Enforced Naming Convention for all newly uploaded AMVs

darkpeace · Post by **darkpeace** » Sat Oct 16, 2004 4:57 pm

Enforced Naming Convention for all newly uploaded AMVs

Would be good if all the AMVs where actually named in a useful manner.

eg: "AMV <Author> - <Name given to AMV> - <Animé used/Various> - <Band who released song> - <Song Used>"

ie: "AMV - DarkPeace - SomeVideoIMade - Various - Cranberries - Zombie"

This is useful in so many obvious ways and some less obvious ways.

If I host AMVs on WinMX, I rename them all to add AMV to the front, so that if people search for "AMV" they will get 100's of hits from myself.

It also means if you search for say "Evangelion" in your Media Player application, then all AMVs which are mainly Evangelion based will be displayed.

For <Animé used/Various/Many>, you usually want to list up to 3 animé, before classing it as 'Various or Many"

Just an idea, but I think ENFORCING the above naming convention would be a grand idea.

(As would author title, info frames, etc in the movie file, but to many AMV authors don't do this, and the lesser quality clips get deleted eventually)

It would be esp useful for users of the D-Link DSM-320 (See D-Link websites) , and other similar devices, for watching AMVs on their Televisions.

Org Profile · Post by **LantisEscudo** » Sat Oct 16, 2004 5:14 pm

This topic came up recently here, and derobert in this thread noted plans to allow users to specify the naming convention they wish to have their downloads named to, which removes the need to enforce a naming convention on uploads.

Your naming convention, while complete and informative, can cause problems when the files are burned onto CD or DVD (as I do), since the ISO9660 and UDF formats have, respectively, 221 and 256 character limits on their filenames, forcing the names to be truncated if the title of the AMV, anime used, and song are long. I've run into truncation issues even with my personal series-title-author convention.

darkpeace · Post by **darkpeace** » Mon Oct 18, 2004 3:54 am

Personally I would impose a limit of 128 characters and use a TEXT file if it needs to be longer than that.

The files do support alot of tagging information aswell, and it is a shame to see even the better quality vids not using it (thus I am staring to award a 4 to videos that are 'incomplete')

A truncated name could still be used, I am sure that given a standard format a simple rename application could be coded up so all parties win out.

Even having a TEXT file of the same (truncated) name (.txt) with all the information in it would be a very useful idea also.

Also a database with all the original names, sizes in bytes, MD5, and other additional information (if available) would be VERY useful.

I have duplicates of some of the files (same name, same size, different MD5), obviously one of them is corrupt in someway (they all play fine), but if I share AMVs via WinMX then I want to share the version the author uploaded with the correct MD5.

(For info on MD5, just goto www.google.com and search for "MD5", it is similar to CRC32, but far better in many respects)

I am sure there are around 3 to 4 terabytes of AMVs here, so the benefits and feautres of MD5 would be greatly useful.

Also is there any way to get all the current AMVs ? (I am willing to pay) or at least all the 3,4,5 star ones, and any decent ones of note.

Brsrk · Post by **Brsrk** » Mon Oct 18, 2004 8:07 am

Or just rename it yourself...

darkpeace · Post by **darkpeace** » Mon Oct 18, 2004 8:16 am

No offence, But was there even meaning to that post ?

Lets just assume you've got 3 Terabytes (3,072 Gigs) of Anime Music Videos, just how the hell are you meant to rename apx 45,000 files.

***Hell if you are volunteering yourself for the task go ahead, obviously your post indicates what a simple task this is.***

OR the authors could just spend a minute (if that) naming the clip.

Sure as hell beats sorting through 150 days of NON STOP footage.

But since you volunteered yourself, I'll get Phade to tie you up and feed you nothing but water crackers & old bread and you can slave away for a few months since you just advised what a simple task this clearly is.

Any ways back to reality, does anyone have any non spam replies that are meaningful and thought out ?

Declan_Vee · Post by **Declan_Vee** » Mon Oct 18, 2004 8:29 am

Or the creators could enter accurate information to the database and people who download can use this to rename the files themselves.

Org Profile · Post by **LantisEscudo** » Mon Oct 18, 2004 9:36 am

I agree with Declan. If you have a naming convention you use for the AMVs you watch, the onus is on you to rename your downloads to match your convention, especially since your convention might not be the same as anyone else's. The only obligation the creators have is to make the filename uniquely and recognizably identify the video, not something like "myamv.mpg" or "awesomeDBZvid.wmv". As I noted above, derobert already has plans to implement customizable filename conventions for local downloads, so your downloads will automatically be named to whatever fits your convention.

I'm also a believer in "One Video: One File." If you want selectable subs or something similar, use a format that supports it, not a separate file. Having a separate text file with the video's information seems redundantly useless to me.

Dinging videos for not having the tags filled out seems a bit petty to me. I don't fill them out or care about them because I never see them. I configure all the media players I have (that support viewing the tags at all) to not display them, since they're just screen clutter to me. I don't share things on P2P networks (in fact, I don't use P2P networks at all), so having all this information for other people to see doesn't seem that important to me. As long as I can identify the video's contents from its filename, that's enough for me.

The idea of having an MD5 checksum to verify downloads is an interesting one, though. I don't know how computation- or disk-intensive the algorithm is, but having the server calculate the MD5 of videos when they're confirmed and listing it on the download page (it's only, what, 128 bits? So it's an easy copy/paste to the utility to check) could be useful. People that really want to make sure they got the file correctly can use the checksum, and people who don't care can just ignore it.

trythil · Post by **trythil** » Mon Oct 18, 2004 9:43 am

This is easy to do on a per-user basis for the more <strike>anal</strike> detail-oriented among us:

http://us2.php.net/manual/en/function.header.php

As far as MD5 goes, I don't see a need for it. Two reasons:

1) MD5 summing is slow. On my Athlon 64 3400+ (I'll give you more specs if you really want it) it takes around 1-3 seconds to give the MD5 sum of a 50-100 MB file, respectively.

The .org systems run on a quad Opteron configuration, but they also have a lot of data to comb through as well as a lot of uploads per day to work with.

This wouldn't be so bad if there was an actual need for MD5 summing. However, as you point out...

2) There's no real need for it. Files that are identical save for MD5 sum, as you stated, play back just fine.

We're not sharing source code or anything similarly critical here. These are fan-made music videos. If this was a general-purpose download site then I'd probably agree with you, but as the Golden Donut currently is there just is no reason to toss on an extra layer of metadata.

trythil · Post by **trythil** » Mon Oct 18, 2004 12:09 pm

trythil wrote: 1) MD5 summing is slow. On my Athlon 64 3400+ (I'll give you more specs if you really want it) it takes around 1-3 seconds to give the MD5 sum of a 50-100 MB file, respectively.

The .org systems run on a quad Opteron configuration, but they also have a lot of data to comb through as well as a lot of uploads per day to work with.

Quick note, because I just thought of it:

This, obviously, isn't a formal proof of "MD5 is slow", and I probably shouldn't have written that. You can't say "MD5 is slow" in any really strict context without having done a formal algorithm analysis beforehand.

However, the empirical data should provide some indication that MD5 adds a non-trivial amount of work on the system, which I don't think is offset by any benefit it might provide.

YouTube · Post by **derobert** » Mon Oct 18, 2004 6:58 pm

trythil wrote:1) MD5 summing is slow. On my Athlon 64 3400+ (I'll give you more specs if you really want it) it takes around 1-3 seconds to give the MD5 sum of a 50-100 MB file, respectively.

That's not slow. We have slightly under 20,000 local downloads. Assuming 4 seconds each, that'd take under a day to md5sum them all. That 4 seconds is even more insignificant when compared to the hour or so it takes an average broadband connection to upload the video. And it's CPU time on Nago, which is a dual xeon with almost no processing to do. Even at 100 uploads a day, it'd be all of ~10 minutes of CPU time.

[BTW: Org has 1 dual opteron, 1 dual xeon, and 1 P4. More servers coming soon. ]

2) There's no real need for it. Files that are identical save for MD5 sum, as you stated, play back just fine.

Not really. Normally, you get an artiface of some sort from the file corruption. As the file gets copied through more and more defective machines, the corruption builds. md5sums give a quick way to be nearly certain if corruption has happened. [I've seen videos and audio from file sharing programs. The bit errors tend to accumulate until the quality is atrocious, especially on file sharing programs that don't internally md5sum the downloads.]

Second good use for it is to check if your upload is successful. Right now, you have to re-download the file you just uploaded. That takes at least 5 minutes. Instead, you could do a several second md5sum calculation.

Third, storing md5sums would allow .org to verify the integrity of our copies of the videos. Right now, we really have no good way to confirm that the video has not been corrupted.