From: Helsinki, Finland
Registered: Dec 2017
posted February 03, 2019 10:26 PM
Personally, I love the “grain” of 8mm film and, consequently, consider cleaned-up footage inferior. (Unless the grainness / color noise is far too prevalent; for example, the Wittnerchrome 200D has (had) pretty awful and, what is worse, colored grains – see for example https://www.youtube.com/watch?v=9pUlIeoT8AA ) This is why I personally don’t reduce video noise before uploading my 8mm scans.
As anyone that use YouTube for distributing 8mm scanned material knows, YouTube applies recompression, which also results in (at least) luminance noise reduction, I made some direct original-vs-YouTube comparisons to find out how much damage this process results in when you upload non-noise-reduced footage showing all the original grain in the film. All my scans are done with low sharpness to minimize in-scanner post-processing.
Case 1: one of my latest shots (May 2018, Helsinki) of the Finnish Defence Forces band members, including their leader, Tommi Suutarinen, leaving the Church of St. Lawrence
(Incidentally, if you like Apocalyptica, the same band also recited Bittersweet applied to classical instruments. Please see https://www.youtube.com/watch?v=rnVBN4ahLrg&t=2827s (that is, from 47:07).)
The scanned DS8 Fomapan R100 film is at https://www.youtube.com/watch?v=RlEjkBijD5Q&t=208s .
Here’s the original (offline) scanned frame: https://www.flickr.com/photos/33448355@N07/32032477437/in/album-72157706336891755/
And here’s YouTube’s rendition (more than a day after upload): https://www.flickr.com/photos/33448355@N07/32032477167/in/album-72157706336891755/
As one can expect, there’s only difference in the low-contrast, larger areas like the white stone on the right. I couldn’t spot noise reduction resulting in real detail smearing effects elsewhere.
Case 2: another (now, color) example using Kodachrome input from one of my recent Vietnam Military Police scans from around Christmas of 1969 (link: https://www.youtube.com/watch?v=jXD1UQIGZtU ). It’s a bit overexposed; hence the washed-out colors.
I’ve chosen this frame as it has low-contrast(!) writing (the “Remove hat...” board on the left) suitable to spot the adverse (detail-smearing) effects of noise reduction and a lot of homogenous areas where the grain is certainly visible.
A framegrab of the original scan: https://www.flickr.com/photos/33448355@N07/40013052593/in/album-72157706336891755/
And a screenshot of YouTube’s rendition: https://www.flickr.com/photos/33448355@N07/40013052413/in/album-72157706336891755/
The legibility of the text isn’t really affected by YouTube’s recompression, fortunately. (And, again, this is the worst-case scenario with low-contast input to begin with.)
All in all, while YouTube does recompress & consequently reduce noise, it’s not as bad as I’ve originally thought.
| IP: Logged