I'm sorry that I'm not familiar with the twaindotnet project, but I do have a lot of experience with document scanning through TWAIN.
First a note: Not all document feeders can feed in single-page mode; some important scanner families always scan everything in the feeder once started. And, quite a few TWAIN drivers won't honor XFERCOUNT=1, no matter what the standard says.
If you try to solve the problem by forcing the scanner to scan "one page jobs", you will be limited to the (indeterminate) set of scanners that happen to support that. The TWAIN standard just doesn't require this feature. (But yes - CAP_AUTOSCAN=FALSE and XFERCOUNT=1 would be the combo to try.)
There is a better solution (time & patience permitting) - It sounds like what you want to do is process and dispose of each image as it arrives, instead of collecting them all in memory. Figure out how to get your TWAIN library to hand you each image (or write it to a file) as it arrives instead of stacking them up in memory, and you'll have a solution that works with all document-feeding scanners. And it will scan quite a bit faster with most scanners, too...
与恶龙缠斗过久,自身亦成为恶龙;凝视深渊过久,深渊将回以凝视…