EddieLarkin wrote:Look at the sky in
this comparison; the way the grain/noise just falls apart into an electronic mush is very noticeable to me during playback. That's soley down to poor encoding (both discs have the same transfer and very high bitrates).
EddieLarkin and Tenia: you are sort of both correct.
In the City Of The Living dead link, what you highlight here is tube telecine noise being "eaten alive" by compression. This high frequency noise is difficult to compress and very space-hungry. It looks a lot like whoever was producing the master applied sharpening to try and bring out the details, and in doing so of course also exaggerated the noise.
IMHO, the proper thing to do in these cases is to lowpass filter the image - rolling off the high frequencies. If you "look through" the telecine noise, you'll see that the film image below it is not hugely sharp anyway. Whether or not that's down to the film element itself (printed-in softness) or something else in the telecine system (out-of-focus CRT beam) is up for debate. Reducing the appearance of the machine noise in this way is something I advocate because it allows the noise (bad) to blend in better and pass for film grain (good). Using specialist denoising filters tend to operate temporally so any attempt to reduce the machine noise also reduces the grain and produces electronic sludge.
Tenia: I don't think that what you point out in the Once Upon A Time In America shots is telecine noise, but rather film grain. Well, what's left of it after the encoding, anyway.
But I think Arrow's poorer presentation might be coming to an attempt to denoise this FUBAR source in order to hide the defect. I'm not sure it's 100% down to the encode anyway.
I can't speak for Arrow because this was before my involvement with them. However as a compressionist I can tell you that the smoothed-over areas in the Caps-a-holic comparison is 100% down to compression. There are no signs of prefiltering in those grabs.
Compression issues and noise and/or grain reduction often get confused with each other, because both tend to operate by reducing the high frequency content of the image. In the case of compression, that's a function of the encoder trying to meet a target bitrate. In the case of grain reduction, for aesthetic effect. The easy way to tell if it's down to compression is to look at the structure of the area where the detail differs. You'll see there that there is an 8x8 block pattern to the areas being discussed.