Hi,
I am trying to process a png image sequence.
The problem I am having is that some images result in a different outcome - as if they are being processed slightly differently.
The result is that a section of the image (the top right) appears to flicker when played back. The problem image has less black in the top right and it doesn't seem to match the source image ie it seems to be making stuff up?
This happens with more than 1 operation type but I am mainly concerned with image operation-> adaptive filter 2 which I used for the example below.
This is a source image of a good item:
and its result:
And this is the next item in the sequence (the one causing the flicker):
and its result:
I can see this happening when processing the images manually ( as I did with the examples above)
I have tried editing the settings but the inconsistency remains.
I not sure if the problem is with the source image, studio artist, or just the settings I'm using or something else?
Could the alpha in the source png be causing a problem?
As stated in the title, I'm fairly new to studio artist (I'm using version 4.05 on Windows 8.1 64 bit) so I may have missed something?
Any help appreciated
Thanks
Replies
The adaptive filter ip op effect is extremely configurable. So depending on the settings you are working with, it can be doing all kinds of different adaptive filtering effects. Many of which are non-linear. And depending on the Modulation option, the filter locally self adjusts (adapts) based on that specified modulation signal.
That effect can also be configured so that you might be processing the source image, but be modulating the adaptation based on attributes of the canvas. If that's the case, then what is in the canvas is going to greatly affect the outcome of the processing for a given frame. So without looking at the specific preset or the control parameters for what you are working with, it's hard to speculate.
So the ip op effect is not 'making stuff up'. It's doing whatever you have it configured to do, based on it's parameter settings. The effect is not working with alpha, so i doubt that alpha in the png images is a factor, at least as far as the processing is concerned.
If you want to attach a preset, or a screen shot of the parameter settings, i can comment further on what the adaptive filter ip op is configured to be doing in your particular situation.
Flicker associated with movie processing is caused by temporal variations in sequential frames. Some ip op effects generate a lot of temporal variation, others do not. So you have to experiment with what works or doesn't work for movie processing. Effects that work well for processing single images may not necessarily translate to processing movies without introducing some flicker.
Depending on what you are doing with the adaptive filter effect, perhaps using something like the Smooth ip op on the input signal to the adaptive filter might help. Another thing some people do is frame averaging as a way to reduce flicker with some processing effects.
There has been a lot of discussion concerning flicker reduction over the years here on the user forum. So you might want to do some searching for previous conversations associated with that topic. I quickly did a quick search to provide a few links below to some previous discussions about flicker.
Here's one post.
Here's another post.
Here's another post.
Here's some discussion concerning different paint animation strategies.
Thanks for your reply John.
I am just using the default setting 'AdaptiveFilter2' and that causes the issue. When processing in a PAS I have just 2 steps
1-Set Canvas -White and
2-AdaptiveFilter2 - with the _Ip source set to 'Current Layer'
I have since converted the png sequence in to a mov file and when that was processed the issue did not occur.
By the sound of what you have said, there is obviously a lot for me to learn about this but it does seem odd that the presumably lower quality images in the mov file remove the problem - does that sound possible?
I will have a play with the smooth operator and look up frame averaging as these sound like useful things to know.
Thanks for your help
You are adaptive modulating based on inverse source texture, so if the movie frames are compressed, then the derived source texture might be somewhat different than the original png image frames (which would have more fine detail, hence a more fine detail texture field associated with them). And you are modulating the orientation and the size of the adaptive filter based on that derived source texture (actually the inverse of it), so local changes in the source texture will cause local modulation of the filters orientation as well as it's size.