Jump to content


Photo

Borrowing real grain?


  • Please log in to reply
10 replies to this topic

#1 Paul Bruening

Paul Bruening

    (deceased)

  • Sustaining Members
  • 2858 posts
  • Producer
  • Oxford, Mississippi

Posted 09 January 2009 - 09:43 PM

Digital grain doesn't always impress people. Has anyone tried overlaying scans of real grain patterns from 35mm onto video? Seems like 48 frames of overexposed, V2 from a white or gray card could be cycled undetectably through the video frames (two of the same per each field kind of thing) with a transparency render or the like. What do you think? Has it been done to death already? Does it suck?
  • 0

#2 Scott Fritzshall

Scott Fritzshall
  • Basic Members
  • PipPipPipPip
  • 584 posts
  • Other
  • Los Angeles

Posted 09 January 2009 - 09:59 PM

Yeah, you've got to spend some time setting it up, but there are features in Nuke and some other programs to do this. I haven't used it, but I think one of the problems is that- at least with compositing- it's not nearly as flexible as synthesized grain, so if it doesn't already match perfectly, for whatever reason, there isn't as much you can do to it.
  • 0

#3 John Brawley

John Brawley
  • Sustaining Members
  • 834 posts
  • Cinematographer
  • Atlanta Georgia

Posted 10 January 2009 - 01:24 AM

Digital grain doesn't always impress people. Has anyone tried overlaying scans of real grain patterns from 35mm onto video? Seems like 48 frames of overexposed, V2 from a white or gray card could be cycled undetectably through the video frames (two of the same per each field kind of thing) with a transparency render or the like. What do you think? Has it been done to death already? Does it suck?



It tends to look fake I think because grain size on film varies with exposure. If you composite a flat field of grain it looks exactly like that, a *film* of grain laid over the top of an image. It's not integrated in a way that we're used to seeing. We're used to seeing grain size change with the texture and exposure of the actual objects contained within the scene.

jb
  • 0

#4 Paul Bruening

Paul Bruening

    (deceased)

  • Sustaining Members
  • 2858 posts
  • Producer
  • Oxford, Mississippi

Posted 10 January 2009 - 01:48 AM

I had some odd ideas about up-ressing 960 wide video from my XL2 to, say, 4K. Then overlaying some real grain patterns into it. Then, knocking the image back down to 1,920. I guess the idea is to impose grain into the low res pixel areas to break them up. Then doing the down-res math to further the randomness of the grain patterns into the resolution math and image. I guess I'm thinking that imposing grain might give a fine degree of randomization to the higher pixel count and yield a sense of variation to the pixels of the down-res'd image. Maybe it could give some sense of grain as well as break-up the big blank areas of those original pixel blocks. It's probably a stupid idea. I may try it anyway.
  • 0

#5 Scott Fritzshall

Scott Fritzshall
  • Basic Members
  • PipPipPipPip
  • 584 posts
  • Other
  • Los Angeles

Posted 10 January 2009 - 04:05 AM

I just tried that, and it looks like it can potentially give you a slightly sharper-looking image, obviously at the price of giving you grain on top of your existing video noise. It's pretty quick to throw together an experiment, so I guess it's worth a shot.
  • 0

#6 Paul Bruening

Paul Bruening

    (deceased)

  • Sustaining Members
  • 2858 posts
  • Producer
  • Oxford, Mississippi

Posted 10 January 2009 - 10:20 AM

I just tried that, and it looks like it can potentially give you a slightly sharper-looking image, obviously at the price of giving you grain on top of your existing video noise. It's pretty quick to throw together an experiment, so I guess it's worth a shot.


I got the idea from a recent Seurat thread. The problem with up-ressing is those enormous blank pixels. 960 to 1,920 gives you 2 high by 2 wide pixels of the same color. 960 to 3840 makes 4 pixels high by 4 pixels wide blocks of chunky goodness. At 4K film grain becomes distinguishable. Put the two together and you might just trick the eye and brain that the detail is really there, somehow. Add to that the possible aesthetic desirability of film grain... well, maybe it could work. Finally, the down-res math might promote the randomness of film grain inside those big blocks to replicate an actual high res film image thats been down-res'd, to the point that it might trick the eye a step further. Add, finally, to the idea that at 24 or 30 fps that new randomness imposed by the grain and down-ressing will blend in the brain and further the illusion of sharpness as a motion illusion. I don't know yet. What do you think, Scott? Maybe add the fractal engine thing first to sharpen up the edges then run this grain switcharoo trick?
  • 0

#7 K Borowski

K Borowski
  • Basic Members
  • PipPipPipPip
  • 3905 posts
  • Camera Operator
  • I.A.T.S.E. Local # 600 Eastern Region

Posted 10 January 2009 - 10:30 AM

Why would you take the worst part about film, the grain and add it to video? If one is using video, I think it is best to embrace and work with the characteristics of that medium.
  • 0

#8 Paul Bruening

Paul Bruening

    (deceased)

  • Sustaining Members
  • 2858 posts
  • Producer
  • Oxford, Mississippi

Posted 10 January 2009 - 10:38 AM

Grain is a major reason why film is film; why it is so beautiful. It's the dance of grain at 24 fps that gives life and vibrancy to the screen. Video is flat, sterile and lifeless in comparison. But, in this case, I'm just trying to steal some of the characteristics of film to fake a sharper image out of my crap-res XL2.
  • 0

#9 Tom Hepburn

Tom Hepburn
  • Basic Members
  • PipPipPipPip
  • 341 posts
  • Other
  • Chicago-land

Posted 10 January 2009 - 01:13 PM

I know what you're saying Paul. I have in the past also tried to emulate some of the characteristics of film on video.
I've even used plugins designed to look like film......I'm so ashamed :( I was trying to save money in my case, by not shooting real film while hoping to obtain some of the quality. I think that says something about the quality of film that many shooting video strive to get that look. I have never even come close (as you would expect).

Very slightly off topic as I'm talking about 16mm film, I just shot a little short, with Tri X(7222) but I ran out as my kids could not get it in 4 takes or less!!! Anyway, all I had in the fridge was Plus x. So I shot 10 seconds of Tri X on foamcore with the flattest lighting I could get in hopes of getting a shot with just grain on white. If the cuts that go from TriX to PlusX are too jarring because of the different grain structure, I'm going to try and use a blending mode to overlay the grain. I'll post some still if I need to resort to that.

Tom
  • 0

#10 Scott Fritzshall

Scott Fritzshall
  • Basic Members
  • PipPipPipPip
  • 584 posts
  • Other
  • Los Angeles

Posted 10 January 2009 - 08:53 PM

I got the idea from a recent Seurat thread. The problem with up-ressing is those enormous blank pixels. 960 to 1,920 gives you 2 high by 2 wide pixels of the same color. 960 to 3840 makes 4 pixels high by 4 pixels wide blocks of chunky goodness. At 4K film grain becomes distinguishable. Put the two together and you might just trick the eye and brain that the detail is really there, somehow. Add to that the possible aesthetic desirability of film grain... well, maybe it could work. Finally, the down-res math might promote the randomness of film grain inside those big blocks to replicate an actual high res film image thats been down-res'd, to the point that it might trick the eye a step further. Add, finally, to the idea that at 24 or 30 fps that new randomness imposed by the grain and down-ressing will blend in the brain and further the illusion of sharpness as a motion illusion. I don't know yet. What do you think, Scott? Maybe add the fractal engine thing first to sharpen up the edges then run this grain switcharoo trick?

Up-resing isn't necessarily quite so simple as doubling or quadrupling a single pixel. There are at least half a dozen different sampling algorithms, all of which give slightly different results, and give the image different qualities of sharpness. There are a lot of things potentially at play here, and I think that basically adding grain at high resolution gives the downsampling different stuff to play with, and breaks up the image slightly, resulting in more sharpness. Or maybe it's just a perceptual thing; I don't know. There are a ton of things you could play with, and I'm sure that the people who up-res SD video for film or film for IMAX have a bunch of other things they do.

Yeah, you've got to spend some time setting it up, but there are features in Nuke and some other programs to do this. I haven't used it, but I think one of the problems is that- at least with compositing- it's not nearly as flexible as synthesized grain, so if it doesn't already match perfectly, for whatever reason, there isn't as much you can do to it.

That's a pretty subjective thing; a lot of people quite like it.

Very slightly off topic as I'm talking about 16mm film, I just shot a little short, with Tri X(7222) but I ran out as my kids could not get it in 4 takes or less!!! Anyway, all I had in the fridge was Plus x. So I shot 10 seconds of Tri X on foamcore with the flattest lighting I could get in hopes of getting a shot with just grain on white. If the cuts that go from TriX to PlusX are too jarring because of the different grain structure, I'm going to try and use a blending mode to overlay the grain. I'll post some still if I need to resort to that.

You've got to be very careful, because it sounds like you're just going to overlay the grain from one type of film over another, which will mean you'll have the grain from both, and that obviously won't look right. You would need to degrain the Plus X first, which frankly will be difficult on 16mm because it's got so much grain. Doesn't mean you shouldn't try, though.
  • 0

#11 Crell Lee

Crell Lee

    New

  • Basic Members
  • Pip
  • 9 posts
  • Student

Posted 05 April 2009 - 10:26 AM

i want to do some research about control of digital noise and film grain, i think first we might clear vedio noise and then add film grain,and in DI how did we control noise,when i change color or other think? do you know sth about it email me thanks ! e-mail:9393573@qq.com
  • 0


FJS International, LLC

Willys Widgets

CineTape

Ritter Battery

rebotnix Technologies

Wooden Camera

Opal

Tai Audio

Aerial Filmworks

The Slider

Rig Wheels Passport

Technodolly

Media Blackout - Custom Cables and AKS

CineLab

Abel Cine

Paralinx LLC

Broadcast Solutions Inc

Gamma Ray Digital Inc

Visual Products

Glidecam

Metropolis Post

Metropolis Post

Paralinx LLC

Gamma Ray Digital Inc

FJS International, LLC

Broadcast Solutions Inc

Ritter Battery

Technodolly

Willys Widgets

Tai Audio

Opal

The Slider

Wooden Camera

Abel Cine

Glidecam

CineTape

CineLab

Rig Wheels Passport

Aerial Filmworks

Media Blackout - Custom Cables and AKS

Visual Products

rebotnix Technologies