Mermouy/flowblade

logic of alpha filtering confusing

Opened this issue · 1 comments

GoogleCodeExporter commented 
When using a compositor (dissolve, ...) the alpha opacity can be set to 
over/and/or/xor. The result when using 'over' is quit clear. But with the other 
values, it seems to me that something is going wrong:
- when alpha in the source varies from 99-1%, starting from 49% the result is 
reverted in some way (49% becomes less transparent than 1%)
- Using an opacity of 99% may give a significant different result than using 
100% opacity
- Using an alpha value of 99% in the source may give a significant different 
result than using 100%

These strange results do occur in version 12 and 14. I don't know if it's 
Flowblade mixing up or if it is MLT. You can find the file I used for testing 
in the attachment. It's rather difficult to catch the logic behind what is 
happening.

I wanted to create a mask that adds transparency to an underlaying layer. By 
trial and error I got to the following solution (see also last sequence in 
attached test project):
V3: mask
  --dissolve(xor,opacity 99)-->
V2:layer
  --dissolve(xor,opacity 100)-->
V1:background

Anyway, the effects that can be created by mixing opacity are really great. If 
I only could understand the workings behind ...

Bart.

Original issue reported on code.google.com by bart2...@gmail.com on 5 Aug 2014 at 2:54

Attachments:

GoogleCodeExporter commented 
I downloaded the files and I was not able to create the artifacts with the 
"testmask1/2/3.png" files and "Dissolve" compositor.

a) There was no jump in result going from opacity 100% -> 99%
b) Result changed linearly in one direction over 49% opacity

Flowblade currently does not support relative paths so I'm was able to open the 
.flb file, there was also a custom Project Profile used that I obviously do not 
have.

I'm leaving this open in case there is more information on issue forthcoming.

Original comment by janne.li...@gmail.com on 26 Aug 2014 at 11:50

  • Changed state: NeedsMoreInfo