20171104_0014_Working_20200327_3200dpi.jpg
20181104_0014_Master_202001011_20201025_Working_4000dpi
“You often feel tired, not because you've done too much, but because you've done too little of what sparks a light in you.”
― Alexander Den Heijer, Nothing You Don't Already Know
I'd seen Daniel's project before, but not yours. I'm taking a shortcut by using one of the cheap chinese CNC kits and re-purposing it.
Yep, that's a good option.
Daniel's design was compact and very sturdy, as the system clamped the lens in place.
I used a Velmex 4000 unislide to position the camera + lens system. That was more complex and expensive, but it allowed me to change out lenses. I tried about 20 different lenses, and I ended up preferring a Rodagon D 75mm 1x lens at 1x magnification. I expect there are better options now...but even with the ancient 55 Nikkor, the system outperformed an Epson flatbed, and that was tested using resolution targets and stepwedges for density measurements. My system outperformed the Epson by about 1 stop more dynamic range. Today's cameras can likely do significantly better, and that's not even considering multiple exposures and different exposure levels. I tested that back then too, and it gave great results, but it did require a lot of extra exposures and processing.
“You often feel tired, not because you've done too much, but because you've done too little of what sparks a light in you.”
― Alexander Den Heijer, Nothing You Don't Already Know
Which is the best setting for the V850? In my test of it vs a Howtek 8000, I did the scans on my V850 using Epsonscan. The first scan was done at 4800 ppi reduced to 4000 for the first comparison set. The second scan was done at 2400 as I found for some reason it was sharper after sharpening than the 4800 scan. That could be an error on my part. Both scans were adjusted to set the clipping points only.Originally Posted by grat
True, in all regards. However, the v800/850 holders are still "stepped"-- 5 steps instead of 2 (and an irritating tendency to shift when I'm loading film), but the screw-type adjuster on the (for instance) betterscanning holder, is superior. Unfortunately, he's still apparently either catching up, or incommunicado.
In the case of the Howtek, it's got multiple sensors for each color, and (I assume) uses the rotating drum to pass the negative across each sensor, meaning each 1/4000th of an inch of negative passes over three sensors, instead of the Epson which uses 1 sensor to decode all 3 colors at once (which can be done, but it's effectiveness is entirely based on the algorithms that are trying to recreate that color based on three (four-- RGGB) adjacent grayscale readings).
The native resolution of the v800/850 appears to be 3200 PPI (in the central high resolution band), with it appearing to be able to do a "pixel shift" type maneuver to reach 6400 PPI. 12,800 PPI is just making up (interpolating) pixels to go in between the 6400 PPI results. If it did three pixel shifts (one for R, G and B) it might produce a more detailed scan, but otherwise, the effect of the bayer demosaic process will reduce the resolution slightly. By scanning at a higher resolution and "binning" the results down to a lower resolution, Steve is effectively raising the signal-to-noise ratio of his scans, and throwing out some of the less reliably interpolated results.
In theory.
My biggest gripe with the state of film scanning in 2020 is that the difference in RAW processing between my 2007 Canon EOS 30D and my 2019 EOS 90D is truly breathtaking-- the difference in sensor processing between the Epson v700 from 2006 and my Epson v800 from 2015 (purchased this year) is... Not much.
Here's the complete thread for the comparison of the v850 vs Howtek 8000.
https://www.largeformatphotography.i...ghlight=howtek
Flickr Home Page: https://www.flickr.com/photos/alanklein2000/albums
Did I have to reduce the 4800 to 2400 first? to 3200? something else?
Flickr Home Page: https://www.flickr.com/photos/alanklein2000/albums
I honestly don't know myself. I've been doing various tests, trying to reduce the number of variables, but I've lacked *ahem*... focus.
Personally, the variables for me are holder height (reasonably sure I've got that one nailed down), scan resolution, down-sampling, and film quality-- and right now, I'm not convinced my negatives are of sufficient quality to make definitive judgments.
Gonna start with buying some already made from Jason Lane. Then start making my own. I have like 20 plates. The hard part will be coming up with an interface from camera to scope and of course focus.
The way to reduce the ppi is to do a 2x2 binning. This is different than resizing in photoshop or other programs. 2x2 binning turns 4 pixels into a super pixel. Reducing the file using photoshop or epson scan will yield inferior results. To my knowledge only Vuescan offers this option unless you use a different program.
I honestly don't know myself. I've been doing various tests, trying to reduce the number of variables, but I've lacked *ahem*... focus.
Personally, the variables for me are holder height (reasonably sure I've got that one nailed down), scan resolution, down-sampling, and film quality-- and right now, I'm not convinced my negatives are of sufficient quality to make definitive judgments.
Focus is number one. From there, scan at the highest resolution you can do and then do a 2x2 bin before saving. That takes a 6400 to 3200. Vuscan is the only program I know that can do that right now. If you have to save the file and use a different program, you will need to keep the file just south of 4gb for a tif file. Do not compress, even if you use lossless.
Don't buy into the scan only for the size/s you plan to print out. The smaller the dpi the more information you throw away. Which is not true for binning.
I believe you could. Have not tried it yet.
Posting Permissions
Reply With Quote
Bookmarks