focus in Zerene
Moderators: rjlittlefield, ChrisR, Chris S., Pau
-
- Posts: 684
- Joined: Thu Dec 16, 2010 2:49 pm
- Location: Barcelona, more or less
focus in Zerene
Is it a feeling of mine, or the parameters I use in DMap are influencing the combined image final focus ???
- rjlittlefield
- Site Admin
- Posts: 23626
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
Well, the DMap parameters are supposed to affect the final result. If they didn't, there would be no point in allowing them to be adjusted.
In general, DMap works by choosing each output pixel from just a single input frame, or sometimes a weighted average of two consecutive frames in transition regions.
The parameters affect the details of how those choices are made.
So, to understand your question we're going to need more information.
Can you show some examples of differences that you find confusing, and what parameter settings produced them?
--Rik
In general, DMap works by choosing each output pixel from just a single input frame, or sometimes a weighted average of two consecutive frames in transition regions.
The parameters affect the details of how those choices are made.
So, to understand your question we're going to need more information.
Can you show some examples of differences that you find confusing, and what parameter settings produced them?
--Rik
-
- Posts: 684
- Joined: Thu Dec 16, 2010 2:49 pm
- Location: Barcelona, more or less
obviously the differences are not great, but it does seem to have more fine detail somestacks than others. Anyway, I still use PMax by default, I have not yet learned to fight against those waves of color that DMAP creates. It could be better, but I'm already happy.
Default DMap
parameters larger than those of DMap defect
PMax
Default DMap
parameters larger than those of DMap defect
PMax
- rjlittlefield
- Site Admin
- Posts: 23626
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
The effects that you're seeing are typical.
PMax is relentless about finding focused detail at all size scales. At the finest scale, it also finds pixel noise and treats that as detail to be preserved, if there's no stronger real detail to find. That's why PMax accumulates noise. Another unfortunate side effect of PMax is that it alters colors and contrasts, especially with deep stacks where the front and back of the stack look very different.
DMap looks at only one size scale, whatever is specified by the radius settings. This makes it less capable of tracking complex geometry, but in exchange it can be perfectly faithful to the colors, contrasts, and noise levels of the original inputs.
In the two DMap results shown in your example, the one with the smaller settings has been more successful at tracking the in-focus surface, so it ends up correctly choosing focused detail in more areas and thus produces an image that looks sharper overall. Neither DMap captures as much detail as the PMax does, especially in low contrast regions where DMap cannot track the surface at all. On the other hand PMax has apparently introduced colors in areas where the original source frames did not have any, for example in the upper left corner of these crops.
Again, this is all typical, and it hints at the reasons why I say that often the very best results come from combining DMap and PMax outputs. The idea is to retain the DMap wherever it did a good job, and retouching from PMax in places where DMap got lost and missed detail. This way you get to use your human judgement to decide in each area whether you prefer the better color/contrast/noise of DMap, or the better detail retention of PMax. It would be great to get both with no human intervention, but I don't know how to do that.
--Rik
PMax is relentless about finding focused detail at all size scales. At the finest scale, it also finds pixel noise and treats that as detail to be preserved, if there's no stronger real detail to find. That's why PMax accumulates noise. Another unfortunate side effect of PMax is that it alters colors and contrasts, especially with deep stacks where the front and back of the stack look very different.
DMap looks at only one size scale, whatever is specified by the radius settings. This makes it less capable of tracking complex geometry, but in exchange it can be perfectly faithful to the colors, contrasts, and noise levels of the original inputs.
In the two DMap results shown in your example, the one with the smaller settings has been more successful at tracking the in-focus surface, so it ends up correctly choosing focused detail in more areas and thus produces an image that looks sharper overall. Neither DMap captures as much detail as the PMax does, especially in low contrast regions where DMap cannot track the surface at all. On the other hand PMax has apparently introduced colors in areas where the original source frames did not have any, for example in the upper left corner of these crops.
Again, this is all typical, and it hints at the reasons why I say that often the very best results come from combining DMap and PMax outputs. The idea is to retain the DMap wherever it did a good job, and retouching from PMax in places where DMap got lost and missed detail. This way you get to use your human judgement to decide in each area whether you prefer the better color/contrast/noise of DMap, or the better detail retention of PMax. It would be great to get both with no human intervention, but I don't know how to do that.
--Rik