On Film Grain Alien Skin Exposure, Capture One, DXO, Etc?

The Kodak Black And White Classics Presets for Capture One I put out there a week or so ago has generated more than one conversation, a bunch of questions, and as usual far different from anything expected. Interaction has ranged from the “hey these are great”, to the philosophical, to the technical. I’ll not repeat all of them here but I do want to point you to a project and a post from Andy Farrell more on the philosophical kind of internal debate. I think what he’s made is really great — how about you…

On a more semi-technical note I received more than a few comments/questions regarding film grain emulation specifically and my opinion on various products as of right now. One from this morning via a comment on another post that was really too long to answer as a comment. I’ll try to wrap up a couple of thoughts along with a response to that comment as briefly and as visually as possible. Here we go…

Film grain is a funny thing to nail down in terms of visual appearance. Even when it’s not really visible in small print sizes there seems to still be an influence of some sort on the overall aesthetic. As you print/magnify more and more and can see the construction of the image it becomes more and more apparent that the image is made out of grain rather than some sort of image that’s being obscured by the stuff. Real film grain doesn’t give the perception of an overlay of stuff on top of the image at all. It’s part of it. It’s easy to see in a big print from a small film negative. That’s what I’m doing with my project from 2014 that’s wrapping up. That’s why I spent a lot of time again on comparing software and seeing what would most closely match real film shot side by side of the same subject on the same day.

Warning And Background

I shot some Kodak TMAX3200P for the above referenced project. It was old, in fact it had various differing expiration dates. I shot that film months apart. I probably varied development here and there. The PH of my water was different depending on what month or day it was. Even with the same developer the contrast/grain/all of it will look different. Exposure variations roll to roll and frame to frame may give slightly differing grain rendering depending on where the subject and scene landed on the response curve. If you put it under a microsope film shot on different occasions and developed the same way WILL look a little different. How about completely different developers like HC110 or XTOL? Hell yes. General characteristics will be the same but they will not “match”.

So that brings us to what I look for in a grain simulation. Why I think Alien Skin Exposure 7 is at the top of the heap for me and why I was shocked at how good Capture One 8 was.

  • I absolutely do not expect any simulator to look exactly like my film out of the box.
  • I do have some broad criteria on how the grain interacts visually inside an image across various scenes and specific things, get to that in a second.
  • Specifically I am looking for controls that will easily allow me to match my real film look at big print sizes. Knowing that I may have roll to roll scene to scene variations and definitely some variations over months.
  • Whatever that recipe is I want it to visually look the same across cameras at the same print dimensions no matter what the input file size is. In other words I want my X100S to look like my D600 with the same “settings” and I want them both to look the same as my film all printed at say 20” x 30” without screwing around frame to frame camera to camera.

What I Look For In Grain Simulation

The image at the top is one I happened to have laying around I used for yesterday’s discussion — I think. It happened to be what came up when I fired up Capture One. Whatever, it happens to be using the Kodak TMAX3200P preset. Those presets match up the average look for that film they way I happened to process it last year. It’s not any particular “match” with a particular frame. Doesn’t matter, especially at web-size. The image directly above is the same thing processed with Alien Skin Exposure 7. Never did that before.

Let’s take a microscope to those results shall we…

First up above is Capture One the way I plastered on the presets yesterday. If you open that image in a new window and look at it full size you can see all the pixels. Roughly a 4 foot x 6 foot print on a screen of around 100DPI. A microscopic view.

Next up we have Exposure 7…

I didn’t start from scratch in Exposure 7. I started with they’re box stock TMAX3200P and went top to bottom in Exposure 7’s list of controls. Could I even get closer — sure if I went back and tweaked some of the controls based on my subsequent adjustments. This was less than 10 seconds in Exposure 7 to match something I spent a far far far longer time getting to in Capture One. Hence my easy/quick matchup criteria… one point for Exposure 7. The controls are more direct and less fiddly than what we have in Capture One 8 in terms of grain.

Forget that as it’s my criteria mostly because I want to match specific results… for those that don’t it’s a wash. Instead let’s talk about what we see under that micro scope and what I look for.

  • Neither have an impression of an image masked behind some grain pattern over it. Both have a similar “the grain is actually the image” look. Can’t put it into better words. This is the big deal for me.
  • Next up is the way the grain interacts with detail in relation to the size of the grain. This is complicated and not at all an attempt to reverse-engineer how either vendor’s algorithm works. Doesn’t matter it’s more of a visual thing no matter how either gets there. This image happens to have a great mix of sharp and soft edges and gradations. A good mix of in focus out of focus along with motion blur. What I’m saying is has a huge range of detail of various frequencies. the way that real film grain looks in relation to edges/detail at various frequencies is interesting and a little difficult to describe if you’ve not looked at a lot of films before. Real grain tends to mask detail — even sharp detail above a certain frequency (fine detail) depending on that film’s grain size. Then there’s a sort of transition — a soft transition as that fine detail approaches the threshold of the film’s ability to resolve and grain size where out of the blue something that’s sharp and fine just jumps out visually as if grain that’s larger than the detail magically lines up to be smaller…. like a static hair here or there without rendering the ones that are too fine at all. On detail that’s above that threshold there’s no grain appearance at the edges at all. The detail doesn’t follow the shape of the grain at all. That changes slowly depending on how soft the gradation is.
  • Shit, that last one was long but I wanted you be able to look and see it. Okay that brings us to smooth gradations which this has plenty of… See the way the grain works, and looks in mid-tone gradations from dark to mid to light on diffuse things. The way that it’s dithered… to me that can either look digital or film grain like. Both of these look similar.
  • Last take a look at the areas of no gradation — even tone. Look at even tones in shadows, midtones, and highlights (not whites) See how they are different? Also note that all of them have a very random and pleasing “clumping” that’s not at all pattern like. Pattern as in repeating pattern that’s at all detectable.

Those are what I’m looking at. The fact I find the rendering comparable from a dedicated thing that’s at the top of the film emulator food chain like Exposure 7 to a full-blown raw processor/workflow tool is what shocked me. I was expecting something more along the lines of a LR grain — not that great and definitely not able to match up my real film results.

That Matchup Via Exposure 7

Now the more nit-picky of you will get all over me about how I didn’t match the renderings precisely. Nope, when it comes to blog posts I’m really lazy. A bit of context should assure you that my “easy” matchup words match the fact.

First off I didn’t use Capture One to process the RAW file and then use that as input to Exposure 7. Nope, I used the idiotically lousy ability of Exposure 7 to deal with the RAW directly. Completely different input data. For the Exposure “match up” here’s what I did.

  • Opened the RAW file directly in Exposure 7
  • Hit the Kodak 3200P factory defualt
  • Yanked the exposure control to -0.25
  • Went down to the tone curve control which is very unique and interesting for Exposure 7. The sliders make modifications to the preset tone curve but keeps the base shape. Kind of nice. I upped the contrast a hair. Then pushed the highlights slightly whiter and the shadows slightly blacker.
  • Under the grain control I modified the default factory TMAX3200P by upping the shadow amount and the highlight amount and added a touch more “roughness”.
  • I was still laking a hair of contrast so I used the curious “push control” in the grain section. I say curious because upping this increases grain size and increases contrast that shows up nowhere else, not in the curves, not in the other grain/size/appearance sliders.

Yep, if I re-tweaked any of the other stuff I could get exactly. Not my mission. Mission was to get it close enough so that you could see the characteristics well enough to see how similar they are in end result. Why didn’t I go the other way around? Well, based on screwing with the Capture One 8 controls I already knew they are not nearly as strait forward to get that matchup. In other words you cannot really match a control up with what you see. I saw the shadows looked different hence I messed with shadows, etc. Matching using a Capture One to real film or even the Exposure 7 results is not as strait forward and requires trial and error by messing with all the controls including the “type”.

One last thing to beat the horse even dead-er. The fact that the out of box Kodak TMAX3200P didn’t happen to match my results has nothing to do with “accuracy”. What do those OOB results look like? Well, they look like this.

Does this mean Exposure 7’s version is wrong? Nope, to me that looks way more like TMAX3200P developed at maybe Kodak’s recommended CI in like XTOL or something. Me, I expose at 1600 or more depending on how I meter the scene and then develop the living shit out of it. Especially when it’s not brand new. That gives me shitloads more contrast and then I print for the highlights and let all that contrast push the blacks to crazy black and the shadows down.

Hope that helps answer a few people’s questions. The bottom line is that Capture One 8’s grain emulation does not fail big-time at any of the criteria I look for when matching the real thing. It’s really close to Exposure 7 but Alien Skin’s controls are far easier to match up with very specific film results side by side if you do that sort of thing often.

RB

blog comments powered by Disqus