What's new

A method to generate procedural Grain

Sorry, I had trouble setting up Gizmos in my Nuke application, which is why I never got around to exporting the Gizmo. The Gizmo is still unfinished, especially for things like the Halation node tree, but I wanted to share it anyway for anyone who wants to play around with it. It also includes the "film scratch and damage node" by Taukeke, a variation of CalvinSilly's gate spill node (it's much worse, but I just wanted to try), and the halation is done using Chris Fryer's True Exponential Blur. I'm planning to add Chris Turner's Edge Flare node to 2D image workflows. In hopes of intuitively, procedurally obtaining those edge flares, that's still a work in progress and will likely be a minimal tool.

It is, unfortunately, a Nuke gizmo and not a DCTL, which is probably. Means not many people will get to try it. It's very compute-intensive, which is a shame, but that's just how it is right now.

I'm pleased with the UI layout; I tried to keep it simple and minimal.

AnalogyUI.jpg
 

Attachments

  • Analogy_Gizmo.zip
    41.4 KB · Views: 21
Recently I have spent some more time zooming into some of my own analog shot footage and carefully observing the grown textures. I noticed that the textures were very similar to fBM noise. I started playing around with Nukes noise node, which has a fBM feature.

Ultimately it led me to a sort of combination of my procedural grain method and the plate grain method, but instead of plates using fBM noise layers.

The method is now:
Gaussian Noise (3 Splines with 11 control Points) + Voronoi Noise + 5 different tweaked fBM noise layers with a spline that mixes them at different exposures, plus another 3 sets of splines with 11 control points to further tweak them.

I feel like the results visually look pretty similar to actual grain, but perhaps more importantly it allows for a lot of custom tweaks.
 
GRainComparisonThree.jpeg
Left is 16mm grain scan, top right is a digital grey patch and bottom right is the grey parch with the grain algorithm (settings were tweaked for visual match, which was a bit too strong for me when used on images). The colours are off, but I feel like the texture looks pretty similar.

PatchesGrainMatch2.jpeg
PatchesGrainMatch.jpeg
 
Silvan Schnelli said:
Sorry, I had trouble setting up Gizmos in my Nuke application, which is why I never got around to exporting the Gizmo. The Gizmo is still unfinished, especially for things like the Halation node tree, but I wanted to share it anyway for anyone who wants to play around with it. It also includes the "film scratch and damage node" by Taukeke, a variation of CalvinSilly's gate spill node (it's much worse, but I just wanted to try), and the halation is done using Chris Fryer's True Exponential Blur. I'm planning to add Chris Turner's Edge Flare node to 2D image workflows. In hopes of intuitively, procedurally obtaining those edge flares, that's still a work in progress and will likely be a minimal tool.

It is, unfortunately, a Nuke gizmo and not a DCTL, which is probably. Means not many people will get to try it. It's very compute-intensive, which is a shame, but that's just how it is right now.

I'm pleased with the UI layout; I tried to keep it simple and minimal.

View attachment 13173
Click to expand...
Eager to try this! Are you able to provide the grain plate used in the gizmo?
 
Oskar Larsson said:
Eager to try this! Are you able to provide the grain plate used in the gizmo?
Click to expand...
To be honest wasn’t planning on doing it. Seeing as grain plates are looping images, I wanted to have grain plate sequences that I could personally use and would know is unique to the project. I’m not sure if that makes sense.

I might upload them eventually though, especially since if I’m honest I much prefer the newer method I just posted about using fBM noise, it looks pretty similar to the grain plate and it’s all procedural. The gizmo I posted isn’t using this method yet, I’ll post the newer one soon.

However, the gizmo does have the grain plate all set up, it just needs you to input your own grain plates and reference images (a frame of the grain plates average color). I talk more about it here:

Thread 'A method to employ varying grain plates at different exposure levels'
http://www.liftgammagain.com/forum/...in-plates-at-different-exposure-levels.19497/

Important to note is that I noticed 5 exposure levels to be a bit too rough for the plates. As in the visual disparity between the grain at -4,-2,0+2 and +4 to be too large. This makes the dithering zones, depending on the shots, noticeable. This is another reason I suggest to use some other grain plates. Or to use the fBM method.
 
Also something I’ve noticed is that some vendors sell monochromatic grain plates instead of coloured ones. So it’s worth double checking when purchasing plates, because you lose out on the real colors of the dye clouds, which is a big part of the authenticity.
 
You must log in or register to reply here.
Back
Top