Want to take part in these discussions? Sign in if you have an account, or apply for one below
Vanilla 1.1.10 is a product of Lussumo. More Information: Documentation, Community Support.
The sickness of a webserver at Bangor means that theses of n-POV interest and use to n-lab readers /users are currently not available. This raises a question as to whether we should not have a repository / archive on the n-Lab with links to theses (not only Bangor’s but generally). Would there be any objections to this? (Only theses with clear access to everyone would be put there, so no problems there, but some may be scanned and that gobbles up space.)
My draft structure would be a page called theses with a list of those available, then, where possible, a small summary (lifted from the thesis) of each with a link to an archived version. (If the arXiv already has a copy this would just be a link to that, but some may not be available in that form, e.g. being too old.)
There are some free online resources which allow for sizeable repository of shared files. Maybe we should have some stuff at places like that for a while with links somewhere at nLab. Once the repository becomes important enough, one can estimate the scale and the purpose and tie it more closely to the future backbone of nFacility. One day we can grow large and important enough to justify applying for grants and other infrastructure help. Unplanned large addition of files to the central infrastructure of nLab should be less preferable at this point; slowing down the backuping and other things. But if we have a side structure like a linked repository somewhere near by growing and justifying future use, we can estimate its value and invest thinking and resources into it. This is hypothetical thinking, I am not one of the people maintaining the infrastructure (yet at least), so it is just a suggestion.
If scanning, the modern and free format of djvu should be used: for the same quality, if at more than 300 dpi it will have better compression than other, older formats, like pdf.
Use Any2DjVu to convert many older formats into DjVu. You can get OCR with this service too, making the file searchable (although it can’t handle math, or even French).
Converting into djvu is worthy if the original is in 300 dpi or higher. If you have below 200 dpi scan in pdf it will result in worse djvu per unit size than he original pdf. This phenomenon described by some djvu people is that in such a case it is better to print it, rescan it at higher dpi and then convert. This does not sound logical but some of the official djvu FAQs say that.
If you have below 200 dpi scan in pdf it will result in worse djvu per unit size than he original pdf.
The usual DjVu converters don’t allow one to create a DjVu file with lower resolution than this. I wonder which is cause and which is effect here.
This phenomenon described by some djvu people is that in such a case it is better to print it, rescan it at higher dpi and then convert.
Printing can’t create information that’s not there. Possibly printing at a low resolution spreads out ink so that the scanned result looks better, but if this works, then one could in principle write a computer program to do it virtually. At best, this advice is a work-around until someone does. It almost sounds like the sort of thing that somebody would say as a joke, with the point being that something at such low resolution really isn’t a document suitable for convenient reading at all.
I think it is something more subtle about the algorithm used for djvu compression, that it starts being worse in comparison to pdf when the things are more granular in input, but I don’t know.
1 to 6 of 6