[X] My Assistant

It seems, surprisingly, that the art of creating a good imageset is rare. Most imagesets are crap. Seriously, I don't understand why. I mean, I get it why most are crap, leechers just try to get few GP, but still, there are so very, very few good ones... so I decided to create a short guide on how to create what I consider to be a good imageset.

If the below seems excessive, keep in mind that it's not easy to find images for some type of content (certain rare parodies or characters, for example), so looking hard and digging in weird places is sometimes necessary. And also, I am telling you how to create a nearly perfect imageset, and that does means many hours of work (but you can fap to help pass the time, so hey, it's not that bad).

Theme

First, you need a theme. This can be either a parody (anime, manga), a specific character, or a fetish (big breasts, futa, and so on).

Completeness

Second, you should try to make your gallery as complete as possible. Because it's possible, because incomplete galleries mean you did a half-assed job, and because selecting only images you like means, well, it's incomplete. Ok, ok, even I used to draw some lines at some really crappy images, and you should probably avoid non-h stuff, through I personally think ecchi is totally fine. And for some content, usually major fetish stuff like big boobs, it's pretty much impossible to collect all images, so... Still, the bigger, the better.

Sites

Third, you need to find pictures which fit. Our e-h galleries are a start, but they are hardly complete. You need to go beyond them. Where? Several sites.

1. Start with [booru.org] boorus. You'll need more then one, because for whatever annoying reason - morons struggling for power, I'd guess - there's more then one, even for anime and hentai, and they don't all share the same images; not to mention some seem to have our equivalents of fjords, too. I used to stick to those three: [danbooru.donmai.us] danbooru. [gelbooru.com] gelbooru and of course [rule34.paheal.net] rule 34. On those sites, remember: tags are your friend, as they support individual image tagging (unlike our e-h site, to my constant annoyance). Few tips about using boorus: register for full (or fuller) access, remember their tags don't support spaces so use _, and make sure to download full resolution images, rather then downsized previews they prefer to load. So, three boorus - you think you are done? Ha. Boorus are far from complete, either.

2. Another great site is [www.pixiv.net] pixiv, where many Japanese artists upload their works (it's a bit like [www.deviantart.com] deviantart for Western artists, I guess). Register, and search - make sure to know the Japanese kanji for whatevery you are searching, too, and search for both.

3. Our site, e-h galleries, is good too, but since it does not support tagging for individual images you may need to do some weeding out (I'll have a section about it a bit later). Just download all related imagesets. It's a judgement call, but I think artistcg, gamecg and even non-story art pages from dojins (that also includes covers) should be included, too. Images are images, right?

4. Almost done. Go to [www.doujinshi.org] mugi and look for covers/sample pages for dojins we don't have.

5. Google image using English and Japanese phrases, I assume you've turned any explicit search filter off, and see what comes up. If you're lucky you may get some artist website or a fansite.

Important tips while saving files: you'll save yourself some work if you add the first letter of a chracter, parody, fetish or whatever you'll want to sort those images later. I usually sort them by character, so I'd save files with the first letter of a character name in front of whatever you get. (or another letter, if there's more then one character, or two letters, or whatever).

Duplicates and best quality

Ok, so you have a ton of images. That almost certainly means a bunch of duplicates. That will not do. So, now it's time to employ some tech solutions: [www.visipics.info] Visipics and [www.clonespy.com] CloneSpy. They will help you find any duplicates. Start with CloneSpy, which will find all identical files (hash). Then run Visipics, which does the magic stuff, i.e. shows you pictures which look similar. This will help you find downsized pics, and select the best to keep. Keep in mind sometimes pics are similar but not the same, such as the small variations often found in those annoying imagesets (cum, no-cum, open eyes, closed eyes, and so on). Be careful here.

Naming

Either before or after using Visipics, one of the last things to do is to rename and reorder the images. How, it does depend on your theme. In my case, when I uploaded my sets, they focused on characters in a given parody, so I usually named them for characters, ex. Asuna_001, Asuna_002, Suguha_001, etc. For groups, I suggest group_Asuna_Sugha_001, and such.

Such renaming is easier if done in several steps. First, use a quick image view tool, I prefer the old [www.oldversion.com] ACDSee32 2.41 (make sure to get the code to unlock the full version by googling for 30s) because it's fast (starts in ~2s) and does renames easily, but there are some other good solutions, probably. Total Commander in thumbnail mode would be my second choice, probably.

If I have multiple characters in the folder, I drag and drop them into folders named by characters (all of this in the thumbnail mode). If you did what I told you earlier and saved files with a first letter for a character, it will be much easier now. You can try to group them by series, whether by sets, similar poses, fetishes, whatever.

When they are in the folder and grouped, use the mass rename function (which ACDsee supports).

At that point your gallery, after hours, maybe days of work, should be ready to upload. Enjoy,

Sample galleries I made

Few sample galleries I prepared using this method:
* Sekai de Ichiban Tsuyoku Naritai
* Rocket Girls
* Arpeggio of Blue Steel
* Shangri-La
* Shion no Ou
* Vampire Princess Miyu
* Sora no Woto
* Blassreiter
* Mutio (Blue Submarine No. 6)
* DearS
* RIN ~Daughters of Mnemosyne~
* Kaze no Stigma
* Angel Links
* Rune Soldier Louie
* Daphne in the Brilliant Blue
* Yui (Sword Art Online)
* Sakuya (Sword Art Online)
* Alicia Rue (Sword Art Online)
* Sachi (Sword Art Online)
* Lisbet (SAO)
* group (SAO)
* Leafa (SAO)
* Suguha (SAO)
* Silica (SAO)
* Asuna (SAO)
* Ai Haibara (Detective Conan)
* Ayumi Yoshida (Detective Conan)
* Eri Kisaki (Detective Conan)
* group pictures (Detective Conan)
* Kazuha Toyama (Detective Conan)
* Miwako Sato (Detective Conan)
* Ran Mouri (Detective Conan)
* Sonoko Suzuki (Detective Conan)
* Yukiko Kudo (Detective Conan)

Bounty offer

I have a bounty for people who create quality galleries using the tips above. See http://ehentaihip.com/bounty.php?bid=13992

Comments?

By all means, if you have comments - other sites to look for images at, helpful software, and so on, do reply here and I may update this.

This post has been edited by Red_Piotrus: Feb 7 2016, 20:02

TL;DR: - Tools you will need for this (official sites for each for security reasons; if a site dies or gets hacked, do not download, also check signatures where available):
- [www.bulkrenameutility.co.uk] Bulk Rename Utility (mandatory)
- [www.f2ko.de] Duplicate File Eraser (optional but recommended)
- [www.voidtools.com] Everything (optional but very recommended)

If you try to upload a gallery that contains files with the same name, the system will only upload one of them, and reject others.

That's by design as Tenboro pointed out.

This is a problem when uploading some zip archives from DLsite or such, some artists tend to have the annoying file structure, in extreme cases it can be dozens of folders, each with a file of the same name (different image, same name, ex. folder1/001.jpg, folder2/001.jpg, etc.). This requires manual renaming (yeah, I know, batch renaming is a thing, but if you are dealing with bunch of folders with a single image, I don't think any batch renaming can help much).

YMMV, but Bulk Rename Utility (see wiki) does a perfect job with this. If you're extra paranoid (or because the software is just that good!) then you should also use Everything. And lastly, you have Duplicate File Eraser for hash-wise dupes!

If you also need to pad with zeroes, BRU can do that for you. You don't need any more than (int) log10(2000)+1 = 4 digits as a prefix/suffix (0000-1999, or 0001-2000 depending on how you choose to index).
Interesting note: Should someone grab your gallery with the doggiebag archiver, if it's padded in the same way the DBA pads the files, they will be left as is!

Let's start with Bulk Rename Utility.
- Before proceeding and if you have enough free disk space, backup the work (copy the folder, keep the archive, whatever, just make sure you can restart from scratch in case shit happens). There is no undo after you rename, so you'll have to delete the files if you screw up.
- Run Bulk Rename Utility.
- Browse to the work's root directory.
- Under Selections (12), bottom-right, tick Subfolders to display the content of subfolders as well.
- Under Append Folder Name (9), change Name to Prefix, Sep. (separator) to the separator you wish to use (I go with underscores, but your call). Default depth Levels is 1. If you need to go deeper than that, increase it accordingly.
- Now you're ready to proceed. Select all files that show up and make sure the new name matches the folder structure they come from (and there is no repeat). If there is a repeat, you have many ways to bypass this, pretty sure you'll find an original way on your own if depth itself doesn't work. It's pretty much instant.
- Once the files are renamed, read below. You'll want to grab all files in one go to make things faster, and if you did everything right, they're all named just fine so you don't have to worry about anything anymore.
- Done? Exit BRU.

If you're asking yourself where Everything comes into play: it helps you grab the files after you rename them en masse (this way, you don't have to check each folder; you can just use Everything and cut-paste into a temporary one with all files). Here are the steps for Everything:
- Run Everything. We'll assume you have an /ehtemp/ folder ready to paste the files into (you can create it now or later, doesn't matter). We'll also assume your freshly renamed folder (with now empty subfolders) is named /mystuff/.
- Tip: Add file: to the query to force a file-only search (Help>Search Syntax).
- The entire list of files that was shown in BRU will show up here too, but now you can cut (or copy) and paste them. Create a temporary folder if you haven't done so yet, then paste everything (no, not the program, all the files) in it.
- Done? Keep Everything open if you copypasted, or you can just exit it if you don't need it anymore for now.

Now, navigate to your temporary folder, and voilà! You have a ready set of files to upload, all organized, and all with unique names! You're ready to upload. Well, almost.

Need to remove duplicate files? Let's do that too. Grab Duplicate File Eraser from the official site.
- Tip: If you feel like a 3x hash check is kill (or you're doing tons of files), CRC32's the fastest but most likely to collide and SHA1's the slowest but least likely. Combine as you see fit for your specs. Using all three methods pretty much guarantees no collusion collision will happen.
- Before moving on, again, just to be safe, you should also copy that folder. Duplicate deletion cannot be undone.
- Run Duplicate File Eraser. Make sure it's the Unicode version. Don't run x64 if your system is not 64-bit. (Goes without saying!)
- Browse to your temporary folder.
- Go nuts and collision-check for MD5, SHA1 and CRC32 at the same time. You'll only be doing 2000 files at most, so you don't have to worry.
- Make sure no limitation is checked for both file and filesize (unless you need that somehow). Unless you're worried about accidentally hitting your root folder (remember, it won't delete anything until you tell it to, and there is a confirmation prompt - you can always kill the process), you can leave all search options on, they're pretty useful for deeper deduping (if you have the bad habit of hiding your stuff when this does absolutely nothing, you'll definitely want Search hidden files on, for example). If you're really paranoid, uncheck Search system files (there's hardly a reason to dedupe that except for DLLs very rarely, lol).
- Enough text! Let's proceed. Click the magnifying glass icon to Start searching. Depending on your CPU and disk speed, it should still be fairly fast. Be patient, won't take more than a couple minutes.
- Found dupes? Well, that's a good thing because that's what you're running it for! :lol: If you don't find any dupes, it's all good, exit and upload away!
- To get rid of dupes, carefully click Select duplicates (this will only keep the first version of the file namewise in each "group"). If the selection doesn't satisfy you, you can invert the selection, reverse it, or mess with it manually. When ready, make sure all is well, click Delete, and wait for the files to get nuked.
- Done? Exit DFE!

Now you're ready to upload a perfect gallery without any dupe whatsoever.

The major upsides:
- This effectively keeps the folder structure intact by appending the folder name as a prefix (and prevents the chances of any name dupe).
- You can potentially get rid of a shit ton of hashwise dupes you didn't even suspect, and you can be damn sure they're actual dupes. Very useful for imagesets!
- You don't have to archive anything, so there won't be any encoding conflicts. Non-ASCII filenames, should they exist, will be parsed properly by the PHP engine.

So bottom line, uploaders have to manually rename such files. Which works, to a point - I am sure many uploaders, perhaps even myself, missed a few files, because some archives are just a pain due to complex (assnine, annoying, dumb, chose your favorite) folder structure. (I just now uploaded a gallery, thought it had too few pictures, on a hunch checked folders, and found a few extra images with same name... how many uploaders won't double check this or miss it?)

I can reassure you on this, we all make mistakes. What's important is to spot them ASAP and fix them. Idiots will blindvote, and everyone will be held responsible, starting with the person who started the mess. (Not trying to be offensive here; people just don't pay enough attention to details. Whatever Alpha may say, DO sweat the small things, it helps the entire community tremendously and prevents people from having to fix a metric fuckton of old derps a decade later. (That doesn't mean spend one hour on something, that's just silly. That does however mean when you tag retarded shit like fleglock because shortcuts are fun - they aren't if you use them wrong, and nope, I am (politely) looking at the topmost senior tag moderator here, not you, Red :) - you should ALWAYS, always double-check to make sure your actions spawn the right tag! Now that we have namespaces, there is ZERO excuse for such situations! But yes, accidents happen. They always happen. It just becomes a problem when they happen almost all the time. ;) </rant>

I am not sure if the gallery upload does generate error reports when identically named files are uploaded, but some uploaders won't read or understand them. And this is, in the end, our loss, as we get incomplete galleries.

It doesn't, they get discarded.

A simple fix would be to automatically rename dupe files. If the uploader doesn't like how this turns out, then they can go and manually rename stuff and reupload it. This would eliminate the problem of incomplete archives.

As pointed out by Tenboro, not an option.

Well, I hope my little guide for BRU, DFE and Everything was useful. :)

- AL