Hi,
As discussed with boucman on irc, I've implemented a new "sticky"
preview mode in the branch sticky_preview in the main repo. Not a lot of
changes...
For now you can have a "sticky" preview mode only by exploiting a "bug"
: pressing 'd' while in preview mode and coming back to lighttable.
I
Hi all,
This a call for testers. I've done a lot of cleanup/refactor in the
collect code : "collect_rework" branch in main repo.
The actual code was... well... I've dropped around 1000 lines of dead
code and unused fct...
Once the cleanup done I've corrected some bug/strange behaviours :
- same
arktable.dir/all] Erreur 2
> make: *** [all] Erreur 2
>
> Regards
>
>
>
> Le mardi 30 septembre 2014 à 22:21 +0200, AlicVB a écrit :
>> Hi all,
>>
>> This a call for testers. I've done a lot of cleanup/refactor in the
>> collect code : "c
Hi,
just for your information, spots use masks, and, basically, just store
masks ids in the iop params. For that it use a static array of 64 items.
It could be increased, thought, as 64*sizeof(int) is not really big. but
imho, there's no real need for that, 64 is more than sufficient for most
u
This should be fixed now in master by commit
37c205bb5ef3a8d5d87295dd3cfd1481f063a5da
thanks for reporting.
Aldric
Le 14/12/2014 17:32, Patrick Shanahan a écrit :
> * Patrick Shanahan [12-14-14 11:31]:
>> * Matthias Bodenbinder [12-14-14 11:15]:
>>> Am 13.12.2014 um 20:32 schrieb johannes hani
w the rendering code very well, it's really possible
that my little hack cause weird situations...
- Changing the prefs doesn't invalidate relevant cache numbers.
- My English descriptions for the prefs are awefull, I know it ;)
Thanks for testing
Aldric
Le 15/01/2015 22:21, Al
Nice...
To add the polygonal drawing tool, I have introduce a new "clonetool"
variable in the spot params. So I'd propose to add here a third type of
tool when your superb curve tool become ready.
What do you think ?
AlicVB
Le 16/10/2012 08:30, Henrik Andersson a écrit :
>
module (not a lot of work for the majority), or is there a
way to reuse blendif ?
Thanks for your insights...
and sorry for my English.
AlicVB
--
Everyone hates slow websites. So do we.
Make your web apps faster
ion approach for spot removal,
>> AlicVB check
>> http://cs.engr.uky.edu/~jacobs/classes/2010_photo/readings/PoissonImageEditing.pdf
Will have a look, but don't expect too much from me on this side, I
don't think I'm able to do such complicated things...
>> /H
&g
[...]
>
> For the feature bloat, I just wanted to add darken, lighten and such
> basics blending modes, nothing more.
That's done. Use https://github.com/AlicVB/darktable.git branch
"clone-blendif" if you want to try.
Now it's time to sleep
rwise, we would have a _very_ small image !)
Here, if you test with CPU, we see two black zone on the resulting image
(normal)
with Opencl, you get some strange lines...
Here is a patch which correct this by testing in the opencl routine if
the pixel we want to interpolate exist.
Hope everythin
I've done a Pull request on github for this bug, and another one (always
in the clipping module)
AlicVB
Le 13/11/2012 22:03, Ulrich Pegelow a écrit :
> Oh great, thanks for reporting, and even with a patch! In fact we should
> deliver the same result on OpenCL and CPU (modulo flo
stand it's not a big deal, but if someone with openmp skill can
explain me why...
Thanks in advance.
AlicVB
--
Monitor your physical, virtual and cloud infrastructure from a single
web console. Get in-depth insight
Ok, thanks.
and sorry for this "off topic" noise.
AlicVB
Le 25/11/2012 02:55, johannes hanika a écrit :
> even though the threads are most likely held idle in the background
> and not created new for every openmp loop, there will be an overhead
> starting a parallel section
atures"
I just made a pull request for this.
thanks
Aldric (aka alicvb)
--
Monitor your physical, virtual and cloud infrastructure from a single
web console. Get in-depth insight into apps, servers, databases, vmw
>
> cheers,
> jo
>
>
> On Mon, Nov 26, 2012 at 11:21 AM, AlicVB wrote:
>> Hi all,
>> Now that release has been done (great work, btw), it's time for new stuff,
>> right ?
>> So I have on a branch a new system for keystone modifications.
>> Th
ldric
Le 28/11/2012 11:23, AlicVB a écrit :
> ok, thanks
> I answer you directly there too...
> ...and I add a new commit to change boolean for int and to correct an
> obvious bug.
>
> Aldric
>
> Le 28/11/2012 04:20, johannes hanika a écrit :
>> i put some comments o
Hi all,
Just a thought : I think module should expose function if they do image
distortion. Now we have just the spot iop which suffer of problem, but
the mask tool will have this problem too :
example :
- take an image and apply some rotation.
- now when you want to define some spots, the image
Hi all,
I've just done a pull request for my multi instances branch.
I don't know if the PR is the right way to go, as it may need further
tests... Don't hesitate to reject it !
And I'm sorry for the multiple commits in the pr... I don't know why
they are here...
Thanks
Aldric
---
Oh good catch, thanks.
I've corrected this bug and the one with the history stack... Need some
tests and it will be merged soon, I think.
Aldric
Le 20/12/2012 17:36, Alexander Rabtchevich a écrit :
> Hello
>
> Ctrl+C - Ctrl+V in darkroom doesn't work also - it crushes dt.
>
> With respect,
> Alex
Hi all,
just some quick words to let you know :
- I'm still alive :)
- I'm not at home, so I will not be back to dt dev before 5 days at least :(
- I can't reproduce the bug mentioned here, no pb to save / apply xmp
and styles from lt, even with multi instances...
So I someone manage to reproduc
Hi,
thanks for correcting this one. I've already got some problems with
empty strings and xmp when coding multi instances, so I think your
workaround may avoid other bugs like this...
Aldric
Le 01/01/2013 22:32, johannes hanika a écrit :
> thanks for reporting, i worked around the null pointer
083a680cab79
>
> and
>
> b522516dd891b21c5b0ce7bb833e2593b42090f9
>
> when you're back and see if that makes any sense.
>
> cheers,
> jo
>
> On Mon, Dec 31, 2012 at 6:30 AM, AlicVB wrote:
>> Hi all,
>> just some quick words to let you know :
>>
Hi all,
I think that now a lot of testers have tried the multi instance feature.
Maybe it's time to set which iop can't have multi instances. The goal
here is not to say "it's usefull" or not, but "It may cause troubles"...
Here is the list of iop I think we should exclude :
1000 gamma
800 color
t blendif to :
> framing
>
>
> apart from these, the blendif == multi-iop rule works
>
>
> (just sending to the ML so it's not lost)
>
> Regards
> Boucman
>
>
> On Thu, Jan 3, 2013 at 6:20 PM, Ulrich Pegelow
> wrote:
>> Hi Aldric,
>>
>&g
Hi all,
As discussed with hanatos, I've created a new branch on the main repo
called "distort"
This branch in an implementation of distortions management.
What's new ?
Modules who take care to positions no longer need to deactivate
distortions iop :
- before : You rotate an image. As soon as you
9:27, Pascal Obry a écrit :
> AlicVB,
>
> That's awesome :)
>
>> Please have a look, test it and if everything goes well, I think we can
>> merge it to master :)
> I have tested it a bit, I found an issue with the spot removal, when I
> click the size of the two circle
This is because when you move a spot, the circle might be distorted too
(try setting a big keystone correction and then add spot, for example)
and recreate the spot on each mouse move would be too slow... And here,
we speak just about a spot, but with curves, it will be even more obvious !
I agr
; like commented out set_sensitive calls and such..
>
> apart from that it workes like a charm!
>
> merged to master for more testing :)
>
> -jo
>
> On Sat, Jan 5, 2013 at 8:01 AM, AlicVB wrote:
>> This is because when you move a spot, the circle might be distorted too
btw, can I delete the distort branch now ?
Aldric
Le 05/01/2013 15:34, AlicVB a écrit :
> ok, I've done done some cleanup (sorry for that...)
> I've changed a little the spot drawing when you move the spot, to
> avoid confusion.
> Pushed to master
>
> Aldric
> Le 05
>
> On Sun, Jan 6, 2013 at 3:49 AM, AlicVB wrote:
>> btw, can I delete the distort branch now ?
>> Aldric
>> Le 05/01/2013 15:34, AlicVB a écrit :
>>> ok, I've done done some cleanup (sorry for that...)
>>> I've changed a little the spot drawing
Hi,
the code is now on the main repo, branch "masks2"
no big changes, except that xmp read/write is almost ok (still a little
bug with double value, ...)
Feel free to test and comment.
Aldric
Le 16/01/2013 21:25, AlicVB a écrit :
> hi all,
>
> After some evenings o
Ooops... sorry for that. Fixed now.
thanks
Aldric
Le 18/01/2013 12:28, Ivan Tarozzi a écrit :
> Il 17/01/2013 23:28, AlicVB ha scritto:
>> Hi,
>> the code is now on the main repo, branch "masks2"
>> no big changes, except that xmp read/write is almost ok (still a
hi,
Le 18/01/2013 19:40, Ivan Tarozzi a écrit :
> Thanks, now it works :)
>
> I done a quick test using monochrome plugin, and I noticed a problem
> using 2 near circles.
>
> Here a detail of my image:
> https://www.dropbox.com/s/49em9mj6p59pttn/mask2-problem01.jpg
Yes, I'm aware of that. not very
pe you don't mind if I make a
> suggestion regarding the organisation of the regions.
>
> On 16.01.2013 21:25, AlicVB wrote:
>> for users :
>> - forms can be added with the "multi-instance" menu. for instance you
>> can just add circles
> > ...
>
one mask from another and
> to apply an iop to the composed region.
>
> On 19.01.2013 15:14, AlicVB wrote:
>> in ASP, everything is display, as if it works with layers. to resume
>> in an example : ... This is not the way dt work as the pipe order is
>> fixed. dt works lik
Hi all (and hi Pascal)
I have a little concern about the actual lightroom implementation in the
code.
The context : As some of you know, I'm currently working on mask
implementation. This implies some changes to spot removal iop, as it
will reuse the forms used by masks.
So I have updated spot
Thanks for your answer,
The solution you propose is not really good imho. Because, from a dev
point of view this mean that each time you update a module, you would
have to ask you how to handle this in lightroom code. And as you know,
final design need lot of try, which doesn't suffer any delay
P
Le 21/01/2013 22:12, Pascal Obry a écrit :
> Aldric,
>
>> parafin propose a solution : duplicating struct and iop version in your
>> header, instead of including *.h
>> That way, you can create safe history stacks and if the iop version has
>> been updated, the iop legacy_params fct will update p
Pascal,
I just have done a quick look at your code. You add a button somewhere
in darkroom, isn't it ?
So you read xmp from darkroom mode and change iop params directly ?
IMHO it's not the right place to do such stuff. AFAIK xmp
importing/exporting is usually done in the "file manager" (or call i
Le 21/01/2013 22:39, jeremy rosen a écrit :
>> He propose to add asserts to drop params inclusion if an error occurs
> more precisely, add
>
> assert(iop version = last version lr knows about)
>
> somewhere in DT init for all iop used by lightroom
>
> this way whenever a dev changes a iop version,
Pascal
What a quick fix ! thanks for that :)
just to clarify, that was not "my" problem. My concern was more the
modularity respect, my work on masks was just here to illustrate that.
Anyway, thanks, I'll merge that on my branch soon.
Aldric
Le 22/01/2013 17:03, Pascal Obry a écrit :
> Aldric,
>
Hi,
Le 23/01/2013 21:45, Pascal Obry a écrit :
> Aldric,
>
>> For the curve tool :
>> - the tool is based on Henrik's work, with some additions/changes
>> - This is a bezier curve, initialized as a catmull-rom
>> - Points are supposed to have aligned control lines, to simplify the ui
>> - you can a
)
Anyway, there's lot of chance that mask ui will change soon.
Aldric
2013/1/24 Ivan Tarozzi mailto:itaro...@gmail.com>>
Il 23/01/2013 21:27, AlicVB ha scritto:
> Hi,
> I just want to let you know that I've done some enhancements
(imho) on
> the ma
Hi, thanks for all the infos.
I've briefly looked, and my first conclusion is... I need to read it
again, but not at the end of a work day :)
All that is completely new for me (mesh, tessellate, quads...), and I
fear it will be too difficult... but let's try !
btw, can you point me to the page w
Hi,
The bugs with the ui (remove forms, etc...) should be resolved now.
Thanks for the report
Aldric
Le 24/01/2013 09:45, alic VB a écrit :
Hi Benoît,
thanks for your feedback.
there is definitely a bug with form removing. Will have a look asap.
For your other concern, that's a limitation of dt
Hi Ivan,
yes, this is certainly your "masks" test. That's why we always say to to
backup everything before testing.
In your case, I even think that's not masks but multi-instance. This
mean that this concern all tests you may have done on git master.
Best solution would be dropping your db and re
Hi,
Le 19/02/2013 10:17, Ivan Tarozzi a écrit :
> Hi Aldric,
> and thanks for your response :)
>
> Il 19/02/2013 08:35, AlicVB ha scritto:
>> Hi Ivan,
>> yes, this is certainly your "masks" test. That's why we always say to to
>> backup everythin
Le 19/02/2013 14:49, Ivan Tarozzi a écrit :
> [...]
> Here my history table schema in faulty database
>
> $ sqlite3 library.db ".schema history"
>
> CREATE TABLE history (imgid integer, num integer, module integer,
> operation varchar(256), op_params blob, enabled integer,blendop_params
> blob, ble
Hi all,
I finally let the mask in paused for 1 or 2 days and work on correcting
bug in C&R.
I've created a new branch on main repo for that : CR_fixes
Currently here's what has been done :
1- save changes done with cropping area even if the pipe is still
working (bug #9245)
2- save ratio in image
Le 01/03/2013 15:03, Richard Levitte a écrit :
> Hi,
hi
> I just had a closer look, and summary of it all: I like!
>
> I have found one thing that differs from before... not sure if this
> is something you care about or not, but... before your changes, the
> code would adapt the aspect ratio to t
Le 01/03/2013 22:05, johannes hanika a écrit :
> cool, seems to work fine.
>
> one question, what's with:
>
> -roi_in->x += interpolation->width;
> -roi_in->y += interpolation->width;
> -roi_in->width -= 2*interpolation->width;
> -roi_in->height -= 2*interpolation->width;
> +//r
Hi all,
interesting discussion indeed. But please don't change too much things
now in spot removal, as in a (near ?) future, this iop will use the
masks tools...
So we will need a more global discussion about that.
FYI, I have already this change working at home :)
thanks
Aldric
Le 03/03/2013 2
Hi,
problem seems to come from my last commit on this file (big patch to
correct lot of bugs and strange behavior of the module).
I see the root cause : self->dev is NULL. The question is more how to
get a work around to get the image size.
Will try to correct this as soon as possible.
thanks
ok, I just push a fix for this.
Please try again and report if it works now...
Aldric
Le 17/03/2013 12:59, AlicVB a écrit :
Hi,
problem seems to come from my last commit on this file (big patch to
correct lot of bugs and strange behavior of the module).
I see the root cause : self->dev
oh oh, my bad here...
I just had a quick look.
this shouldn't be too hard to solve for most cases, by restricting the
line to the visible area, but there is some cases way more problematics
: if the handles and the lines are outside the visible area (gnd is
before crop in the pipe, so it can ha
I just fix it in master by allowing zoom under the "zoom to fit" level.
This should solve some pb in vignette too.
Aldric
Le 31/03/2013 14:32, AlicVB a écrit :
oh oh, my bad here...
I just had a quick look.
this shouldn't be too hard to solve for most cases, by restricting t
n a écrit :
On Sat, Apr 6, 2013 at 9:18 PM, Alexandre Prokoudine
<mailto:alexandre.prokoud...@gmail.com>> wrote:
On Sat, Apr 6, 2013 at 10:52 PM, AlicVB wrote:
> I just fix it in master by allowing zoom under the "zoom to fit"
level.
> This should solve
Hi,
about dt hang, I can't reproduce here... Is this reproductible ?
do you, by any chance, have a backtrace of the crash ?
Aldric
Le 13/04/2013 21:28, Alexander Rabtchevich a écrit :
> Alexander Rabtchevich wrote:
>> Pascal Obry wrote:
>>> The size of the spot is not remembered. This makes applyi
Hi,
I just push a fix which hopefully correct your bug bug too.
Aldric
Le 13/04/2013 21:54, Alexander Rabtchevich a écrit :
> AlicVB wrote:
>> Hi,
>> about dt hang, I can't reproduce here... Is this reproductible ?
>> do you, by any chance, have a backtrace of the cras
Hi Ulrich.
First thanks for your comments :)
about the mask line in the module, I agree it's actual position is not
optimal, but I don't agree with your solution either, because it would
mean that you have to do a lot of "click" to add a mask : 1- select
blend to whatever except off, 2- select
Hi,
I've just pushed to master a new icon like Ulrich suggest. I'm not
really happy with it, thought : it's too small, so if someone with a
better drawing skill or idea want to enhance it, no problem ;)
I've fixed some pb with the icon state too.
Thanks
Aldric
Le 05/05/2013 11:06, Ulrich Pegelo
Hi Olivier,
Le 05/05/2013 20:50, Olivier Tribout a écrit :
> Hi Ulrich, Hi Alic
>
> Thanks for the impressive job.
Merci :)
> After playing with it I have some remarks:
>
> 1) Wording of the latest change
> May I suggest
> - to rename "drawn mask" by "geometrical blend"
> - to rename "parametric m
hm...
The only pb I manage to reproduce is that the color-select mode is not
disabled by the click on one of the icon to draw the mask.
I will try to correct this.
Thanks for the report.
Aldric
Le 14/05/2013 10:22, David Vincent-Jones a écrit :
> darktable 1.3+292~g1f7297d
>
> I am finding a pro
hi,
I think I've fixed this in git master. can you test and confirm ?
thanks
Aldric
Le 14/05/2013 10:22, David Vincent-Jones a écrit :
> darktable 1.3+292~g1f7297d
>
> I am finding a problem creating masks in color-zones when 2 instances
> are in play.
> The process works correctly when there is
Hi,
thanks for the report. The backtrace is useful, but I don't understand
why it crash at this point.
Some infos :
do you manage to reproduce the crash ?
what is the size of your image ?
what is approx the size of your shape ? in % of the total image
do you use sharp or smooth corners ?
anything
hi,
I just push a fix for point 2
For point 1, I have to admit that I don't fully understand (I don't
speak English very well) can you please explain exactly the problem, and
detail a "step by step" procedure to reproduce the bug ?
Thanks for reporting
Aldric
Le 18/05/2013 12:55, Roumano a
hi,
That should work... everything should be handled in the legacy_params
fct. and the process routine should reproduce exactly the old algorithm
too...
I've tested it when I've implement the change. Seems it need more tests ;)
I'll have a look asap.
thanks for the report
Aldric
Le 26/05/201
hi all,
There seems to be a problem with blend_params update.
Let's describe one of them :
houz added recently a duplicate feature for multi-instances. (everything
is in dt_iop_gui_duplicate in develop/imageop.c)
I've added the duplication of blend_params to this fct.
The problem is that line 109
Hi,
I've already fix a bug about dble-click in the past as well as a crash
with straight lines... and can't reproduce your pb anymore :(
Just to be sure, what exactly version of dt do you have (Pascal's ppa
have been updated in the night, but I don't know at which revision
previous ones have bee
oo lazy to test them
one by one :(
But be carrefull, I'm a noob in sql, so I don't now if adding index can
have weird effect elsewhere or not !
AlicVB
Le 15/09/2013 09:13, Pascal Obry a écrit :
> James,
>
>> The snark in me suggests that if you were running the Solaris v
Hi all,
I don't know if I am the only one, or if I miss something, but I find
the keyword module quite confusing and not really user-friendly.
To replace in context, I've about 50K photos all tagged, with a heavy
use of hierarchical tags (ex : Iceland|Myvatn|Hverir or Landscapes|volcano)
Here i
.
I can give it a try, if you want.
Thanks
Aldric
Le 22/10/2013 10:30, johannes hanika a écrit :
heya,
On Mon, Oct 21, 2013 at 10:28 PM, AlicVB <mailto:alic...@gmail.com>> wrote:
Hi all,
I don't know if I am the only one, or if I miss something, but I find
the keywor
Hi all,
There seems to be something wrong with styles and multiinstances.
a simple way to reproduce the problem :
1- take a new photo
2- set some params to a module (ex : gnd with saturation=0.10)
3- create a new instance of this module
4- set some other params to this instance (ex saturation=0.90
rule
Note that I've not reverted the previous change to the keyword module,
but now that this part is included in the collect module, we can drop
it's "select" part move it to the right panel, and focus on tag
modifications.
Thanks for your feedbacks
Aldric
Le 22/10/2013 12
Hi Pascal,
Le 23/10/2013 09:28, Pascal Obry a écrit :
> Aldric,
>
> Seems like there is an issue with the way the multi_priority is set.
>
> /.../
>
> Then add a new instance of exposure and back to lighttable, then:
>
> sqlite> select num,operation,multi_priority,multi_name from history
> where i
Hi,
I have some change pending, to include the keyword module into the
collect one. It should be ready to be merged in 1 or 2 days. Is it ok ?
Aldric
Le 28/10/2013 15:42, Tobias Ellinghaus a écrit :
Hello everyone,
as you might have heard we are thinking about doing the next minor release
(1.
Hi all,
a quick update :
- tag collect is now integrated in the collect module, with tree display
of hierarchical tags
- keywords module has been move to the right panel and its "collect"
functionality has been dropped
Now it would be great to delete keyword module completely, but it still
has
hi,
at least Henric is not alone with this bug now !
He has reported that "freeze" on irc, but nobody else manage to
reproduce the pb :(
If you have some time, can you please answer some more questions/tests ?
1- what is your distro, your gtk-glib version ?
Can you try to delete (or rename) your
Hi,
With the help of Ulrich, we finally have found how to correct the
problem, the fix has been pushed to master. Note that we revert to the
old "double-click" behavior.
Thanks for the report
Aldric
Le 07/11/2013 23:14, Togan Muftuoglu a écrit :
>>>>>>
80 matches
Mail list logo
