'll buy my accessories, in America, that's my business. Live free or Die Hard, ya heard?
All of the hoopla about these cameras, though, (even including the Scarlet) is pretty much like talking about a Justice League movie, IMO. I mean, sure, eventually there SHOULD be one, but who knows? And who knows if it will be any good, right?
Advertising and promotion is fine. But let's be real, until there is some real world application of any of these "specs" or "prototypes", it's all just nonsense. Evidence and proof, baby -- that's all we care about.
All of the hoopla about these cameras, though, (even including the Scarlet) is pretty much like talking about a Justice League movie, IMO. I mean, sure, eventually there SHOULD be one, but who knows? And who knows if it will be any good, right?
Advertising and promotion is fine. But let's be real, until there is some real world application of any of these "specs" or "prototypes", it's all just nonsense. Evidence and proof, baby -- that's all we care about.
DavidNJ
Carbonite Member
"That's not very precise. Kodak do mention 64dB which is good in theory. Dividing 20,000 saturation electrons / 12 noise electrons is 1666 which is about 64dB but there are other sources of noise in a real implementation. Are the electronics in any camera perfect without electromagnetic interference? How hot is your chip in a low power passive cooling situation and how does this impact noise performance?"
I imagine of the photo cell is large enough the signal to noise ratio could be high enough to increase the dynamic range. However, that really isn't necessary.
If you have enough photo sites and expose them differently you can obtain the ability to increase the dynamic range. Fuji did this with their S sensor pairing large and small photo sites. They do it electronically with their EXR sensor now.
I imagine of the photo cell is large enough the signal to noise ratio could be high enough to increase the dynamic range. However, that really isn't necessary.
If you have enough photo sites and expose them differently you can obtain the ability to increase the dynamic range. Fuji did this with their S sensor pairing large and small photo sites. They do it electronically with their EXR sensor now.
Otis Grapsas
Carbonite Member
That's interesting. How large is the frame size compared to 2/3"? 2/3" for 16:9 is 9.6x5.4 or 11mm diagonal.
Tom Lowe
Well-known member
i'm not the one starting a camera company. i'm the guy telling you what shooters want. is it not possible to find such a sensor? 2/3 is going the way of the dinosaur. S35 and FF35 are the way of the future. you guys don't want to be shooting at where the target was a year ago, you need to lead the target and aim at where it will be next year.
Nik Manning
Well-known member
I think 2/3 size chip is great. You can't find any cameras list price under 10k with that size sensor. Cheapest is HPX500 and that has a SD size chip. So your cam should be at the top for 2 years. Then 2 years from now you can make it bigger.
DavidNJ
Carbonite Member
I don't think everyone appreciates these aren't sensors designed for cinema. More likely, they are designed for automation, robot vision, quality inspection, etc. It just happens that the demands for that are now overlapping high end video needs.
All of these parts are made with cost constraints; robots don't need bokeh, they need quite the opposite...to always be in focus. Quality inspection is usually done under good lighting. So the advantages of huge sensors are probably not that critical.
All of these parts are made with cost constraints; robots don't need bokeh, they need quite the opposite...to always be in focus. Quality inspection is usually done under good lighting. So the advantages of huge sensors are probably not that critical.
filmmaker's gang
Carbonite Member
:thumbup:
Otis Grapsas
Carbonite Member
CCD is not a low cost choice. It's CMOS that is used from consumer to prosumer and CCD above that. How many cameras are 1920x1080 2/3" CCD using I-frame formats (not to mention uncompressed) and how much do they cost? Check the Sony and Panasonic range. In the near future, 2/3" 1080p CCD will only come with 50,000 euro cameras requiring very expensive lenses.
The 2/3" sensors have many advantages in actual production, especially in low cost projects that are the ones requiring 5,000 and 10,000 euro cameras. Everything by Sony and Panasonic is only up to 2/3" 1080p CCD except a camera from Sony at the astronomic cost of 400,000 or something like that for a complete package. And even that camera is only 1080p. Are all these guys wrong about sensing what the market needs or do they lack full frame technology and knowhow? Their idea of what the market needs is not full frame sensors, it's obvioulsly 2/3" 1080p CCDs with an I-frame format.
A 16:9 2/3" 1920x1080 sensor is not designed for automation. 4:3 makes a lot more sense for that and 1:1 is usually preferable. The DOF of 2/3" is also too shallow for sensing a 3d environment and adequate DOF in these environments requires use of strobes. 2/3" would be ok for sensing something that is completely flat but it still wouldn't be as good as smaller sensors.
Low lighting requires smaller sensors, it's not the other way around. If there is enough light, you can stop down all you want (until you start getting blur). It's the pixel size that is an indication of low light performance, not the sensor size.
The 2/3" sensors have many advantages in actual production, especially in low cost projects that are the ones requiring 5,000 and 10,000 euro cameras. Everything by Sony and Panasonic is only up to 2/3" 1080p CCD except a camera from Sony at the astronomic cost of 400,000 or something like that for a complete package. And even that camera is only 1080p. Are all these guys wrong about sensing what the market needs or do they lack full frame technology and knowhow? Their idea of what the market needs is not full frame sensors, it's obvioulsly 2/3" 1080p CCDs with an I-frame format.
A 16:9 2/3" 1920x1080 sensor is not designed for automation. 4:3 makes a lot more sense for that and 1:1 is usually preferable. The DOF of 2/3" is also too shallow for sensing a 3d environment and adequate DOF in these environments requires use of strobes. 2/3" would be ok for sensing something that is completely flat but it still wouldn't be as good as smaller sensors.
Low lighting requires smaller sensors, it's not the other way around. If there is enough light, you can stop down all you want (until you start getting blur). It's the pixel size that is an indication of low light performance, not the sensor size.
lawriejaffa
Veteran
Yep id agree with that Otis.
That is why i dont quite buy it that these huge camera manufacturers just so blithely missed the boat on the supposed revolution...
I mean we'll wait and see... but when you have dudes scrambling everything they can get to produce a micro at 4k (or whatever it really is) while top digitally shot movies are shot at 1080p...
OH OH and i nearly forgot,
This is probably insane Otis, and for Oooze- but have you guys considered customised castings and visual aesthetics for the camera's designs. A lot of chaps that are into PC customisation and so on - may well be the same breed that are into your own wares.
It would be great, for metal / leather features - or cool radical colours hehe. I know that might sound a little crazy - but worth a little focus grouping you know?
That is why i dont quite buy it that these huge camera manufacturers just so blithely missed the boat on the supposed revolution...
I mean we'll wait and see... but when you have dudes scrambling everything they can get to produce a micro at 4k (or whatever it really is) while top digitally shot movies are shot at 1080p...
OH OH and i nearly forgot,
This is probably insane Otis, and for Oooze- but have you guys considered customised castings and visual aesthetics for the camera's designs. A lot of chaps that are into PC customisation and so on - may well be the same breed that are into your own wares.
It would be great, for metal / leather features - or cool radical colours hehe. I know that might sound a little crazy - but worth a little focus grouping you know?
It's an interesting moment to say that when we have a 8 oscar movie (including cinematography) shot with one of those 2/3" robot vision sensors you talk about. I don't know if you've ever shot with 2/3". I have and I can tell you the lack of shallow DOF is not a problem. Specially if you're shooting with F1.4-2.8.
craigbowman
Veteran
There is always room in the marketplace for innovative new products. Entrepreneurs make the world go 'round!