Hi Roman, Perspective correction always requires cropping, so from a GUI standpoint, I can understand why the keystone feature was added to the C&R module. But true, each interpolation step add artifacts.
I can check the Python script, it doesn't seem too difficult to rewrite it in C. Cheers, Nicolas On Thu, Aug 13, 2015 at 5:28 AM, Roman Lebedev <lebedev...@gmail.com> wrote: > On Wed, Aug 12, 2015 at 7:21 PM, Nicolas Lannuzel <nlannu...@gmail.com> > wrote: > > Hi, > Hello. > > > This is a small patch for the "crop&rotate" module. It adds a slider to > > fine-tune the amount of vertical keystone correction. The slider can move > > from 0 (no correction, original image) to 100% (verticals are verticals). > > > > Sometime architecture photos can look artificial if the keystone > correction > > is fully applied. Setting the factor to 80 or 90% gives a more natural > look. > Indeed, makes sense.. > > > I've done it by translating C and D points linearly on the X axis (in > > clipping.c, function commit_params): > > float kf = (float)(p->vert_keystone_amount / 100.0); > > > > d->kxc = d->kxb + kf * (d->kxc - d->kxb); > > d->kxd = d->kxa + kf * (d->kxd - d->kxa); > > > > > > I've also updated the legacy_params function, and the > LRDT_CLIPPING_VERSION > > from 4 to 5. Did I miss anything and what do you think of this patch? > Some not really meaningful words: > Right now, there are 2 IOPs that do pixel interpolation - lens > correction and C&R. > This is slow[er than doing it in one step] and probably produces > slightly lower quality than 1 step would. > C&R only needs to do pixel interpolation for this perspective > correction and rotation, > just cropping can be done in without it. > > So there is a push to move perspective correction [and rotation?] into > lens correction module. > New implementation algorithm will be a bit more mathematically > sophisticated, > and in it's current version It already does have a similar function, see: > > http://sourceforge.net/p/lensfun/code/ci/a532325328117177d3cefdd7d9fcea7aafaeb805/tree/tools/perspective_control/perspective_control.py#l291 > [I have CC'd tools author, Bronger] > > But there is no ETA on when it will happen. > > Now, C&R iop is, well, a delicate piece of code :), so we are a little > scared of touching it... > NOTE: i'm not saying that this patch is bad or that we do not want it. > > So my personal suggestion would be: try that tool > [perspective_control.py], find a few issues there, > discuss it's functionality, etc. > And if you can, help rewriting it in C. I'm afraid i do not know > Python that well do be able to do it by myself. > Once that tool is in C, integration of this functionality will be much > closer. > > (probably that was not what you have wanted to hear, but that is my > view on the subject) > > > > > Cheers, > > Nicolas > Roman. > > > > > > ------------------------------------------------------------------------------ > > > > _______________________________________________ > > darktable-devel mailing list > > darktable-devel@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/darktable-devel > > >
------------------------------------------------------------------------------
_______________________________________________ darktable-devel mailing list darktable-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/darktable-devel
