douglasf13 wrote:I\'m just not convinced the tonality will be a huge upgrade. Everything I read about these b&w sensors is in regards to detail, not being more b&w film-like in terms of tonality. It\'s still a digital sensor, after all.
+1 @ more detail
As to tonal range ... a total elimination of the CFA would render all pixels capable of recording the same amount of light energy, but wouldn\'t the range of a pixel\'s response to the light energy still be limited to the individual pixel\'s capabilties. In that regard, it would seem that the DR wouldn\'t be any greater ... but the transitions within that DR would have more accurate/precise increments than those interpoplated from from the RGB CFA (with different energy response @ R vs. G vs. B). To me, this is where the gain would be.
Even though we can currently convert the color to B&W, and play with it greatly in PP, the initial values are a guess/reverse engineering of the light that has passed through the CFA. Without the CFA ... it is no longer a reverse engineering effort, but rather a direct response. To me, that would render the accuracy/precision greater from the onset, leading to Deming 101 principles.
Here\'s a twist that might be interesting. Given the use of the RGB CFA approach ... consider if you were to use a variable-NDFA, somewhat similar to Fuji using the multi-size pixel approach in their super-CCD, or the variable-sized grain structure of some films, that yielded greater DR than their contemporaries. In doing so, you might be returning to a semi-interpretive approach, but it could offer some ability to handle those scenes that exceed the pixels DR for protecting highlights.
If the non-NDFA pixels are not \"clipped\", then the amount of ND could be offset accordingly via the algorithm. If the non-NDFA pixels are clipped, then the algorthm could utilize the NDFA pixels \"as is\". A profile could easily be used for on/off in simple form, or even better programmable to those pixels @ varying ND value.
It\'ll be interesting to see how much DR can be improved without a CFA (assuming it will not be replaced with an NDFA). My expectation is the improvement in tonality \"within\" the DR more so than the expansion of the DR. Of course, both would be nice.
Now, I just need to get the patent for my variable-NDFA finished, so I can start collecting my royalties. Of course, I have to finish working on my tilt-shift sensor patent first.
douglasf13 wrote:I\'m just not convinced the tonality will be a huge upgrade. Everything I read about these b&w sensors is in regards to detail, not being more b&w film-like in terms of tonality. It\'s still a digital sensor, after all.
+1 @ more detail
As to tonal range ... a total elimination of the CFA would render all pixels capable of recording the same amount of light energy, but wouldn\'t the range of a pixel\'s response to the light energy still be limited to the individual pixel\'s capabilties. In that regard, it would seem that the DR wouldn\'t be any greater ... but the transitions within that DR would have more accurate/precise increments than those interpoplated from from the RGB CFA (with different energy response @ R vs. G vs. B). To me, this is where the gain would be.
Even though we can currently convert the color to B&W, and play with it greatly in PP, the initial values are a guess/reverse engineering of the light that has passed through the CFA. Without the CFA ... it is no longer a reverse engineering effort, but rather a direct response. To me, that would render the accuracy/precision greater from the onset, leading to Deming 101 principles.
Here\'s a twist that might be interesting. Given the use of the RGB CFA approach ... consider if you were to use a variable-NDFA, somewhat similar to Fuji using the multi-size pixel approach in their super-CCD, or the variable-sized grain structure of some films, that yielded greater DR than their contemporaries. In doing so, you might be returning to a semi-interpretive approach, but it could offer some ability to handle those scenes that exceed the pixels DR for protecting highlights.
If the non-NDFA pixels are not \"clipped\", then the amount of ND could be offset accordingly via the algorithm. If the non-NDFA pixels are clipped, then the algorthm could utilize the NDFA pixels \"as is\". A profile could easily be used for on/off in simple form, or even better programmable to those pixels @ varying ND value.
It\'ll be interesting to see how much DR can be improved without a CFA (assuming it will not be replaced with an NDFA). My expectation is the improvement in tonality \"within\" the DR more so than the expansion of the DR. Of course, both would be nice.
Now, I just need to get the patent for my variable-NDFA finished, so I can start collecting my royalties. Of course, I have to finish working on my tilt-shift sensor patent first.
douglasf13 wrote:I\'m just not convinced the tonality will be a huge upgrade. Everything I read about these b&w sensors is in regards to detail, not being more b&w film-like in terms of tonality. It\'s still a digital sensor, after all.
+1 @ more detail
As to tonal range ... a total elimination of the CFA would render all pixels capable of recording the same amount of light energy, but wouldn\'t the range of a pixel\'s response to the light energy still be limited to the individual pixel\'s capabilties. In that regard, it would seem that the DR wouldn\'t be any greater ... but the transitions within that DR would have more accurate/precise increments than those interpoplated from from the RGB CFA (with different energy response @ R vs. G vs. B). To me, this is where the gain would be.
Even though we can currently convert the color to B&W, and play with it greatly in PP, the initial values are a guess/reverse engineering of the light that has passed through the CFA. Without the CFA ... it is no longer a reverse engineering effort, but rather a direct response. To me, that would render the accuracy/precision greater from the onset, leading to Deming 101 principles.
Here\'s a twist that might be interesting. Given the use of the RGB CFA approach ... consider if you were to use a variable-NDFA, somewhat similar to Fuji using the multi-size pixel approach in their super-CCD, or the variable-sized grain structure of some films, that yielded greater DR than their contemporaries. In doing so, you might be returning to a semi-interpretive approach, but it could offer some ability to handle those scenes that exceed the pixels DR for protecting highlights.
If the non-NDFA pixels are not \"clipped\", then the amount of ND could be offset accordingly via the algorithm. If the non-NDFA pixels are clipped, then the algorthm could utilize the NDFA pixels \"as is\". A profile could easily be used for on/off in simple form, or even better programmable to those pixels @ varying ND value.
It\'ll be interesting to see how much DR can be improved without a CFA (assuming it will not be replaced with an NDFA). My expectation is the improvement in tonality \"within\" the DR more so than the expansion of the DR. Of course, both would be nice.
Now, I just need to get the patent for my variable-NDFA finished, so I can start collecting my royalties.
douglasf13 wrote:I\'m just not convinced the tonality will be a huge upgrade. Everything I read about these b&w sensors is in regards to detail, not being more b&w film-like in terms of tonality. It\'s still a digital sensor, after all.
+1 @ more detail
As to tonal range ... a total elimination of the CFA would render all pixels capable of recording the same amount of light energy, but wouldn\'t the range of a pixel\'s response to the light energy still be limited to the individual pixel\'s capabilties. In that regard, it would seem that the DR wouldn\'t be any greater ... but the transitions within that DR would have more accurate/precise increments than those interpoplated from from the RGB CFA (with different energy response @ R vs. G vs. B). To me, this is where the gain would be.
Even though we can currently convert the color to B&W, and play with it greatly in PP, the initial values are a guess/reverse engineering of the light that has passed through the CFA. Without the CFA ... it is no longer a reverse engineering effort, but rather a direct response. To me, that would render the accuracy/precision greater from the onset, leading to Deming 101 principles.
Here\'s a twist that might be interesting. Given the use of the RGB CFA approach ... consider if you were to use a variable-NDFA, somewhat similar to Fuji using the multi-size pixel approach in their super-CCD, or the variable-sized grain structure of some films, that yielded greater DR than their contemporaries. In doing so, you might be returning to a semi-interpretive approach, but it could offer some ability to handle those scenes that exceed the pixels DR for protecting highlights.
If the non-NDFA pixels are not \"clipped\", then the amount of ND could be offset accordingly via the algorithm. If the non-NDFA pixels are clipped, then the algorthm could utilize the NDFA pixels \"as is\". A profile could easily be used for on/off in simple form, or even better programmable to those pixels @ varying ND value.
It\'ll be interesting to see how much DR can be improved without a CFA (assuming it will not be replaced with an NDFA). My expectation is the improvement in tonality \"within\" the DR more so than the expansion of the DR. Of course, both would be nice.
May 08, 2012 at 08:58 AM
Previous versions of RustyBug's message #10615855 « Leica B&W looks like a reality »
