Hi =)
Yes, I'm dumping blind, but I'm capturing the raw stream format, and ADF, using a Kryoflux. The kryoflux does a pretty decent job of attempting track rereads (with a head reposition) when the track looks like a valid Amiga track, but has bad checksums when read etc. With the raw mfm stream, I can go back and have another go at any disk, if I'm willing to attempt an alternative interpretation of the stream data. It's pretty much the best option I had for reading via automation, barring running the disks through multiple brands of disk drive etc.
The vast majority of the disks are Amiga, and are data, (a lot of sound samples, older public domain disks, source code, midi files, soundtracker/octamed, etc). Then there are the inevitable stack of 'yet another cracked version of blah', some of which have been written over originals, in addition, there are some original games/apps in there, a LOT of coverdisks, and a decent smattering of MSDOS disks, since if you are going to read disks.. might as well read all of them =)
As for quantities.. once I'd finally finished digging disks out of cupboards, boxes, etc.. I read just over 4000 disks, the 5000 figure being passed around being an estimate based on how many disks were in each crate etc.. that's still way too many to read by hand =) Of those, around 2000 were my own disks, 500 or so belonged to a friend who loaned me a catweasel to try to read them (and made me realise doing it by hand would be a bad plan...), and the remainder were 'free' when I picked up a 2nd hand A1200 at a carboot sale.
I've now got a Java app that scans the dirs, lets me navigate by the disk image, and view information about the disk. The app runs ADFINFO against each adf, and associates the output to the jpeg, this gives me the volume name for any proper amigados disks. Also gives basic info on checksum failures, limited virus detection, filesystem type, and a crc32. In addition the java app calculates its own crc32 for the image, which over 4k disks has disagreed with the AFDINFO crc32 once.. which is odd.. someone has an error in how to build a crc32.
The app loads all the TOSEC amiga dats, and builds an internal map of crc32 to disk(s) (allowing for clashes). The map is then used to attempt to identify the adfs, where present in the dirs. This approach gives a hit rate of about 1 in 5, that leaves the other 4 in 5 as being not known to the dats, requiring manual handling.
1 in 5 is a good sign, enough to know the rig is properly reading disks.. the 4 in 5 includes non Amiga disks, which are not expected to match, and includes many many disks that likely are in TOSEC already, but my version is altered, either through virus damage, bitrot, icon timestamp updates due to being used write-enabled, 'protective' bootblock installation, different intro/group, etc.
Once I'm done, I'll export a dat from the app containing checksums & info for the disks I have, that other people may have. I doubt you guys will want many (if any) of the entries from it, but it may be of use to anyone following in my footsteps in years to come.
I may make select subsets of the disks available online afterward, especially if they are content not known to TOSEC, but before I can do that, I've got to tag the collection to remove the ones with my old files on =)