Hi Bud,
When I do a histogram transformation....on the live preview......then drag the new instance icon to the ScreenTransfer it gives the same visual look.
Now I understand that this method will keep the image linear...is this correct?
Also, the histogram transformation ..... by itself....is a linear transformation....is this correct?
Whilst you are using LivePreview to make the adjustments to the Histogram sliders, you have still NOT yet changed any data in the image. You then drag the NewInstance icon (the little blue triangle) to the bottom bar of the ScreenTransfer window, and you have still NOT changed any data in the image.
Things will get a little bit confused though at this time, because the STF is an action that is always applied 'between' the image and the image display - and that means that your LivePreview may well now get 'double transformed', once by the actions of the Histo sliders, and once by the actions of the STF. Just something to keep in mind!!
In fact, I now usually come up with a 'first trial' using the STF only - in other words I don't use the LivePreview and the Histo sliders at all. Then I 'apply' the STF instance over to the Histo window (via the bottom bar again) and have a quick 'tweak' as needed, paying attention to the 'amount' of pixels that may be getting clipped at the Balck and White ends (in fact, I would normally not accept ANY clipping, but you hopefully see where I am coming from).
At that stage I would possibly even drop a 'copy' of the STF instance on the workspce desktop, as a 'safety net', and then perhaps 'apply' the Histo transform to the image, clearing the STF as I go. If I don't like what I have done, at least I can backtrack to the saved instance.
As far as 'linear transforms' are concerned, I am sure that you would understand that ALL Histogram transforms remain 'linear', providing the 'MTF' (Mid-tone Transfer Function) slider remains at 0.50000. In my mind, once you have applied any Histo or Curve transform that affects the image 'linearity', then you have made some operations, such as Deconvolution, all but impossible to apply. This is where the STF scores so heavily, because it allows you to 'view' the image AS IF the non-linear transform had been applied, whilst the underlying image data remains capable of being processed by algorithms such as Deconvolution, etc.
As far as the likes of HDRW is concerned, whilst this 'can' be used with data that is still 'linear', there is no advantage in doing so - and in fact the likes of HDRW is actually 'easier' to apply once the image has had a non-linear stretch applied, simply because this gives the algorithm 'more headroom' to work with.
As Sander describes, there are only a very few operations that do NOT affect the linearity of the data - and those which he listed are the same as I apply, almost automatically, at the start of every image workflow. Onto his list are the other obvious stages, such as all the Calibration and Alignment steps, as well as all the Integration steps, and would also include an initial Dynamic Crop as well (essential for the likes of me, where I just do NOT get perfect behaviour from my mount during imaging sessions, and so I will ALWAYS have to 'trim off' the edges where only partial data has been obtained.
Whilst on the subject of DynamicCrop, remember that it is also very easy to rotate your crop outline as well - handy if you are trying to either maximise the number of useable pixels in your final image, or if you are after a specific 'framing' requirement (and this is also where an initial 'hard' STF can let you see all the data you need to see).
Nice to see you 'on-board', and looking forward to seeing what you can make PI do!
Cheers,
