Grain Reduction and Blu Ray transfers

[Steve Zimmerman]

Looks like there needs to be some additional education of the masses about film grain, just like letterboxing in the past. Some uninformed comments mixed in here. If grain is in the original film, it should be there on the High Def media image.

 

http://gizmodo.com/392663/hollywood-attack...ain-for-blu+ray

Edited May 22, 2008 by Steve Zimmerman

[Satsuki Murashige]

Looks like there needs to be some additional education of the masses about film grain, just like letterboxing in the past. Some uninformed comments mixed in here. If grain is in the original film, it should be there on the High Def media image.

My dad recently got a Blu-Ray player and a 50" Panasonic 1080P plasma tv so I've had the chance to watch a lot of Blu-Ray discs. Most of the visible "grain" in newer films like "Casino Royale", "No Country for Old Men", and "Spiderman 3" is actually video noise, probably from the disc compression. For these films, I'd imagine that the digital master from the DI was used as the source element to create the Blu-Ray version, so there should be even less visible film grain on Blu-Ray than in a theatrical print, where grain from the camera original is overlayed with grain from the emulsions used in the various printing stages. I'm not sure you'd even see any film grain at 1080P resolution unless it was quite pronounced in the negative to begin with (think "Eyes Wide Shut", the B&W intro of "Casino Royale", "Munich", etc.).

 

Some discs made from older telecine transfers like "The Road Warrior" do actually exhibit film grain, although much less than a theatrical print (though the print that I saw was an old one in poor condition). Dean Semler admits in the commentary track that he underexposed one of the night exteriors too much, resulting in an extremely thin negative that had to be printed up, so no surprise that there's visible grain in that scene. In general, I'd say that these older films exhibit more grain because: film stocks were grainier back then (especially fast stocks from the 80's), the original source material used in the transfer (interpositives with optical effects in many cases) had more grain in them because they had gone through more printing stages and were sometimes damaged from age and abuse, and the amount of noise reduction/grain reduction used in the transfer was much less than would be used on a modern feature going through a DI for theatrical release today. So it's a combination of factors.

[Antti Näyhä]

Most of the visible "grain" in newer films like "Casino Royale", "No Country for Old Men", and "Spiderman 3" is actually video noise, probably from the disc compression.

I would say it's film grain that has been messed up by filtering and compression. Video compression alone doesn't create texture like that on "clean" footage.

 

I'm not sure you'd even see any film grain at 1080P resolution unless it was quite pronounced in the negative to begin with

Sure you would. You can clearly see film grain in a 2K DCI digital projection, and the resolution of that is very close to 1080P. The difference is that DCI is much less compressed, so the grain looks more natural and less like video noise.

[Michel Hafner]

Most of the visible "grain" in newer films like "Casino Royale", "No Country for Old Men", and "Spiderman 3" is actually video noise, probably from the disc compression.

How do you know? How do you tell real grain apart from video noise without access to the uncompressed master?

I'm not sure you'd even see any film grain at 1080P resolution unless it was quite pronounced in the negative to begin with (think "Eyes Wide Shut", the B&W intro of "Casino Royale", "Munich", etc.).

Of course you see film grain in 1080p as you see in 2K. Just not all of it.

[Satsuki Murashige]

How do you know? How do you tell real grain apart from video noise without access to the uncompressed master?

I don't know for sure, but it certainly looks like video noise and not film grain to me. So, perhaps I should have said, "IMO, it looks like..." ;)

 

Film grain is most visible in midtones and is much less visible in highlights because the the negative there is denser and the smaller less sensitive grains are exposed, filling in the gaps between the larger, more sensitive grains. But the "noise" that I noticed appears in both midtones and highlights uniformly. It also doesn't swirl randomly as film grain does but tends to "shimmer" in place. That's an unscientific description to be sure, but I'm pretty familiar with what video noise looks like (or at least, with in-camera generated noise, if not video compression noise), and well, that's what it looks like to me. I know very little about video compression so Antti may be right that it is a combination of both grain and noise.

 

Of course you see film grain in 1080p as you see in 2K. Just not all of it.

Ok, well then what about all that talk on this forum about how 2K DIs often appear less grainy and thus less like film-originated material than 4K DIs because the 2K scan doesn't resolve all the grain in the 35mm negative to begin with? I mean, naturally you're seeing "some" film grain in 1080p, the exposed grain that makes up the image (wasn't sure if that's what you meant by "of course", if not then please disregard this sentence). But what about the unexposed grain in the image that I unscientifically referred to as simply "film grain"? My supposition is that on a relatively high-key scene, shot on well-exposed, normally-processed, modern 35mm color neg (as much of "Spiderman 3" was shot) the amount of visible "film grain" would hardly noticeable in 1080p. If so, then what I'm seeing must be video noise. Am I wrong about that? :unsure: I realize that film-originated material that is grainer to begin with (like B&W neg, Super 16, pushed stock, underexposed stock, skip bleached stock, x-processed stock, older stock, etc., etc.) is another story.

 

I specifically recall David Mullen referring to the "grain" which somebody had pointed out in a 1080p apple trailer for "Astronaut Farmer", saying "that's not grain, that's noise." That's very similar what I think is going on with a lot of Blu-Ray discs. BTW, Isn't the mpeg4 compression done for those apple HD trailers quite similar to compression done for Blu-Ray discs (mpeg2 and mpeg4)?

[Will Earl]

I haven't seen that much bluray stuff yet, so I can't really comment on the quality of the format. But I've seen enough film scans to know that film grain will show up on 2k scans, even the well-exposed Vision2 (5217) that I'm looking at right now the grain is noticeable on the original scan.

 

The comments in that blog posting are some of the sillyess I've seen in awhile.

[Michel Hafner]

Film grain is most visible in midtones and is much less visible in highlights because the the negative there is denser and the smaller less sensitive grains are exposed, filling in the gaps between the larger, more sensitive grains. But the "noise" that I noticed appears in both midtones and highlights uniformly. It also doesn't swirl randomly as film grain does but tends to "shimmer" in place. That's an unscientific description to be sure, but I'm pretty familiar with what video noise looks like (or at least, with in-camera generated noise, if not video compression noise), and well, that's what it looks like to me. I know very little about video compression so Antti may be right that it is a combination of both grain and noise.

Compressed 1080p will have film grain and some compression noise. How much of each I would not dare to guess for specific examples witout seeing the uncompressed version. But there should not be a lot of compression noise on well compressed high bit rate AVC or VC-1.

Ok, well then what about all that talk on this forum about how 2K DIs often appear less grainy and thus less like film-originated material than 4K DIs because the 2K scan doesn't resolve all the grain in the 35mm negative to begin with?

As I said, you see grain in 1080p and 2K, just not all of it. There is more in 4K.

I specifically recall David Mullen referring to the "grain" which somebody had pointed out in a 1080p apple trailer for "Astronaut Farmer", saying "that's not grain, that's noise." That's very similar what I think is going on with a lot of Blu-Ray discs. BTW, Isn't the mpeg4 compression done for those apple HD trailers quite similar to compression done for Blu-Ray discs (mpeg2 and mpeg4)?

It's not the same although both often use AVC, Trailers are not hand tuned and use lower (+-constant) bit rates. BDs are hand tuned and use variable bit rate and higher averages with peaks reaching > 40 MBits/s at times.

Edited May 26, 2008 by Michel Hafner

[Tom Lowe]

Nearly all the 1080p Blurays I have seen exibit a lot of grain, and it can be distracting when you are viewing the bluray pixel for pixel, on an LCD screen close up. Newer movies like The Matrix have less grain and look much "better" and cleaner to me. Some people might like the look of grain, and I understand that. But most people are not DPs who care about things like "grain"... they just want a clean, beautiful moving picture. 1980s movies suffer terribly from grain. Hopefully new transfers can be done.

 

BTW, far and away the "cleanest" and most beautiful 1080p movie I have seen was an HDTV rip of Revenge of the Sith. So damn clean, such amazing color.

[Michel Hafner]

Nearly all the 1080p Blurays I have seen exibit a lot of grain, and it can be distracting when you are viewing the bluray pixel for pixel, on an LCD screen close up. Newer movies like The Matrix have less grain and look much "better" and cleaner to me. Some people might like the look of grain, and I understand that. But most people are not DPs who care about things like "grain"... they just want a clean, beautiful moving picture. 1980s movies suffer terribly from grain. Hopefully new transfers can be done.

I find watching anything on most LCD monitors distracting. Badly implemented display technology (in the consumer section, professional monitors are better).

And I completely reject the call for DNR on HD transfers as soon as it goes against the authentic look of a film. I want to see films as they were made, not digitally mutilated versions with no more HF detail, waxy skin and worse. People who dislike grain can use the noise reduction in their display chain (and reduce the sharpening and brightness to correct levels) and/or stick to originally low grain/noise material (film or HD camera).

[Tom Lowe]

BTW, I have recently seen two IMAX movies on Bluray and they look FANTASTIC. No grain, no noise, amazing dynamic range. Just beautiful.

[John Holland]

Well if you cant a good image on Blu Ray from a IMAX film then well what to say !!! .

[Steve Zimmerman]

Film restoration expert Robert A. Harris has a great article on the subject

 

http://www.thedigitalbits.com/articles/rob...rris062408.html

[A. Whitehouse]

Nearly all the 1080p Blurays I have seen exibit a lot of grain, and it can be distracting when you are viewing the bluray pixel for pixel, on an LCD screen close up. Newer movies like The Matrix have less grain and look much "better" and cleaner to me.

 

50% of the "grain" I see on Blu-rays (or hd-dvds) is better described as noise. It appears in parts of the image that I wouldn't normally associate with being a target for aggressive grain. Its also sometimes limited to one channel. I wonder if it has something to do with the codec used on the disc (are there three different options now?).

Anecdotally, I was at a home theatre store last year talking to a salesman about bravias and bluray, who incidentally tried to sell me on the motion flow technology, and he was talking about the sharpness of a bluray disc and its ability to see grain in a film image. I didn't say anything but the Bluray demo disc he was using as an example had select scenes from "Click" and some Robin Williams RV camping movie, both of which were shot on the Genesis.

The comments in the thread linked above show just how incredibly ill informed most of the public is and is likely to remain.

[Robert Houllahan]

The comments in the thread linked above show just how incredibly ill informed most of the public is and is likely to remain.

 

 

I was at a small specialty Audio / Video store the other day a friend of mine manages, they sell allot of hi end HD sets. Stu said that they cannot even get Blu-Ray players from Sony as they have some kind of product "sell through" policy and have somewhat run out of players until new models come out. Ironic that they win the little skirmish of the consumer electronics world and then cannot come up with the goods.... :unsure:

 

 

HD looks great when you are sitting in the DSX or Spirit room looking at the D32 and just seems to get ground into crap every step after that. Some Blu disks look ok to me if you don't look too hard and much of HD broadcast is just sad. I know allot of people who have no connection to the film business (other than watching them) who are really dissapointed with Hd for everything except for sports.

 

-Rob-

[Paul Bruening]

Hello Robert,

 

Isn't this kind of backlash interesting. I'm running into what you've mentioned as well. It seems like people get awed by the new technologies right away but, given enough time, make decisions about it that was not the same as when they started. Some of the folkes that I hang out with at the local bar say they don't like their new HDTVs. They bragged about them at first. But, something occurred to them as they had time to think about the images they're getting. When they try to tell me about it, it's in non-technical and non-artistic terms. They don't quite know what's wrong with the image. They just know they don't like it as much as they did. I still don't actually know what their beef is.

 

Aren't we hearing a little of that at the professional DP level? Sure some DPs are in love with digital. But, rumblings are coming out of Mt. Olympus that sound like dissatisfaction. It's no particular vindication for film. It's just interesting.

[Jonathan Bruno]

Satsuki,

 

Not all films that go through a DI have a heavy amount of grain reduction, so you can't base your assumptions on whether or not it went through a DI. In fact, most DIs default to using absolutely no grain reduction. However, I will agree with you that we can't be sure that what we're seeing is necessarily film grain, although the compression settings are blu-rays are pretty damn good.

 

I would say we should all just be happy that this technology exists and that HD DVD is gone. It's silly to fight over issues of grain.

[georg lamshöft]

Grain with current film-stocks at a effective scan-resolution of <2000ppi?

 

Well, I'm not too familiar with current film stocks but this is the result when you scan a current fine-grained 100ASA-slide-film, it's a Imacon/Hasselblad X1 using a 8kpixel-Kodak-CCD-line-sensor which is cooled by peltier-elements. This scanner is nearly (not entirely!) noise-free and translates into the DI-world with a resolution of 8k. The scan/slide itslef isn't perfect (it could be sharper).

 

First JPG-100%-crops of the original 480MB-16Bit-TIFF-file, one without any sharpening or noise reduction, the second one with little sharpening and noise reduction:

 

And downsampled with PS (bicubic) to HD-resolution (~2000ppi):

 

Where is the noise? Are Vision2/3-film-stocks grainier even below 200ASA? I can see the difference between color-noise and grain and MANY Blu-Ray-transfers have heavy color-noise, grain-alaising and other artificats propably not even close to the quality of the original!?

Edited July 25, 2008 by georg lamshöft

[Paul Bruening]

Hello Georg,

 

I have been doing a little research on this Hasselblad, Imacon unit. I have been unable to determine which dimension the 8K follows. Does it follow the 8-perf dimension (long dimension of normal SLR frame) or the film's width (from perf to perf, SLR short dimension)? Will it transport and frame-recognize 35mm cine frames: 4, 3, 2-perf? Does it have some form of registration like precise transport, pin registration or digital registration?

[K Borowski]

Hello Georg,

 

I have been doing a little research on this Hasselblad, Imacon unit. I have been unable to determine which dimension the 8K follows. Does it follow the 8-perf dimension (long dimension of normal SLR frame) or the film's width (from perf to perf, SLR short dimension)? Will it transport and frame-recognize 35mm cine frames: 4, 3, 2-perf? Does it have some form of registration like precise transport, pin registration or digital registration?

 

No Paul, unless the design has changed, it only takes cut film, so it'd have to be heavily modified if that is even possible, to deal with cine film.

 

Same goes for drum scanners.

[georg lamshöft]

The Imacon-Scanners (Hasselblad is just the brand) have a moving Rodenstock lens which enlarges the film from 24mm width up to 9cm so the whole area of the CCD-sensor is covered and the scan-results become ~8kpixel wide (which results in 8000ppi for 35mm film, 3200ppi for medium format and 2000ppi in large format).

It would be really interesting to build an adapter which takes film rolls!? Although the scanner isn't pin-registered and it's not very fast.

 

I just thought it would be interesting to "imitate" a DI-scan with up 8k resolution.

 

Why are so many DIs/transfers so bad? The telecine seems to be like a flatbed-scanner in the still-photography-world: horrible results with good technical data!?

Aren't these people specialists, working in highly specialized companies, getting hundreds of dollars per hour? Can't they use a good scanner (Northlight/Arriscan) to make a good transfer? Is it soo difficult? They did a new transfer for "Blade Runner" or "2001" with great quality but "Juno" (one of the best-selling movies (and DVDs/Blu-Rays), best-picture-nominated) is gets such a bad transfer that people think a simple HD-camera is better for Blu-Ray? I simply don't get it! That's just like the photographers who bought a Epson-flatbed-scanner, scanned their Hasselblad-slides, compared them to their 35mm-DSLR-results and threw away their Hasselblad!