Grain Upon Grain?

Peter Moretti · April 30, 2009

This is something I've been wondering about. During filming, the film itself adds grain to the picture. So even if a DI is made and projected digitally, film grain will show up.

But if a print is made from the DI, isn't another level grain added by the film print? If so, how does this second introduction of grain to image interact with the grain already there from the orginal filming? Is this something to be aware or is it a non-issue?

Thanks much and sorry for asking such a basic ?. Thanks again :).

Paul Bruening · April 30, 2009

Hello Peter,

It's a fine question with no easy answer. First, much of the original's grain doesn't make it into the DI. See, pixels are bigger than grains in varying amounts. A 2K DI captures few to none of the individual grains, even from chunkier-grained stocks. The pixels can only represent grains by providing slight color shifts within the same position from frame to frame. At 4K DI you can actually start differentiating the large-to-medium sized grains. So, basically, the grain survival of the original stock is determined by the grain level of the stock (speed, age, exposure, etc.), it's processing factors (push, pull, bad chemicals, etc.) and DI (capture method, resolution, compression, polishing and alterations). Take all of those variables and flip them over for the recording side (dig-to-film)of the game.

That all means: Who knows?

Andres Pardo aka Gral Treegan · June 16, 2009

Hi!

as paul said it depende on the scan resolution. i sacn to 6k s16mm and you can see the grain, the native grain.

we finally print in 35mm and the grain is the same for my eye. the point is that the print stock are really fine grain!!

bests!

GT

Scott Fritzshall · June 16, 2009

Print stock is really slow, I don't remember offhand but I think it's something like 2 ASA. The idea is that it's got a really fine grain structure that will be minimally apparent on top of the image's original grain. It really shouldn't be much different for film that has gone through a DI versus film that has been photochemically finished in this regard, other than any processing of the grain that may have gone on during the DI or VFX.

Michel Hafner · June 17, 2009

It's a fine question with no easy answer. First, much of the original's grain doesn't make it into the DI. See, pixels are bigger than grains in varying amounts. A 2K DI captures few to none of the individual grains, even from chunkier-grained stocks. The pixels can only represent grains by providing slight color shifts within the same position from frame to frame. At 4K DI you can actually start differentiating the large-to-medium sized grains.

So what are we seeing here apart from 'compression noise'?

Seventh Seal

Each pixel represents several different grain particles? How many?

Paul Bruening · June 17, 2009

Hello Michel,

It's not a fixed amount. Stock, processing, format, scan and any compression will affect the results.

There's a thing that happens commonly to scanned footage that is often mistaken as grain. I don't know if this is what you're asking about. But, I'll fling it out there. Imagine a digital pixel somewhere in the middle of the frame. In this example the actual optical image at that location is a bit of blue sky. Now, at that exact pixel location, different grain patterns will occur 24 times a second. Inside that second, that pixel will change color and brightness 24 times because of the changing grain patterns. When you see the whole frame flipping past at 24 fps (or 2:3'd in telecine) those value shifting pixels seem like the screen crawl of film grain. It's not, but the brain thinks it is. For many producers this approximation of grain is enough. So, 2K is the most common scan size since this pixel crawl is close enough to film grain for them.

If you want to get picky about grain, it could take a resolution somewhere in the neighborhood of 64K to capture even the tiniest grains of an average 35mm negative stock. You see, film stock is actually a pan-resolution medium. Grain sizes and clumpings go from large to tiny. The same goes for the even more vaguely defined and resolved dye clouds. Actually, more so for dye clouds which, ironically, are the larger part of what people are actually talking about when they talk about grain. Dye clouds are really "not-grain" grain.

Is that any use to you?

Michel Hafner · June 20, 2009

Is that any use to you?

Yes, I think so. What I'm not clear about is this distinction between real crawl of film grain and the fake one (with pixels changing intensity quickly).

While real grain particles come in all sizes and some are too small even for 4K scanning so 2K/4K sized pixels can not show all their individual behaviour what makes the grain crawl on film prints more real? The losses from copying and during projection are so massive that no very small grain particles in the negative will ever show up accurately on any screens. Whether it's a digital chain or an analogue chain we always see only average cumulative effects of many grain particles working together, and arguably with more clarity and detail on a 4K projector than on any 35mm print. ? You were not talking about grain aliasing, were you?

(http://www.photoscientia.co.uk/Grain.htm)

Paul Bruening · June 20, 2009

I had to read the link you provided to get your question. I didn't know that was the name for that phenomena. Yes, chunkier film grain makes for greater variation between pixels and within the same pixel location through time. In the larger picture, that makes the image seem grainier or "crawlier". Every scan house does whatever it feels is necessary to compensate for this... or not. I massage my scans pretty heavily. I assume most scan houses do too. But, I don't know that for sure. Nor, do I know they would admit it.

Michel Hafner · July 1, 2009

I had to read the link you provided to get your question. I didn't know that was the name for that phenomena. Yes, chunkier film grain makes for greater variation between pixels and within the same pixel location through time. In the larger picture, that makes the image seem grainier or "crawlier". Every scan house does whatever it feels is necessary to compensate for this... or not. I massage my scans pretty heavily. I assume most scan houses do too. But, I don't know that for sure. Nor, do I know they would admit it.

The only massaging I would want to be done at scanning is proper antialiasing. Which means a lot of oversampling. Conventional printing does just that, antialiasing by random sampling and random dither. And built in low pass filtering.