Pages: [1]   Go Down

Author Topic: interesting discovery to set-up R3D up or downsampling  (Read 50067 times)

fredjeang2

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1376
interesting discovery to set-up R3D up or downsampling
« on: August 02, 2015, 09:56:36 am »

I've just made a discovery.
Not linked to Red only but didn't have time to check with other footage.

It's about the scaling filters used in the decode settings into the Scratch play. I ignore if those can also be parametered inside RCX as well.
That needs to be seen and if you have imputs on that, please post.
BUT...and this is an important "but",
The choice of the right-filter-for-the-right-job affects the results. Just like in stills.

In practise, it means that according to the filters (some are better for upsampling, others better for downsampling wit different smoothness outputs)
one can acheive the right amount of sharpness without using any sharpening tool.
It also means that you can use the correct filters in the case that you zoom into your R3D to reframe a shot and therefore minimize or control the right amount of sharpness or smoothness desired
without having to defocus.

In other words, they affect the decode when up or downsampling.

Here is the list you can apply:

linear
cubic
Sinc
Lanczos
Triangle
Box
Gaussian
Mitchell
Quadric
B-Spline
Mipmapping

Proper tests have to be done but there are infos in internet on how those filters act and for what they are good at or to avoid according to the cases.

The scratch play also allows to load those filters into the dual monitor itself and not only the saver so zooming on your right monitor will show you the effect.
 
Logged

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8913
Re: interesting discovery to set-up R3D up or downsampling
« Reply #1 on: August 02, 2015, 10:21:08 am »

The choice of the right-filter-for-the-right-job affects the results. Just like in stills.

Hi Fred,

The Mitchell filter is generally considered as pretty good, for both up- and down-sampling. Some of the others may do better on specific types of image content, but Mitchell is a good first choice. It does not produce too many artifacts, and attempts to avoid too blurry results with minimal haloing.

Cheers,
Bart
Logged
== If you do what you did, you'll get what you got. ==

fredjeang2

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1376
Re: interesting discovery to set-up R3D up or downsampling
« Reply #2 on: August 02, 2015, 10:56:35 am »

Hi Bart, nice to see you in this motion forum.

I agree. For the reasearches I've been doing so far, the Mitchell is high on my list as well.
You are right, a good first choice.

Except maybe in the case of a cropping into the red file (for ex to do a fake close up)?, the need might be to soften without loosing details
and I haven't had the time to run those testings in particular cases.

Found this link:  http://pixinsight.com/doc/docs/InterpolationAlgorithms/InterpolationAlgorithms.html

Cheers.
« Last Edit: August 02, 2015, 10:58:49 am by fredjeang2 »
Logged
Pages: [1]   Go Up