Apologies, I thought I answered this a couple of days ago...
A more nuanced list would certainly be helpful but to be honest I'd rather keep it here than in an external store (just my biases).
I've just purchased, from the great auction site in the sky, two of the units on the list - the epson 4870 and the HP G4050. I'll see which I prefer (I think the 4870 may require the glass replacing) and then the other might find its way back onto the bay. It'll be interesting to compare the two - apart from anything else, the 4870 is 16 bit and the 4050 only 8 bit. Which is not to say that the one is better than the other, but there are sound signal processing reasons to have as many bits as possible on the input stage, even if the final display is only going to be eight bits deep. But that's perhaps another thread.
Neil
Recently I took apart a 4870 and after a good clean had the oportunity to test it. Inside it looks just like a V700. Comparing scans of the same negative (4x5 color negative), using Epson Scan for both, they are very similar in performance, the 4870 being a bit better on the focus issue.
My stuff for sale is here: https://drive.google.com/drive/folde...-0?usp=sharing
Focus can be corrected by adjusting holder height.
V700 should deliver slightly better performance capability for vertical bars, as it has 6400dpi (hardware) vs 4800 dpi of the 4870, delivering slightly better practical resolving performance, IMHO not much, at all.
Practical max performance values are between 2300 and 2800 optical dpi, depending on x or y axis an on other factors like microcontrast, flat negative, etc.
This is around what a good LF lens can resolve if workig in a very good shooting situation, so at the end it is not much a limiting factor, but close of it, IMHO.
Those flatbeds are less ideal for 35mm film, where a fraction of the shots may outresolve the scanner capability.
Ah, my mistake reading a spec somewhere.
They haven't arrived yet, anyway...
Mathematically (from a signal processing point of view) one should use the default resolution of the sensor element, rather than allowing the interpolated/decimated output from the machine. There is no anti-aliasing optical filter in the system other than the limited resolution of the film/lens combination, but sample rate conversion is *not* as simple as replicating or discarding samples, or even averaging them.
[If you really want to know: first find common divisors of the input and output sample rate. If the default is (say) 2400 ppi and you want 2000 ppi, these ratios are 6 and 5. Expand the input signal by the output ratio, inserting four zero samples between each actual sample - now the signal has five times as many effective samples.
Low pass filter the result to the Nyquist limit of the input sample rate; in this case under 1200ppi (probably 1100-1150ppi to make a usable filter).
Now decimate the filtered result by discarding five out of every six samples - reducing the sample rate to the desired 2000ppi and filter again to the Nyquist limit of the output - nominally 1000ppi but more likely 900-950ppi. You now have a mathematically accurate, and if you used the right filter, a phase correct output at the lower pixel rate. The critical thing is that you don't want any frequencies in the output signal which were not there in the original, or in the case of reducing the sample rate, that would otherwise alias in the output.]
Neil
I have a microtek i900. It serves my purposes. It cost around 80$ off Craigslist.
First observation: the HP G4050 covers up to 'about' 8x10 on the transparency unit (with holders for all the usual suspects: 35mm strip, a guide to locate 51mm square (mounted slides?) and 120 strip and 4x5).
Normal (print) use looks fine, but... there is apparently no TPU driver for linux, so no negative scanning. And I have been unsuccessful in installing the HP software in the W7 image running in VirtualBox. So not a lot of use for me, so far... possibly Vuescan, but while I can make the unregistered version do a flat scan, it silently fails when I try and do a transparency scan. No idea why. In the windows VM, Vuescan fails to see the scanner at all.
Neil
My stuff for sale is here: https://drive.google.com/drive/folde...-0?usp=sharing
Bookmarks