James O'Neill
Veteran Member
Try to keep a long story short.
I first used a computer at school which was obsolete already and used paper tape. One row of punched holes was one byte and IIRC 1/10th inch so 1MB would be ~100,000 inches 2.54KM / 1.6 miles. When I started using floppy disks they were 100K. The first hard disks I saw were 20MB and took about 20 minutes to back up to tape. So the K1 writing 40MB files is a sign of how far we've come.
I upgraded my laptop earlier in the year and have a bigger drive, I just cleared 120GB of raw files off it. A couple of laptops ago the hard drive wasn't that big. I did a search for duplicate files, which meant reading 220GB from my "my pictures" folder, generating a hash of each, and seeing if that hash had been seen before. It crunched them at 420 MB/Sec. That's 21 of those 20MB disks a second, or paper tape travelling at 1000KM/sec - air friction would cause it to catch fire long before that
Since I had my archive drive out I went back to some old portraits. Some I'm quite sentimental about from the *ist-D, some with my original K5 and some with the K5-IIs a lot used the FA-50 f/1.4 lens. I only have a handful of shots with this lens and the the K1, but I should do an objective comparison (pun intended). I wondered what would happen if I reprocessed some old images - some from 2004 / 2005
The first thing was whether "enhance and super resolution" which is now in Lightroom - make any difference? The answer was a firm NO. It HAS given me amazing results with some shots of the New York Skyline I took with the *ist-D, but portraits don't have enhance noticeably, with the current tech at least.
The shock was the different way I (and I assume everyone) look at pictures.
I've made good A4 prints from 1MP, and have good A3 prints from the *-ist-D.
But the degree of zooming in I've grown used to reveals how little detail there was in an A3-printable image. Blemishes that I might retouch out on some pictures today wouldn't even resolve on the 6MP sensor which (AIUI) made matters worse with an anti-aliasing filter.
That FA-50 is a lens I liked, I was using it with studio flash and quite small apertures (mostly around f/8 or f/11) but for 3/4 portraits the head is towards the edge of the frame where the lens performance drops off, where I now see individual eyebrow hairs is just mush on a *ist-D image - partly the lens but mostly the sensor.
And zooming in I really notice noise. The base iso on the *ist-D was 200, and I was reprocessing some pictures which for reasons unknown used ISO 400 and f/11 instead 200 and f/8 - one of those shots I'm sentimental about was a "save" when the flash didn't fire and instead of reducing exposure by about 1/2 stop I had to increase it by 2 2/3, so net it's about ISO 2500. Pattern noise in the shadows, and intrusive noise is everywhere, needing heavy noise reduction at the cost of more detail and shadows need darkening to black.
Really, if I were doing this properly I'd find some K10D and K7 shots to compare, because there is a night and day difference between 16MP on the K5 and 6 on *istD.
The FA-50 delivers a ton more detail. I couldn't do a proper comparison of noise, but I found some where I'd used ISO 12,800 on the K5-iis to get a grungy look and it's far less than that "save". I took the K5 diving 10 years ago (!) and I needed ISO 6400 and posted https://www.dpreview.com/forums/post/50779054 on the amazing performance. 2000-2010 was a period of massive sensor improvement, the improvement was something like <10% efficiency to >50% efficiency and obviously you can't double efficiency once it is over 50%., so the last 10 years couldn't deliver the same progress.
Examining K5 shots with the same critical eye there are "portfolio" shots which don't stand up to a 2023 sharpness benchmark, degrees of mis-focus which were acceptable then would make me unhappy now, and skin blemishes which are only noticeable at 300% I'd now retouch and didn't in 2013/14. I can, just, find details in shots with the K5-iis without the AA filter using the 77ltd which the combination of FA50 and K5 with AA filter didn't capture. If you know the scene from Blade Runner where Deckard zooms in and in and in on a picture, I think there is some detail I worry about now which only shows when given "the full Blade Runner treatment". Yes it would be visible if I made prints the size of tennis courts, and got down on my hands and knee to look, but in normal use... it really doesn't matter.
Editing today doesn't give dramatically different results. 2023 software instead of 2005-2015 software, and with all the extra experience I have doesn't give many improvements, but I will do shots today which I couldn't have made work then.
I worked with softness and a lack of absolute detail before, I seem to demand more now. My (maybe everyone's) idea of good evolves with the tools being used, and pixel peeping at a 36MP DFA* 85 image shows faults which were once invisible - I can't help thinking I'd be happier if I didn't put everything under the microscope at the the edit stage.
Thanks for reading to the end. Didn't intend to ramble so much.
I first used a computer at school which was obsolete already and used paper tape. One row of punched holes was one byte and IIRC 1/10th inch so 1MB would be ~100,000 inches 2.54KM / 1.6 miles. When I started using floppy disks they were 100K. The first hard disks I saw were 20MB and took about 20 minutes to back up to tape. So the K1 writing 40MB files is a sign of how far we've come.
I upgraded my laptop earlier in the year and have a bigger drive, I just cleared 120GB of raw files off it. A couple of laptops ago the hard drive wasn't that big. I did a search for duplicate files, which meant reading 220GB from my "my pictures" folder, generating a hash of each, and seeing if that hash had been seen before. It crunched them at 420 MB/Sec. That's 21 of those 20MB disks a second, or paper tape travelling at 1000KM/sec - air friction would cause it to catch fire long before that
Since I had my archive drive out I went back to some old portraits. Some I'm quite sentimental about from the *ist-D, some with my original K5 and some with the K5-IIs a lot used the FA-50 f/1.4 lens. I only have a handful of shots with this lens and the the K1, but I should do an objective comparison (pun intended). I wondered what would happen if I reprocessed some old images - some from 2004 / 2005
The first thing was whether "enhance and super resolution" which is now in Lightroom - make any difference? The answer was a firm NO. It HAS given me amazing results with some shots of the New York Skyline I took with the *ist-D, but portraits don't have enhance noticeably, with the current tech at least.
The shock was the different way I (and I assume everyone) look at pictures.
I've made good A4 prints from 1MP, and have good A3 prints from the *-ist-D.
But the degree of zooming in I've grown used to reveals how little detail there was in an A3-printable image. Blemishes that I might retouch out on some pictures today wouldn't even resolve on the 6MP sensor which (AIUI) made matters worse with an anti-aliasing filter.
That FA-50 is a lens I liked, I was using it with studio flash and quite small apertures (mostly around f/8 or f/11) but for 3/4 portraits the head is towards the edge of the frame where the lens performance drops off, where I now see individual eyebrow hairs is just mush on a *ist-D image - partly the lens but mostly the sensor.
And zooming in I really notice noise. The base iso on the *ist-D was 200, and I was reprocessing some pictures which for reasons unknown used ISO 400 and f/11 instead 200 and f/8 - one of those shots I'm sentimental about was a "save" when the flash didn't fire and instead of reducing exposure by about 1/2 stop I had to increase it by 2 2/3, so net it's about ISO 2500. Pattern noise in the shadows, and intrusive noise is everywhere, needing heavy noise reduction at the cost of more detail and shadows need darkening to black.
Really, if I were doing this properly I'd find some K10D and K7 shots to compare, because there is a night and day difference between 16MP on the K5 and 6 on *istD.
The FA-50 delivers a ton more detail. I couldn't do a proper comparison of noise, but I found some where I'd used ISO 12,800 on the K5-iis to get a grungy look and it's far less than that "save". I took the K5 diving 10 years ago (!) and I needed ISO 6400 and posted https://www.dpreview.com/forums/post/50779054 on the amazing performance. 2000-2010 was a period of massive sensor improvement, the improvement was something like <10% efficiency to >50% efficiency and obviously you can't double efficiency once it is over 50%., so the last 10 years couldn't deliver the same progress.
Examining K5 shots with the same critical eye there are "portfolio" shots which don't stand up to a 2023 sharpness benchmark, degrees of mis-focus which were acceptable then would make me unhappy now, and skin blemishes which are only noticeable at 300% I'd now retouch and didn't in 2013/14. I can, just, find details in shots with the K5-iis without the AA filter using the 77ltd which the combination of FA50 and K5 with AA filter didn't capture. If you know the scene from Blade Runner where Deckard zooms in and in and in on a picture, I think there is some detail I worry about now which only shows when given "the full Blade Runner treatment". Yes it would be visible if I made prints the size of tennis courts, and got down on my hands and knee to look, but in normal use... it really doesn't matter.
Editing today doesn't give dramatically different results. 2023 software instead of 2005-2015 software, and with all the extra experience I have doesn't give many improvements, but I will do shots today which I couldn't have made work then.
I worked with softness and a lack of absolute detail before, I seem to demand more now. My (maybe everyone's) idea of good evolves with the tools being used, and pixel peeping at a 36MP DFA* 85 image shows faults which were once invisible - I can't help thinking I'd be happier if I didn't put everything under the microscope at the the edit stage.
Thanks for reading to the end. Didn't intend to ramble so much.
