Ralf Brown
Ralf Brown
Still present in 4.6 / 4.7.0+91.
The underlying issue is that the crop module's parameters are merely what percentage to remove from each side of the image *it is given*, followed by forcing the crop back...
@jenshannoschwalm Note that the new box-filter code in this PR only does one iteration of box mean, while the calls to `dt_box_mean` which were replaced each did **eight** iterations....
> The walking box filter would definitely be faster ... If you mean the new `blur_2D_bspline` in this PR, it's unlikely to be faster than `dt_box_mean` with one iteration (the...
> Besides having to implement a 2ch kahan version there seems to be "something else wrong" in the box filter codes. Still investigating. It occurred to me (independently, a few...
For that one, I would suggest `piece->data = dt_calloc1_align_type(dt_iop_colorequal_data_t);`
Does this also happen with the PIxdl plugged in first and then adding the LG? It's either a bug in GPhoto2 or the darktable code gets confused by having two...
High brilliance values can produce extremely large pixel values (e.g. in the billions, where 1.0 is pure white) -- see #12442. If you have highlight reconstruction turned on in FilmicRGB,...
It's the same underlying issue as in filmic's reconstruction - both that and the guided filter work by blurring the input, which means the supernova-bright pixels generated by overuse of...
> I'll make a PR that adds soft limit of ±20% for all brilliance sliders. Really only of use for the global and highlights sliders, as the pixel must be...