Adding noise anyone?

[den]

In other words, how do I take a digital file and use PWP so that it fools the film enthusiasts into thinking it was shot on film?

OKay... let's set aside affordability and quality of hardware that separates film and digital to produce prints either as a hobbyist, professional, or commercial enterprise and address the impact that PWP as software offers regarding digital image data access and manipulation.

As background take a look at this "DPR Feature: Apical dynamic range interview" [ http://www.dpreview.com/news/0903/09031801apical.asp ] to see possible trends towards processing digital dynamic range in-camera.

I believe that in order to achieve a 'film look', a digital camera image file with near full dynamic range, good exposure, and good focus after a basic RAW conversion needs to be (1) separated into visual elements; (2) each visual element optimized for brightness, contrast, and color; and (3) the visual elements re-combined into a single 0 to 255 gray tone image file of infinite colors without objectionable halos (auras), artifacts, or tone reversals.

PWP5's 3-Zone Adjustment transform using (1) symmetrical tone range masks; (2) tone range specific contrast, brightness, and saturation adjustments; and (3) image version blending in the HSV color space model offers easy user access to an image's data and its manipulation to achieve a 'film look'. See: Multi-Zone Adjustment: http://www.dl-c.com/Temp/pwp50beta/down ... stment.pdf.

For serious digital image data access and manipulation, I tend to use manual techniques with: (1) Mask Tool symmetrical and asymmetrical tone range/image area masks; (2) Brightness/Color Curves for brightness, contrast, and color edits; (3) highlight tone range edits in the HSL color space model, mid-tone tone range edits in the RGB color space model, and shadow tone range edits in the HSV color space model; (4) image version blending with Stack Images/Composite-Blend; and (5) optimized specific image area 'One to One' cloning... followed by crop for (6) composition; (7) 'Creative' sharpening; and (8) 'Output' sharpening.

As digital images are not created 'equal' there may not be a 'magic bullet' or a generic workflow applicable to all images... I tend to enjoy taking my time with an image to maximize its potential... which at times results in 'over-processing'. You are welcome to visit: http://www.ncplus.net/~birchbay/09/IMAGEsbyDEN_09.htm to see my latest recorded near world view or experiment... ask if there are questions.

[couman]

In all fairness, not all films are equal, so you must define exactly what "look" you are seeking. Norman Koren discusses some differences between film and digital responses and also differences between some films. He also suggests a particular PWP curve form for simulation of film contrast. However, as I think Den has suggested, simple application of that curve (or any other transformation) without a clear concept of direction; understanding of the tools; and understanding of the psychological impact will not necessarily lead to a pleasing result.

[tonygamble]

I read Den's response to say that digital looks like digital because of its imperfections. He listed all the criteria he sought and the assorted techniques to achieve them. I'm assuming, Den, that you are saying get a digital image as close to perfection as you can and it will look like film.

When I started using digital I used still to take my negs to my local pro lab when prints were needed. There were always prints taped to walls, or on the retouching desks. There was no mistaking which were from negs. They had a certain look about them. I repeat that they were not 'better'. It was that they were different.

I guess the first thing is that they were softer, not as sharp, or however one wants to put it.

They did a lot of work for Alison Jackson
http://www.alisonjackson.com/index.html

She worked with film, and unashamably badly, as she could never be bothered to work out how her equipment worked. They were fuzzy and grainy. It was the reportage look she wanted for her exhibitions.

At the other extreme they did the prints for the London exhibition of Eve Arnold's Marilyn Monroe shots. You'd never say they were fuzzy and grainy - but again it was clear that the source was negs.

The last poster makes the point that the mindless application of a curve is unlikely to achieve anything. Hence my quest to dig deeper into what makes the images look 'different'.


Tony

[den]

Tony...

...for me the 'film look' is what can be achieved similar to large format film cameras and sea-landscape/nature image content, preferably with vibrancy and contrast, yet containing crisp detail as the eye/mind moves from one visual element to the next....

[tonygamble]

Den,

I had to chuckle when I read that reply above.

Here is me trying to suck you down into emulating Henri Cartier Bresson with his early Leica and there is you teaching me how to use my 350d to emulate a 10 by 8 plate camera !!

No wonder I felt I was getting the wrong answers !!

You know the sort of pix I mean.

Bresson:-
http://www.google.co.uk/imgres?imgurl=h ... 1&ct=image

Horst:-
http://www.horstphorst.com/works.php?ca ... nvno=yw472

National Geographic covers:-
http://ngm.nationalgeographic.com/2002/ ... index-text

Beaton:-
http://www.guardian.co.uk/artanddesign/ ... is-beetles

We could go on forever. Almost certainly the shots you'd never want to print, but I'd quite like to be able to take a shot in 2009 and get that effect.

Tony

[tomczak]

Simulating film look is an interesting question. There is an old discussion on it here:

http://www.dl-c.com/discus/messages/2/1808.html

Other then lumpiness of the grain and its dependency on luminance, I noticed a few other things:

1) reducing pixel level chroma noise tend to make luma noise more prominent and grain-like.

2) black and white images bring nostalgia and people automatically more easily associate them with film than digital regardless how they were done, I think.

3) There are subtle colour and contrast signatures perfected for years by film manufacturers (e.g. it's hard to mistake Velvia for Kodakchrome), that are not that easily simulated digitally.

4) for the longest time it was trivial to distinguish between movies shot on actual film and those taped by a video camera. I always wondered why. One suspicion that I had was that film movie cameras were used for expensive productions, with good lighting, equipment and competent people. Videos, not always so. So, at least sometimes, maybe it's the quality of the set and lighting not some specific technical difference?

[tonygamble]

Maciej,

Thanks for that reply.

It's a pity that the thread you found, from 2003, does not seem to show any particularly convincing examples of simulated grain. What it suggests is that simply overlaying a texture is not going to work. It's also a pity that those images in the final posts never appeared. I wonder whether the poster actually had cracked the challenge.

Your statement about grain being lumpy is maybe the key. I wonder if, with the extra power and experience of using PWP there is a way of getting this effect. And grain is more noticeable in the darker elements (yes?) so that might be a factor to incorporate.

Your point (2) about nostalgia is highly relevant. I shoot quite a lot of amdram and the casts now like their musicals printed in colour and their plays printed in b&w. Now easy with digital - but in my days of film this sometimes caused problems as the shots were also used for other purposes other than for the cast albums.

Point (3). I use Bibble to convert my RAWs and there are plugins that are supposed to emulate various well known film stocks. A nice touch. Does anyone remember the Dufaycolour sequence in The Aviator? Miss Hepburn playing golf on blue grass!

Point (4). I used to work in television and remember the problems of making dramas where most of the action took place in the studio, but we used filmed inserts for the external sequences. They really looked different. One of the techniques used today, when shooting on tape, is to tinker with the frame speed to create the slight strobing effect that comes with film. And now people are much cleverer about tinkering with the tonal values on tape.

It is my last sentence, above, that makes me think we have, with PWP, the ability to take one of Den's lovely Birch Beach scenes and make it look as if Henri Cartier B shot it with an old Leica through a train window! Quite why is another matter! I suppose, like many of us around here, life is nothing without a challenge or two.

Tony

[tomczak]

Thanks Tony,

I think it's worth experimenting. I'm guessing, but what Den started to do, illustrating the old thread, shows a promise (i.e. texture - between power and frequency sliders maybe there is a setting that that can look like film grain...?). Textures can be made pretty fine but still lumpy, and with 1024 panel size the grain may be irregular enough.

What I tried, and what seems to be encouraging, is to make a black and white texture with Power that happened to be close to -0.4, and Frequencies between 0.1 and 0.9, using the large panel size, and then used the texture either in Additive Filter with some pretty large Exp. compensation or in Composite/Soft Light (using reversed HSV-V as a mask - it's easy to do, just open the Mask and mask with the straight Brightness tool then reverse the black and white Amount sliders when filtering or compositing, while keeping the mask tool opened). If I did it a bit more carefully, I think I could make it look really close to film... But the jury is out...

I was going to bring it up elsewhere, but since Tony mentioned it already and it has to do with satisfying colour gamut of films: simulating film in RAW (or more generally, changing the RAW colours, either in RAW dialog or after, the way we want them). It's not that easy, I found. Sure, one can fix a colour or another, but it's the overall gammut that I don't seem to be getting right. Most of the time I'm getting haphazard results, regardless colour spaces and adhering to them - far from consistency and feel of the slide film, scanned or otherwise.

Jonathan did some testing on colour redention of Velvia and Provia: http://ftp2.bmtmicro.com/dlc/Velvia%20v ... 20100F.pdf. Den, Elie and others already designed a Colour Correct or Colour Map procedures to selectively saturate colours. I was wandering it this could be done a little more 'scientifically', for lack of a better word. If Jonathan could make available the actual files of Makbeth charts from the above white paper, then a Makbeth chart could be shot with a specific camera with specific illumination colour temperature, then RAW processed with fixed adjustments, and then Makbeth patches mapped to velvia or provia, and used for all other RAWs. Perhaps Match Reference could be persuaded to make it happen too (kind of doing it backwards). Does it make any sense?

[tonygamble]

I'm bumping this as there is a relevant thread running on DPR

http://forums.dpreview.com/forums/readf ... d=34664232

You can read the stuff yourself, but I'll copy one post:-

"Do it all the time. Exhibition prints - Lambda 32 inch. From A dsl (5d/5d2).

I only use guassian noise (applied selectively using layer blends). I suppose these plugins do a good job, but to my eye PS Guauss works just as well.

BW and colour. Another technique is to add progressively as you interpolate up.

Works fine, with a bit blur up to 50x50 inch prints.

This has worked fine for me and no one has complained (exhibited in major galleries).

Digital looks terrible w/o noise when blown up to such dimensions.

But I even add it to smaller prints (eg 20x32 inch)."

Tony

[den]

TonyG... Others...

Also on the DP Retouching forum was a HDR challenge thread where the OP supplied three image versions. I took them and in PWP5: image blended with StackImages, 3Tone Range tone mapped, converted to BW, more tone mapping, and then... added simulated grain [noise] per recent postings in the High ISO monochrome image thread....

My DPR posting is here if of interest:
http://forums.dpreview.com/forums/read. ... e=34683487