There is a contrast slider but the description in RawPaedia does not say whether 0 gives all the first algorithm and none of the second or all of the second and none of the first.
Also I use the contrast mask for the capture sharpening and using AMAZE+VNG4 dual demosaicing to see which areas would use AMAZE and which use VNG. Would the white areas use AMAZE and the Black VNG or vice versa? (It would also be useful if the two contrast sliders used the same scale or there was a note in RawPaedia about how to convert.)
They use roughly the same scale. The main difference is, that the dual-demosaic contrast threshold uses the result from left demosaicer whereas the capture sharpening contrast threshold uses the result from dual-demosaic.
The dual mosaic contrast slider goes from 0 to 100 and the one for capture sharpening goes from 0 to 200. I assumed that if using Amaze + VNG that 2 on the capture sharpening would equate to 1 on the demosaicing slider.
I have now done some testing with 3 versions of the same picture, one with AMAZE only, one with VNG4 only and one with both with the contrast slider on the demosaic setting set at 14. The split of areas that used AMAZE vs VNG matched the contrast mask when the contrast mask setting for capture sharpening was set at 28. I also did a dual demosaic at 28 and the areas matched the capture sharpening mask set at 56.
My method now is to decide which areas I want to use AMAZE and which to use VNG and set the demosaic method to VNG. I then enable capture sharpening and the contrast mask set the slider to 200 and bring it back down until everything I want in AMAZE is white. I make a note of the value and then halve it. I select the dual mosaic and use that halved value as the value for the contrast level there. I then recheck the mask and make sure the correct areas are all white where they should be and black where they should be. I adjust the capture sharpening contrast until things are OK and not the level. I then put half that level in the demosaic contrast box, etc. (usually only the one adjustment is needed).
Having got the demosaicing where it should be I then use the contrast mask for capture sharpening to sharpen the areas I want. As it happens that is usually lower than the split of areas that I want to use for the different demosaicers, as I want to sharpen some areas where I am using VNG4, so the two figures may end up close, but not usually the same.
OK so thinking about it a bit, how does the program find the contrast. What I mean is that it has to demosaic before it can find where the contrasty areas are, but can’t demosaic until it knows what level to do the split. Does it just use the first method to find the contrast levels and then use that to find the areas to use the second method on based on the setting. Or does is do that as a first pass then use the setting to fine tune and then do a second pass at split demosaicing based on on the new info?
Dual demosaic uses the demosaicer which gives better results for high frequency details to build the contrast mask to blend with the results of the demosaicer which gives better results for low frequency details.
Btw: I just added Amaze+Bilinear, DCB+Bilinear and RCD+Blinear.
These are intended to be faster then their VNG4 counterpartes, while giving almost equal results.
Intentional use is for mass processing of video raw files
As you correctly supposed, I was talking about dual demosaicing. I assume that the higher up the algorithm is in the list, the better it is with high frequency details.
That being the case, if I want to use the capture sharpening contrast mask to see what areas would use Amaze and what areas would use VNG4, I should set the demosaicing algorithm to just plain Amaze and make a note of the contrast setting that gives me the correct split and then change the algorithm to Amaze+VNG4 and just use a figure of half of what I noted.
Just for information: there is one difference between the mask for dual demosaic and the mask for CS. The CS mask shows areas with clipped highlights in black (to not sharpen them, as clipped highlights are non linear data), whereas the DD mask is just contrast based.
Not possible. The capture sharpening contrast level goes from 0–200 whilst the dual demosaicing contrast level goes from 0–100. More to the point I used the same raw to generate 4 tiffs, one was amaze only, one was vng4 only one was dual with a 14 set on the dual demosaicing, and one was dual with 28 set. What I was aiming for was a the same split as I saw when I had the capture sharpening mask set to 28 (with AMAZE only as the demosaicing).
The one that matched was the one with 14 set on the dual demosaicing mask. The one with 28 set had more VNG and as it happened matched what I saw if I set the capture sharpening contrast mask to 56.
Now, once I have used the mask to work out where I want which algorithm I set that and then use the CS mask to work out where I want the sharpening which might include some of the areas that are VNG4 demosaiced so may well be lower than the figure I halved to set the demosaic split.
I must not have made myself clear. When I used 14 as the setting for the dual demosaicing (DD) contrast slider, the split between the AMAZE and VNG areas matched the split shown in the contrast mask with the capture sharpening (CS) slider set at 28.
If I set the DD slider to 28 the split between the AMAZE and VNG areas does NOT match the CS mask at 28. Setting the same values gives a different split. However, a 28 DD split does match the areas shown when the CS slider is set to 56.
If I want the split that I see when the CS mask is set to 28 why would I not use the DD setting that gives me that split (14) rather than the DD setting that gives me a different split just because the setting happens have the same numerical value on a different scale?
The CS slider goes to 200. If I want the split I see when that mask is set to 125 then how would I set it on the DD mask when that slider goes only to 100?
