Log in

View Full Version : Two question-ideas


improser
4th September 2016, 07:25
Hello, long time reader here.

I have two ideas and Im curious if is there a way to script or even utilize the current available plugins to do this tasks:

1) In cinematography for movies the use of high grade cinema specific lenses were used. This lenses cost 10.000+ USD or more and have a great resolution power along great control of aberrations, and also, there are build with traditional modeling techniques, that develop a characteristic bokeh texture, if not soft almost organic. When you use modern lenses, molded like plastic or so in structures that copied the form, there also copy the concentric texture of the material, resulting in bokehs that have a unpleasant texture or not smooth at all, even sometimes called "".
A way to combat this particular asthetic defect is to blur/median or haze this shape areas, eliminating the texture on these.
It is possible to develop a script that recognizes circular/oval forms, make an automatic mask, and blur/median only this mask area? Then, maybe apply some texture/noise to blend better with the overall image.

2) Experimenting in resizing with NNEDI3 and others method, I come to a idea. Maybe is possible to blend a video with a picture in more resolution of the same shot to "enrich" nor blend more detail to a video? Let me explain better: for a video in FullHD of a general, detailed shot, I have a 16 megapixel still of the same shot, taked to the purpose, to in the resize process, blend altogheter to enrich better te video, in the zones there are no movement, and gain detail present in the still.
If the shot/camera moves, there need to be tracked maybe. One can build this stills in better resolution for that purpose especifically.



I have not all the needed understand in Avisynth, also english is not my native language so excuse me is not the best writing.

Thanks in advance.

improser
4th September 2016, 07:26
reserved to post pictures

junh1024
5th September 2016, 00:44
No scripts exist by they're probbaly somewhat doable.

1) It's a bit hard to understand, mind making it clearer/examples/pictures?

2) Your still shot would need to be exactly the same as the lowres video, except higher res. Then, it's up to mocomp to track the textures as they move, and then it'll become back to blurry until the next still shot.

Really, these things are of limited utility (or so i've herd). Just shoot in higher res & use better cameras to start with.