Log in

View Full Version : DVD Noise


MShoulerUK
29th January 2011, 15:36
I a few questions about the noise found on DVD Movies, as I'm not quite sure how it works and why anybody would use noise if it decreases the quality of the video.

1. Where does this noise come from?

2. If you purchase two copies of same the movie in a video store, will they both have the same noise?

3. If you neglect your dvd's does the noise increase?

Thanks for any detailed explanations, would like to get my head around this.

sneaker_ger
29th January 2011, 16:11
1.) From the cameras, the scanning process, put into it by the director and so on...
2.) Yes, because it's digital. Every copy of a movie is the exact copy of the master in the production facility. Every single bit and every single pixel will be 100% identical.
3.) No, because it's digital. While DVDs (especially if burned) can get corrupted, this will not be seen in increased noise, but instead by very visible errors. The playback might stop completely or the DVD won't start at all.

Sharktooth
7th February 2011, 04:14
also do not confuse noise with film grain (which is intended to be there).

Ghitulescu
7th February 2011, 10:06
Film grain is a marketing issue.
Removing the grain from a film is not always an easy task, and sometimes requires manual tuning. Relying on automatic "denoisers" can produce adverse results - which were equally marketed (see what happens if we remove the grain). Grain removal is a standard procedure in all professional photo labs - so one can print those huge posters without exhibiting the grain and without detail loss, of course.

So it's possible to do it. Why it's not always performed and marketed as "the director wanted to give a realistic perspective" lie? Because 90 minutes of film are some 130000 photograms. Hand processing those 130000 images, one by one, as the pro photographers do, would require some 50 to 200 years (or divided by the number of workers, with the possibility that various people may yield various results). And that costs. A lot. So better launch the "Kinofeeling" lie. It's cheaper and it would catch ... it did.

Also widespread is the lie that film grain is the detail. No, but it can support the detail. It is possible to have a whole photography on a single large argentic crystal (a single grain). The fine graining of a film was there for a different reason - a chemical one.

Coming back to the noise. The film grain is unique to each film roll, so it should stay in the same place if the same roll was used in various editions. The the film grain noise is common to each edition.

But there's another noise, the scanning noise, also known from the argentic photography. It can be removed, too, also like in the argentic photography. Usually seen in dark passages or in uniform coloured areas (cartoons, anime).

And there is the noise of the support itself, better seen in dark areas.

Summa summorum: all these noises can be removed. It takes a lot of processing and man-hours and, of course, money, but it can be done. How perfect is a second issue. In the end, the media content distributors would choose the most advantageous cost:effect ratio and launch the DVD/BD.

Sharktooth
8th February 2011, 14:25
i cant watch movies without grain. they look like soap operas...
grain is a substantial part of the watching experience. however, this is off-topic and should be better discussed in another thread.

Ghitulescu
8th February 2011, 14:42
i cant watch movies without grain. they look like soap operas...Because it's incorrectly done. Lack of time. Lack of money. Lack of interest. The approach pro photographer generally use would take up to 4 hours per image.
grain is a substantial part of the watching experience.That's also a reflex, once getting accommodated to grain-free (or generally noise-free images) you can't watch degraded movies any longer.

Amateur
8th February 2011, 14:49
there are a couple of denoisers that i like to add to all of my scripts because i don't really see any negative effects and they supposedly help with compression. pixiedust and undot. is there ever a time when those are a bad idea to use (i'm sure there are but i have yet to see it so i'm wondering if someone can give me a specific example)? should those be added to the beginning of my string of filters for the best effectiveness? which should be first, pixiedust or undot?

Sharktooth
8th February 2011, 16:01
undot is old and slow. use removegrain(mode=1) instead coz it does the same denoising and it's available with SSEx optimizations.

Amateur
8th February 2011, 19:38
oh ok that's good to know. what about my other questions though. is removegrain(mode=1) really risk free compressibility (since it's the same as undot) and therefor should be used on most/all of my scripts when encoding cartoons? what about pixiedust? and which order should i place them for maximum effectiveness? removegrain then pixiedust or pixiedust then removegrain...or does it not matter?

Sharktooth
9th February 2011, 02:32
removegrain(mode=1) is actualy the same as undot(). im not an expert in denoising/degraining as in my encodings i tend to preserve everything that's in the source (except interlacing of it needs some kind of restoration).
however try asking in the avisynth forum. there is more competent ppl there.