Sony NEX-FS100 with Super 35mm CMOS image sensor

I have wondered if someone has tried using instead of another green, some other color like yellow? I imagine the advantage might not outweigh due to our sensitivity to green. Pertaining to the lens, yes, optically identical, but I think the mechanics do have value. I think a lot of people try to tell people that for some, it is very important. I personally would not buy a cine lens because I don't like the weight. That said, if I were pulling focus all the time, or even Iris, then I'd SERIOUSLY be considering one. Having used that lens he showed, mechanically, it is AMAZING.
 
headroom said:
... its the COLOR LOCKUP TABLE dude, not the Filter array on the Sensor.
Click to expand...

Right, but the problem is if all the colors are being derived from RGB, and then displayed in RGB, then what exactly is this 3D LUT providing? That is the question. I am not saying it is good or bad. But saying it creates colors that you cannot get with RGB makes no sense. What really matters is how many nodes and how good the 3D LUT actually is. I think what he was trying to say is that it preserves your gamma no matter which 3D LUT you use, but that's not how it came out :)
 
Or to put it another way, saying a camera with an RGB sensor is capable of capturing more colors because of the LUT sounds about like saying a camera with an 8-bit ADC is capable of capturing 10-bit because it has an HD-SDI port.
 
Lee Saxon said:
Or to put it another way, saying a camera with an RGB sensor is capable of capturing more colors because of the LUT sounds about like saying a camera with an 8-bit ADC is capable of capturing 10-bit because it has an HD-SDI port.
Click to expand...

Haha, yes. What I think he also might have trying to be saying is that by using a 3D LUT with enough nodes, when applying the LUT, you can get a better color transform and not get quantization in the interpolation. (This is what I have seen people saying a 3D LUT can be good for). Not to say a 1D LUT cannot do this if it is big enough. It seems the trade-off is that a 1D LUT requires a lot of space to do 1:1 mapping, and if you use a smaller 1D LUT table, you require interpolation where you aren't getting an actual curve but piecewise linear segments. (Unless they actually use something like bezier approximations but I doubt that). With a 3D LUT, you can get the same performance as a large 1D LUT but you require more processing power to generate the color transforms.

Which, in summary, is nothing like he said in the video :P
 
3D LTU is much like distributor advance curves in cars.
First they had linear, than progressive and now they have a 3D map, that is much more flexible and precise.
Still only makes the same old simple spark at the right time, that exact moment makes a lot difference though.

Frank
 
Any chance the FS100 could be hacked to deliver 10- or 12-bit HDMI out, or failing that, modified to output RAW, like the Andromeda and Hydra mods for the Panasonic DVX and HVX?
 
Haha, Tzehekh, if I had a dollar for everytime someone asks that question. :) I'm going to start making up camera names and asking that question.

"Do you think the Sony FS250-1 is hackable for 14-bit 4:4:4 4D?"
 
I agree that 3D LUT is not marketing hype in and of itself, as I've been to that site before. What I don't agree with is that it can create colors that you cannot get with a 1D LUT. It all depends on how the mapping is done. If your starting point and ending point is RGB, what it deals with is what type of freedom you have in using LUT matrices to alter your image data. It's like he read a few key words and then just added them in :) I think the biggest difference is the fact that single value triples actually affect ALL mapping to the newer color, and not just within the single value of the 1D LUT. Also, it depends on how many nodes and how the interpolation is handled as to how well it will do. Hopefully not piecewise linear
 
all lot of :-) joking check page 20

"The Color LUT is 3 Dimensional and thats a good thing. More insights from Juan Martinez (Sony Product Manager) shows the new Sony NEX-FS100 camera. He also talks a bit about the upcoming E-mount lenses."

http://www.sonyalpharumors.com/sony-...ntation-video/

Juan Martines and hypercardoid microphone of axis... Dust on the Sensor, Cleaning the same like NEX 3/5/10 ?

http://vimeo.com/22462516

listen at 4:50 HDMI OUTPUT Analog Component 4:2:2 an RGB 4:4:4 8bit !
 
Last edited:
Re:

Re:

headroom said:
all lot of :-) joking check page 20

"The Color LUT is 3 Dimensional and thats a good thing. More insights from Juan Martinez (Sony Product Manager) shows the new Sony NEX-FS100 camera. He also talks a bit about the upcoming E-mount lenses."

http://www.sonyalpharumors.com/sony-...ntation-video/

Juan Martines and hypercardoid microphone of axis... Dust on the Sensor, Cleaning the same like NEX 3/5/10 ?

http://vimeo.com/22462516

listen at 4:50 HDMI OUTPUT Analog Component 4:2:2 an RGB 4:4:4 8bit !
Click to expand...

What does Juan mean when he says "Analog Component 4:2:2 OR RGB 4:4:4". Notice he said "OR". In logic "OR" can mean either/or (not both) or both. Is he implying that we can CHOOSE which colour sample to output? Or he is saying that both occur sequentially after the other? Or is he saying that the output format DEPENDS on the external recording device?
 
shorty15 said:
Oh my god, why is it so hard to find out if the HDMI is 8 bit or 10 bit? Did Sony say anywhere in their offocial product literature? This is really annoying me, why is it so hard to figure these things out?
Click to expand...

I believe ALL HDMI cables are capable of handling up to 12-bit color. That is because (and I'm guessing here) any cable's capabilities is calculated based on the aggregate available bandwidth. What makes it work here is the agreed upon convention used to transmit 12-bit colour with the assumption that your maximum resolution is 1920 x 1080. I think (and again I am guessing) the only thing that needs 'support' are things like DRM such as HDCP since they don't want any of the HDCP to be hacked along the way thourghout the cable.

Having said that, HDMI cables today are 'rated' between 10.2 Gbps (v1.3) and 15 Gbps (v1.4). The former should be plenty to support the analogue 10-bit 4:2:2 color sampling, however I am still guessing at this part.

Anyone know how to calculate the bandwidth throughput of color samples (i.e. 4:2:2 vs 4:2:0)?
 
You must log in or register to reply here.
Back
Top