maybe it’s best if i write a dedicated i-tif input module
I was just comparing VKDT with other softwares (mostly RawTherapee, Affinity, Dxo Photolab)
With RawTherapee, for instance, you can open most formats including TIFF
Do you plan to code an inpaint feature as well, in the future?
I suppose it is not available yet…
Currently, GIMP allows to inpaint an image (through G’MIC filters); KRITA has a similar option.
On this computer (12th Gen Intel(R) Core™ i7-12700H (20 CPUs), ~2.7GHz) G’MIC is pretty fast (inpaint > multi-scale filter) even with big TIFF
Darktable and RawTherapee have “only” the clone tool option so far
I have read GEGL has a similar option but, judging from some comments on gitlab, at present, it sucks (no offence here)
there’s the inpaint module best used with the inpaint.pst preset. it’s not very smart but works to remove some dust or even some cables against a smooth sky.
In case you are interested…
Just tried to export some RAW (NEF) to DNG with DXO Photolab 7.3
Every DNG does not work with vkdt: same display problem as TIFF (see previous report)
These same DNG work fine with RawTherapee - ART - Windows Photo - Affinity Photo (not tested on darktable)
Here one of them in case you want to download and test it on your side (dropbox link):
Basically every dng exported with Dxo photolab 7.3 is broken on vdkt.
With Dxo there is also the option to export your dng with some more tweaks (e.g. the lens corrections applied internally by Dxo) but I didn’t try this option since the standard mode was already not working …
I think there should be a separate saturation module in vkdt. Sometimes the amount of saturation that is possible with the colour module is not enough, plus it could be used in combination with masks. And it should also be possible to reduce saturation with such a module.
Creating a second instance of colour does not seem to work because it also changes the white balance.
Of course, the best thing would be if Aurélien ported his color balance rgb to vkdt.
note how the point cloud extends all the way to the spectral locus. we can limit to smaller ranges too (my screen will not display these colours, also note how i needed to disable filmcurve and llap since they don’t play well with negative rgb values):
… you get the idea. if that is not enough, or you want to selectively saturate only some hues, you can use the colour mapping radial basis function bulit into the colour module, to stretch, say, only the red shades away from white:
Besides, sometimes for very artistic effects, somthing like pop art or so, extreme saturation and contrast are needed, or other times some parts of the photo need to be almost colorless.
Btw I will ask more questions about the grade module and I was wondering if it’s possible to do this with it but so far I had a quick look at it and I find it rather difficult to use…
How do you get such a big brush in the draw module? Whenever I switch it on, the brush is tiny. Sometimes is is bigger but I don’t know how and why it gets bigger.
This is what I was able to do with the ctrl+click on the saturation slider. However, the problem is now that the color of the sky is a tiny bit too intense. And it would be even more intense if the photo wasn’t converted to srgb. Same thing here, the sky is too intense compared to the city:
I tried to fix this with grade+mask but somehow I was not able to do it. As far as I see, the grade+mask module is there to paint with a certain color over a certain area. But that’s not what is needed here, is it?
I mean, in color balance rgb it’s just a click on the chroma/staturation highlights slider.
hm okay just had a look at the colour balance rgb module. i think it does something different entirely. in this case i’d route a graph like so to desaturate the sky:
to play with saturation in a 2nd instance of colour. no idea what you had with the white balancing earlier, but you need to set the module to operate with a rec2020 matrix then (i.e., bypass the colour matrix/clut). saturation set to 3 again:
any idea how to make this easier to understand? currently requires knowledge that the internal pixel format in the pipeline is rec2020. so if you set colour to that it’ll be an identity transform. maybe should i label the combobox entry “bypass” instead?
The problem is more that “image” is the default matrix setting for the second colour instance - how would anybody know that it is necessary to change the matrix setting? There are so many settings in that module… maybe create a preset with a mask if it’s possible to set the matrix to rec2020 in the preset?
hm preset is totally possible. the alternative is to overwrite the matrix in the image information that is passed along the graph. maybe this notion would be a good idea for embedding profiles in exported images too. in this case the module with default settings would pick up the identity matrix (because image in the combobox) and not do anything. probably is a matrix multiply less efficient (don’t think is measurable).