[Konvas] Film grain aliasing
colcam at aim.com
colcam at aim.com
Sun Jul 16 20:09:17 EDT 2017
Oh, and remember this is CMYK based.
From: colcam--- via Cinema <cinema at konvas.org>
Most of the problems with conversion from film to digital files are centered on the number of colors-- get that wrong, the picture looks lumpy and the colors look fine-- and it shows up in the number of steps in dynamic range. You may have to scan something for each color independently AND then scan for brightness of total combined colors AND wind up with a huge file that cannot be used except to produce 4k masters. I had a single scene of 5227 stock that had very bright, changing colors with backgrounds that were flat and had no fill, and it took a dozen scans per frame to get it up for 720p HD format. They just redid that footage for a new 4k master and it took over eighteen scans per frame, each color, one by one, plus scans for the brightest areas, scans for the darkest areas, and marriage scans to allow them to be combined.
The number of powers of enlargement play into it, and that's why the small frames require higher resolution, and you should only count the line pairs of the scanned area.
There are at least a dozen different issues, some based on colors, some on brightness, some on contrast, and to make it worse, the new digital files are different than anything before so the guys doing the work may be guessing, too.
From: Brandon Esten via Cinema <cinema at konvas.org>
Has anyone here ever come across this problem while scanning? It seems to pop up around ~2001 with early still frame scanners and I figured it was a solved problem; however, it seems to be alive and kicking! One of the claims in the forum I quoted is that for smaller film gauges you need a higher resolution scan... but that doesn't make sense. A certain scanner sensor pixel count per mm^2 that is high enough to get past aliasing should be a constant figure based on ISO (ASA) and not the physical dimensions of the celluloid or gate framing.
For a given sensor, a 4K scan of 4-perf 35mm film is apples and oranges to a 4K scan of Super-8? In fact, with a fixed optical pathway, a 35mm size sensor will be 'windowed' to scan Super-8 and thus, with the same pixel density should render the same 'resolution'?
There again... does ISO scale with film gauge? I mean, is the emulsion the same between 35mm 200ISO and Super-8 200ISO? I always assumed 'yes' and think I remember folks mentioning it... but... idunno?
Perhaps I'm getting tangled up in different but intertwined issues.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Cinema