hi,
i have to admit i didn't look at the patch itself yet. but in principle
i can very much relate to the issue of `overstraightened' pictures, i
think they look terrible. providing an easy way to do only 80% sounds
very useful to me, so if we can have that without waiting a year for
something else to happen i'm all for it :)
cheers,
jo
On Thu, Aug 13, 2015 at 9:43 AM, Roman Lebedev <lebedev...@gmail.com>
wrote:
[this is a resend of the mail since OP used old, dead mailing list]
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-de...@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/darktable-devel
___________________________________________________________________________
darktable developer mailing list
to unsubscribe send a mail to
darktable-dev+unsubscr...@lists.darktable.org
___________________________________________________________________________
darktable developer mailing list
to unsubscribe send a mail to darktable-dev+unsubscr...@lists.darktable.org