Digital Prints and Simulated Grain - Challenges

[Ganesh H Shankar]

In past when I used to print I used to waste lots of paper because the 'simulated grain' does not get rendered the way I wanted them to be. I used to do some trial and error, yet the issue remained and I was not happy with how simulated grains get rendered compared to how they get rendered in the traditional wet room prints. Of late I have been wondering about the issue again. I think I now know the root cause of the issue, no solution yet. With the root cause known I may be able to think of some solution.

The issue

Softwares like Nik, TrueGrain etc gives us ability to simulate film grains. True Grain actually overlays the real scanned grains on the images making them as accurate as possible. However, the issue is incorrect relative size of the grain crystals in differently resized images!! These softwares will not have any clue on how much the image has been enlarged/interpolated. Irrespective of the enlargement the software overlays some fixed grain structure on the image. This is wrong!! The grain control sliders, namely "Grain per pixel" in Nik's or "Film size" TrueGrain is hard to tweak. The problem does not arise in wet room printing. The enlarger will proportionately enlarge the grains too on the final print and this is a non-issue. In the digital workflow this crucial information about amount of interpolation and hence the relative size of grain lumps compared to picture elements/pixels is completely lost!! Irrespective of how much you resize the software will overlay same grain structure on image. This results in lost aesthetics.

Here are couple of samples viewed at about print size with added grains simulating Ilford FP4 grains. The first one below is from Nik's SilverFX which I guess is simulated and the second one is from TrueGrain when is the real grain structure of Ilford FP4. Both look very different though!


Nik's Ilford FP4 grain simulated (grain per pixels adjustment appears to give more of noise than grain compared to TrueGrain's version below)




TrueGrain's Ilford FP4 grain simulated (Film size control seem to give far larger grains which does not go well with the image even at 100% (6x8 negative)).

While I like TrueGrain's structure, the relative size of the grain crystals compared to the image elements is completely lost (in both). Interestingly TrueGrain has different grain overlays for the same Ilford FP4 film depending on the format, 35mm, 645 etc. Larger the format finer the grain for the same film (as it looks like). However, scaling information is not taken correctly into account at all. Essentially "Film size" control is hard to tweak.

I think, Raj's scan of his film solves this problem for him since he scans both image elements and the grain together! The relative size/structure of the grains are maintained in the enlarged scanned digital image or the print. This is not possible when we overlay the grain on the final digital files with inadequate control over the grain size. I now suspect this is one of sources of the lovely look of the grains in scanned digital files (from films/slides) which is hard to manage in all digital print workflow.

Too bad, we have no equivalent for the enlarger lens in digital print with grain workflow!!

[Prashanth Sampagar]

Ganesh,

Thanks for the comparison and explanation. Now I know, this is the reason I never liked grain added on the lower resolution images. I used to add grain in the highest resolution(36MP) itself and it seemed to work. On lower resolution images the grain seemed bit too aggressive. Also, I was never happy with the grain on the highlight areas. For RGB values above 230 it always used to mess up, need to recheck it though. Especially for first and last zones of tonal ranges I always felt it a bit out of place. Need to check it and compare.

Best Regards

[Rajkumar]

Yes Ganesh this seems to be the right find. Even though some of these tools seems to have a grain size button. Then again how these grains interact with sharpening etc is a biq question. Yes in film it is baked into the image at the time of taking itself