does CO work directly with sensor data?

Permanently deleted user

• 2017年08月21日 15:41

to the CO engineers... just out of curiosity.

with all the amazing sensitive adjustments that CO can make to our images, I'm wondering if CO works directly with the raw sensor data for much of the 'processing' and touch-up. if all the bayesian R/G/B dots in the camera were combined into RGB values before processing the image, the effective resolution would be cut by 2/3 or more, making each dot 3 times the size of the original dots. the precision of trying to skew/rotate/correct-aberration on an image would be limited by this.

if CO is working directly with individual sensor single-colour values for much of the process, that would be very complicated! it would also explain why CO takes a little while to support new raw formats based on new sensor arrays.

very impressive!

cheers,
Gregory

0

• 

  John Doe

  • 2017年08月22日 00:20

  Well yes, of course CO works from the RAW data. That's the whole point of a RAW converter, right? 😊
  
  What other data do you think it could work from when you feed it a RAW file?

  0
• 

  Permanently deleted user

  • 2017年08月22日 03:14

  😁 my original (wrong) reasoning was that each of the sets of RGB dots were combined into a single RGB value before any processing was done to it, similar to scanning and descreening a newspaper or magazine.

  0
• 

  PhaseoneUser55657

  • 2017年08月26日 12:21

  Gregory,
  You should do a little research on how Baye filters sensors work. No when you convert to RGB you do no lose â…” the resolution. Think of graph paper, the open areas are the photo sites, where the lines intercept is where it calculates the individual full RGB values.
  
  
R G B G R G...
+ + + + +
G B G R G B....
+ + + + +
R G B G R G...
.......

  
  So each pixel is made of 1 Red, 1 Blue and 2 Green sites. (the '+') So if the sensor has 1000X1000 photo sites the system will produce a an image 998X998. Now Fuji X-Trans sensor uses a different color pattern, and a different number of pixels are used to compute the RGB value.
  
  Some sensors are now removing the anti-alias filter. What this filter did was defocus the image slightly so keep sharp lines from only hitting a portion of the RGBG area. With out it if you had a sharp line that happen to fall on only 2 of the 4 sites that made up the RGB value, the color would not be as expected.
  
  Now, on raw converters, they all convert down to RGB values at some point. As that is the only way to actually display what the image looks like. Where the adjustments come in to the process flow is up to the software. It can be done on the raw data or after conversion to RGB. (PhaseOne would have to give up that secret.) My guess, is they use a combination, when you are not at 100% they use the RGB values from the Previews that are generated, but at 100% or exporting they use the RAW data.
  
  
  So a little bit about sensors and Raw and RGB, that probably no body cares about.
  
  Robert

  0
• 

  Permanently deleted user

  • 2017年08月26日 16:00

  thank you!

  0

投稿コメントは受け付けていません。