Soft focus filters or not?

MrAnderson wrote in post #18258236
That made me laugh. Thank you

It may be amusing to hear about something as archaic as a filter, but the question on the table is whether common digital postprocessing found in Photoshop is truly capable of mimicing an actual filter, or not. Tiffen offers software, but unless you have the same shot with both filter and with software, it is hard to do a direct comparison to see to what extent software that is as effective a substitute for something as the widely respected as Softar filter used on so many Hasselblad shots.

fwiw- I sometimes enjoy mixing film techniques in a digital world. I still grab my Softar's from time to time (both as an overall global diffusion as well as introducing the Softar with a separate strobe pop during a time exposure to add a soft diffused highlight). I've also been playing around with 80's light painting techniques, I still drag the shutter to introduce a bit of warm modeling light, etc. At the same time, I have a very competent digital tech that smiles whenever he sees me going in one of these directions, convinced that he can achieve the same results with a few clicks of the mouse. So far, he's tried his hand a few times and admittedly has come close on occasion, but to my eye, it's still not exactly the same (and it generally takes him much longer to produce).

Foodguy wrote in post #18258285
fwiw- I sometimes enjoy mixing film techniques in a digital world. I still grab my Softar's from time to time (both as an overall global diffusion as well as introducing the Softar with a separate strobe pop during a time exposure to add a soft diffused highlight). I've also been playing around with 80's light painting techniques, I still drag the shutter to introduce a bit of warm modeling light, etc. At the same time, I have a very competent digital tech that smiles whenever he sees me going in one of these directions, convinced that he can achieve the same results with a few clicks of the mouse. So far, he's tried his hand a few times and admittedly has come close on occasion, but to my eye, it's still not exactly the same (and it generally takes him much longer to produce).

Absolutely nothing wrong with a blend of the best of both worlds!
I still do not think digital can equal what you can obtain with a black and white enlargement on photosensitized paper! Inkjet pigments layered on top of glossy coated paper just are not as dimensional.
Nor can digital 4K projectors (as expensive as they still are) equal the impact on photography enthusiast audiences that a medium format transparency projected on a large screen will elicit from those accustomed to seeing 135 trannies.

Wilt wrote in post #18258252
It may be amusing to hear about something as archaic as a filter, but the question on the table is whether common digital postprocessing found in Photoshop is truly capable of mimicing an actual filter, or not. Tiffen offers software, but unless you have the same shot with both filter and with software, it is hard to do a direct comparison to see to what extent software that is as effective a substitute for something as the widely respected as Softar filter used on so many Hasselblad shots.

Wilt when I was in university doing my Electronic Engineering degree, UEA 92-95, the course work in my C Programming for Engineers course in the second year consisted of writing several different image filters and even a basic interpolation filter. The lecturer was a biologist with an interest in using machine vision to study cell mitosis and growth. Remember that this was in 93, so well before Photoshop had become a verb. So I have a little experiance in what is required for this sort of stuff, and given the level of the course it was computationally pretty simple stuff, well within the confines of a STEM degree course.

The simple filters like the Gaussian blur, or any of the other types of mathematical distribution are pretty easy to do, since the modulation transfer function is well defined for them. I'm not able to go into details now, since I don't have my notes from back then, nor any of the source code. Finding it all back from scratch would be a very big task too. Still once you know the transfer function the filter implementation is relatively trivial from a computational point of view.

The difficulty for reproducing the things like the net filter is working out the MTF for the filter. For a blur that is going to effectively introduce some repeating pattern I would expect the MTF to be resolution dependent. Considering the location of the filter I would also expect that there will be a degree of focal length, focal distance and aperture diameter dependence too. So to get a perfect digital match to the analogue filter you could need to make the MTF calculations pretty specific. It would then be quite interesting to see just where the limits become visible.

This is where I wish I still had all my notes, and the original source code and ancillary functions, like being able to read a TIFF file. We only worked on monochrome 8 bit TIFFs back then, JPEG had only just been ratified as a standard at that point. I would probably go and have a play with this. Thing is though all of the files were on 3.5 inch floppies, so even if I still had them, I wouldn't be able to read them now.

Alan

Thanks for those comments, Alan. They reinforce the initial impressions that I had, that the Gaussian blur was simply a generalized blur solution, not a parameterized one that might involve variables like FL or the spacing of a net or optical bumps. If I had a spare $250 to throw at my curiosity, I would buy the Tiffen DFx software and also a Zeiss Softar 1 filter, to complement what I already own, and then do some serious comparative and tightly controlled shooting, and then also apply common software filters found in Photoshop or other JPG editor as well as the Tiffen software.

BigAl007 wrote in post #18258465
Wilt when I was in university doing my Electronic Engineering degree, UEA 92-95, the course work in my C Programming for Engineers course in the second year consisted of writing several different image filters and even a basic interpolation filter. The lecturer was a biologist with an interest in using machine vision to study cell mitosis and growth. Remember that this was in 93, so well before Photoshop had become a verb. So I have a little experiance in what is required for this sort of stuff, and given the level of the course it was computationally pretty simple stuff, well within the confines of a STEM degree course.

The simple filters like the Gaussian blur, or any of the other types of mathematical distribution are pretty easy to do, since the modulation transfer function is well defined for them. I'm not able to go into details now, since I don't have my notes from back then, nor any of the source code. Finding it all back from scratch would be a very big task too. Still once you know the transfer function the filter implementation is relatively trivial from a computational point of view.

The difficulty for reproducing the things like the net filter is working out the MTF for the filter. For a blur that is going to effectively introduce some repeating pattern I would expect the MTF to be resolution dependent. Considering the location of the filter I would also expect that there will be a degree of focal length, focal distance and aperture diameter dependence too. So to get a perfect digital match to the analogue filter you could need to make the MTF calculations pretty specific. It would then be quite interesting to see just where the limits become visible.

This is where I wish I still had all my notes, and the original source code and ancillary functions, like being able to read a TIFF file. We only worked on monochrome 8 bit TIFFs back then, JPEG had only just been ratified as a standard at that point. I would probably go and have a play with this. Thing is though all of the files were on 3.5 inch floppies, so even if I still had them, I wouldn't be able to read them now.

Alan

It's interesting, Alan, because I myslef worked for a company (HP) that has had a huge sttake in the digital imagery field, scanners, they pioneered the whole inkjet printer field), and then digital cameras (my first digicam was an HP "compactdigicam
But I actually
So, like you, my early stuff was in the programming field, C and C++

But I actually mever de;vd into the digital imaging progreamming, even though I was surrounded with such folks, I just did software development, and never got into those details, I ahave thought it would be cool to get into some of the development that has gone into the proceessing work we do, Lightroom and PS and such.

In fact in the vrery early years, a C Language program was put out that became the "basis" of much work that followed, but it might have been fun to go over that type of works, line upon line, who knows what could be discovered??!!