Celine2020
Member
Reading a report on the Mr. Magoo's Christmas Carol disc and one detail jumped out more than the usual grain-scrubbing complaints. During the All Alone In The World sequence, the words and images written on the chalkboard reportedly disappear entirely while the camera pans across it. I do not have the disc in hand, so take that as one viewer's account, but if it is accurate that is not a taste question. That is artwork leaving the frame.
Denoising hand-drawn animation is a different kind of damage from denoising live action. On film you are mostly arguing about grain structure and how much of it you want left. On cel animation, the pencil work, the paint edges and the low-contrast detail sit at the same spatial frequency the filter is hunting for, so anything aggressive cannot tell a chalk line from noise. Add a pan and it gets worse, because temporal filters average across frames and lose their tracking the moment the whole image moves. That is precisely when faint detail gets smeared into whatever is behind it, which matches the symptom being described.
What frustrates me is that this stuff is usually not a scanning problem, it is a finishing decision made by someone chasing a clean-looking still frame for a store shelf. A fresh scan with the grain left alone and a light manual pass on actual damage would look better and cost less than the processing that ruined it.
For anyone who has the disc already, does the chalkboard hold up if you pause on it and only fall apart in motion? That would tell you whether this is temporal filtering or something baked in further upstream.
Denoising hand-drawn animation is a different kind of damage from denoising live action. On film you are mostly arguing about grain structure and how much of it you want left. On cel animation, the pencil work, the paint edges and the low-contrast detail sit at the same spatial frequency the filter is hunting for, so anything aggressive cannot tell a chalk line from noise. Add a pan and it gets worse, because temporal filters average across frames and lose their tracking the moment the whole image moves. That is precisely when faint detail gets smeared into whatever is behind it, which matches the symptom being described.
What frustrates me is that this stuff is usually not a scanning problem, it is a finishing decision made by someone chasing a clean-looking still frame for a store shelf. A fresh scan with the grain left alone and a light manual pass on actual damage would look better and cost less than the processing that ruined it.
For anyone who has the disc already, does the chalkboard hold up if you pause on it and only fall apart in motion? That would tell you whether this is temporal filtering or something baked in further upstream.
