search

google-code-export / flowblade

Automatically exported from code.google.com/p/flowblade
1 stars 0 forks source link

compositors failing with Flowblade 0.6.0 #11

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago 
I have 2 general problems with compositors: giving no result at all and 
parameters giving no or limited result. After some testing, I could bring it 
down to following two flaws:

Compositors with geometry setting giving wrong result: when I move the geometry 
one (or more) pixel(s) up, on start and ending keyframe, the compositor does 
work.

Parameters not working: only the utmost positions work.

I added 5 files: 
- bug_workaround.flb: used to test, just using 2 color clips and a region 
compositor. Geometry is lifted up on both keyframes, softness is set to 100.
- bug_workaround.mpg: a rendering of the result. This is OK.
- bug_softness.mpg: softness set to 99 and rendered, the softness is gone ...
- bug geometry.mpg: geometry reset on both keyframes, compositor not working.
- my_system.txt: info about my system, OS and other stuff.

I get similar problems with other compositors and settings.

If you need more information about my configuration, or if you want me to test 
something, just ask. I really like the way the interface is build, and the 
responsiveness is quite impressing. I still have some wishes for aesthetic 
improvements and basic functionality, but, first things first: lets purge some 
damn major bugs out.

Thanks for coding !
Bart.

Original issue reported on code.google.com by bart2...@gmail.com on 1 Aug 2012 at 6:58

Attachments:

GoogleCodeExporter commented 9 years ago 
Investigated a bit further.

The params that Flowblade is passing to mlt.py seem to be alight. It might be 
some bug in mlt itself. Check http://www.kdenlive.org/mantis/view.php?id=2182

Bart.

Original comment by bart2...@gmail.com on 3 Aug 2012 at 12:59

GoogleCodeExporter commented 9 years ago 
Next check (per http://www.kdenlive.org/mantis/view.php?id=2310)

melt does take in account my regional settings:
  melt colour:red -attach brightness:0,5
result is ok (using a comma as decimal sign)

  melt colour:red -attach brightness:0.5
resulting black (dot as decimal sign)

So, is there a parameter to tell Flowblade I'm living in Belgium, or do I have 
to tell Python that Flowblade is coded in another way ?

ex.: Flowblade now passes params for geometry to mlt in the form of
  0=0,0:720x576:100.0
so it ain't taking my regional settings into account ...

Bart.

Original comment by bart2...@gmail.com on 3 Aug 2012 at 1:51

GoogleCodeExporter commented 9 years ago 
Fixed it for the moment by starting Flowblade with:
   env LC_ALL=C flowblade
Don't know if this might introduce other errors ...

Bart.

Original comment by bart2...@gmail.com on 3 Aug 2012 at 10:11

GoogleCodeExporter commented 9 years ago 
I have the same problem. When starting with "env LC_ALL=C flowblade" the 
program does not create the effect properly. Ubuntu 12.04 64bit.

Original comment by apostol...@gmail.com on 10 Sep 2012 at 2:57

GoogleCodeExporter commented 9 years ago 
I agree with the conclusions of bart2pub, thanks for the work on this issue.

I've applied a patch that should fix this. It does two things:

1. Format for giving position information is changed from"x,y" to "x/y" for all 
parameter types involved.

2. This broke backwards compatibility with older project files, so I applied 
fix to loading code that enabled loading those again.

Could someone with the problem test the repository version to see if the patch 
worked. See wiki TestingRepositoryVersion for instructions.

Original comment by janne.li...@gmail.com on 11 Sep 2012 at 9:15

GoogleCodeExporter commented 9 years ago 
Hi Janne,
I'm really enjoying the flowblade. The rendering is very good and it is 
supporting the h264 format (which Cinelerra does not). Just missing the correct 
transition. I followed the instructions. Now is making the transition, but is 
imperfect. The frames of a track are not in sync with the other track as you 
can see in the pictures attached.

Original comment by apostol...@gmail.com on 16 Sep 2012 at 2:58

Attachments:

GoogleCodeExporter commented 9 years ago 
apostologir, I think what may have happened is that you have Issue 11 fixed, 
but now you have Issue 13. Check Issue 13 to see if the symptoms are similar.

What version of MLT you're running? MLT version is displayed at program 
start-up on terminal. Fix for Issue13 is to install MLT 0.8.0 or higher and 
Python bindings for it.

If you're running MLT < 0.8.0 upgrading MLT could fix that. I'll put guide on 
how to that this week, so if you don't how to do that, check that when I've 
added the wiki for it.

Original comment by janne.li...@gmail.com on 17 Sep 2012 at 4:59

GoogleCodeExporter commented 9 years ago 
[deleted comment]
GoogleCodeExporter commented 9 years ago 
[deleted comment]
GoogleCodeExporter commented 9 years ago 
apostologir, now you can try wiki UpgradingMLTForRepositoryVersion to see if 
that fixes your problem.

Original comment by janne.li...@gmail.com on 17 Sep 2012 at 9:16

GoogleCodeExporter commented 9 years ago 
Janne, thanks. It is working 100%.

Original comment by apostol...@gmail.com on 17 Sep 2012 at 11:07

GoogleCodeExporter commented 9 years ago 
Hi. I confirm patch. I used a dissolve composer in FB 0.6.0 between two H264 
(1280x720 50fps) videos and didn't work.

I tested TestinVersion and works! Then I saved project with this version and 
opened with FB 0.6.0 and works too (dissolve composer).

Thanks for that great soft. I used so many video editors under Linux and this 
is, at this moment, my favourite.

Cestomano
Ubuntu 12.04 64bits

Original comment by cestom...@gmail.com on 10 Oct 2012 at 9:41

GoogleCodeExporter commented 9 years ago 
Upss... I've just realized than using dissolve affects geometry/proportion of 
clips. In fact, when I apply dissolve composer (1280x720 @50fps H264), upper 
clip distorts (increases height and width and offset to the right). If I apply 
"align distort", I correct offset but increase width of upper clip.

I attach a example of dissolve. You can see how upper clip distorts and then 
(after dissolve) it shows normal height/width.

Regards! ;)

Original comment by cestom...@gmail.com on 10 Oct 2012 at 11:11

Attachments:

GoogleCodeExporter commented 9 years ago

Original comment by janne.li...@gmail.com on 4 Dec 2012 at 10:51

GoogleCodeExporter commented 9 years ago 
Hi Janne,

The problem still isn't completely solved. The fix you mentioned in comment 5 
works.  I did a quick test and position information seems to be interpreted 
correctly. But with many filters, the problem with the decimal sign still 
persist (as mentioned in comment 2 with the brightness).

Launching Flowblade with the command ' env LC_ALL=C flowblade ' makes it work 
ok. Sorry that I didn't test the repository version, but at that moment I was 
not so keen to install a bunch of other apps on my machine. When I look at it 
now however, that Mercurial thing doesn't seem to be a very heavy load, so next 
time I might try it ?  It runs Flowblade straight from the repository, right ?

I really like Flowblade and the way it works. Thanks for all the efforts !

Bart.

Original comment by bart2...@gmail.com on 5 Dec 2012 at 12:44