Subscribe to this blog via RSS-feed (Android-users click here)


Digital photography is dead? Film photography isn’t?

DALL-E 2 algorithm seems so unbelievably good that I would call it fake. Well, I will call it fake – in spite that few ”influencers” have access to it and have found it amazing and actually working.

In video above the question is what happens to photography in few years because of DALL-E 2 and such AI algorithms – which there must be plenty under development right now.

Will it ”kill” stock / commercial photography? For sure it will have large impact. What about moving picture? I believe the impact is much bigger there once the wheels start to roll. A movie scene generated by just few words instead of thousands of work hours? In some years that is actually possible. Sooner than we are ready, I’m afraid.

Film photography was smashed and brutally killed by digital photography back in first years of 2000. And back then digital photography was sucking really bad; it wasn’t competitive to film iin terms of quality or dynamics but it still smashed film photography like a giant. So if you are still discussing if digital photography has killed film – it has. It has stomped over the corpse for many times.

Could film photography shine again after AI kills digital picture? With authenticity it could. For people who are only scanning their negatives this is bad news. Probably AI will laugh all the way at you. Just type the sentence describing your scene and add ”tri-x film photo look” and nobody every can make difference was it actually shot on film.

Of course there might still exists machines that ”print” digital images on film but those things are rare. So it is pretty safe to say that film photographs are mostly ”real” thing – not AI generated. Real deals, from real situations. And then make print of the real negative in darkroom on silver gelatin; that is the market of authenticity. And let’s not forget wet collodion process and contact printing ..

I’m pretty sure media and masses will fall in love for AI. For film photographers the honeymoon is the time to hit. Maybe people realize that all their videos (and photos) are always faked and some might start to look for authenticity.

BTW: Somewhere there was interesting note that AI generated ”photos” cannot be copyrighted at the moment because the creative work must be output of a human. So what it means is that anyone can use other companies AI stock photos for free, I guess. I’m sure laws will be changed to prevent this when the money starts to talk.

Subscribe to this blog via RSS-feed (Android-users click here)

pinhole technical

Analyzing pinhole with scanner

I’ve read many times that people analyze their pinholes with a scanner. I tried this today with Epson V600 and my conclusion is that the resolution is nowhere near to make any good analysis. The shape / quality of the hole is probably pretty hard to see. I scanned at 2400 dpi and on my file one pixel equals 0.01mm so the difference between 0.28 and 0.32mm pinhole is only four pixels. 

These are two commercial pinholes from eBay. The millimeters are the advertised sizes and the cyan lines are lines placed for the advetised millimeter distance.

Because the resolution is so low even at 2400 dpi scan, I can only say that the pinholes might be as advertised based on the measurements I made. Both pinholes were purchased from same seller (1 year apart). Based on sellers microscopic images the hole should be really clean.

Subscribe to this blog via RSS-feed (Android-users click here)

grain technical

Digitizing film negative with digital camera

I’ve previously tried to digitize my 135 film with digital camera without any great results. Few years ago I purchased Canon FD slide copier + bellow combo and it takes only Canon FD lenses so I tried with 1.8 50mm FD lens and the results were .. horrible. Well, great in stylistic way.

I printed some mounting rings so I could use enlarger lens instead of the smuhy mushy FD lens. I got everything running and ..

I was blown away by the quality.

Left: Digitized version, Right: clip from the 30x40cm darkroom print

I rushed to pick one of my favorite negative which shows Tri-X grain in beautiful way and digitized that. I was very sure I couldn’t pick that grain. But it almost did. So close but remember this is quite a challenge. Grain (or actually; the grain clumps) are very small. The part used as example is a small part of 135 film negative.

The clip analyzed here is about 13x18cm from 30×40 enlargement. Calculate the enlargement amount if you want, it is quite big. The original is 135 negative.

I think digitizing this way is very very good This is just unbeliavable and mind blowing. I never understood digitizing with digital camera could produce such high quality. Also the easyness digitizing a single roll is so much faster than with scanner when the roll is uncut. I believe it takes 2-3 minutes to capture all 36 frames – and the quality is much much better than with flatbed scanner.

So question: is this some kind of hidden secret? Voodoo? Is the flatbed mafia next to my door when I post this?

Digitized with Fujifilm X-T30 + Nikkor 50mm enlarger lens, ISO 160, 0.9s exposure time. On right darkroom print on silver gelatin.

Below is the full frame from 135 film negative (and the previous clip):