Can AI and Film Photography Happily Co-exist with Each Other?
This blog started out as a thoughtful post for Facebook asking if AI could be used to help speed analog, or film image processing once transferred to computer, but Facebook in it’s infinite wisdom took it away again without rhyme or reason. So, I start again with a clear idea of my thought process, but no actual memory of the words.
It started by me looking at an image from our recent Tenerife vacation which I had somehow sneaked my Nikon FE film camera into, together with 3 rolls of Fomapan 100 film. The FE was essentially my carry around camera even though I also had my Fuji XT2 available. The Nikon just felt more portable, less obtrusive and more fun. However, fast forward a week or so and my three rolls of Fomapan are developed in 510Pyro, scanned and awaiting processing which initially is just a straightforward process with Negative Lab Pro (More on that too) Then they sat on my hard drive for a day or two before actual work could begin on the images. First thoughts were that these were a good set of pictures, sharp, well exposed with good detail and well controlled grain. From there, I began to look at patterns and themes of which several emerged, but this post is not about those. Other posts will look at that matter.
No, this is about how when all elements work together, the Analog workflow can be a pleasurable one from start to finish. Also, how AI can be profitably employed by the analog photographer. This image shot of a balcony with angular potted succulents, white paint and a clear blue sky fitted my perception of a place filled with light, contrast and colour, but with little in the way of soft greenery. Although, I was happy enough with the composition, it needed a crop to clean it up, and removal of some distracting elements. Also adjustments to levels in different parts of the image, plus removal of some defects and dust spots. That I accomplished this with just mouse clicks utilising smart objects, Photoshop’s object selection tools and neural filters taking just a minute or two is a testament to the ongoing improvement in AI over time. The ONLY manual work I did was to brush away some of the Smart filter mask to restore a couple of elements accidentally removed by the Photo Restoration tool. Now, I’m not saying this is a world beating image, but it is a satisfactory one, especially in the context of what I was trying to do in a strange place filled with angular modern architecture and spiky desert plants.
Being able to get on with the processing by starting with a well-presented negative is a key element of that and my care in production, plus an element of luck in picking a development time for the combination of Fomapan 100, a film not normally noted for it’s fine grain structure. Fortunately however, I have been using 510Pyro, a newly developed fine grain liquid, one-shot developer recently which seemed like the ideal partner to manage grain in the final result.
So now we have come round full circle back to this image. Above are the final and uncorrected images together plus a screenshot of the PS environment in processing. You will see that this last is extremely simple, there are not dozens of layers and masks, just a very simple setup.
But lets go back even further, as far as loading these 3 films into a developing tank. As I have a darkroom, the loading could be done without the aid of a changing bag or light tent, reducing the risk of introducing dust or other contaminants into the process. Wearing suitable gloves further reduces the risk, Temperature control in development also helps eliminate variables and being careful to follow the process exactly also helps. Post development is also important, so in this case no squeegee was employed to eliminate scratching the soft emulsion. Once dry, the negatives were immediately sleeved, again with gloves worn and then scanned at the first available opportunity, being careful to ensure all surfaces were prepped to mitigate against the intrusion of contaminants. My standard scanning process (to be discussed in a further Vlog/Blog) was then followed. This resulted in images suitable for use in most applications and as discussed above able to be processed quickly and effectively. When you start with 100 plus images, winnowed down to around fifty after initial screening, this is important as a smooth workflow means a more pleasant one which in turn means focusing on each individual image is easier. Negative Lab Pro (NLP) which is my preferred tool for neg-pos conversion is another important part of this process. Recently updated to version 2.4, this is pretty much a one-stop shop for analog image makers. Some I know only use NLP, but I am comfortable with Lightroom (LR) and Photoshop (PS) and this is where the AI tools and Neural filters as discussed above come into play.