Sounds like something I'd me interested in doing, wondering how much effort is required. Also curious if I would be able to work on the filter/search engine which I had been waiting for.

Thanks...
This message was last updated by the user at 09:44, Mon 26 Dec 2016.

Hi Pie,

Thanks for your interest, I'm sure the community appreciates it.  I certainly do.

Shan will probably correct me, I believe the tasks are;

1. Receive the submission (obviously).  Currently this is via email.  portraits@rpol.net are forwarded to wherever you like.
   
2. Check for duplicates.  This can be the real killer, it requires you to have familiarity with most of the gallery and to flip through the sections likely to have said duplicates.  Duplicates (including flipped/mirror images) will sneak in, but we want to minimise them as much as possible.
   
3. Review the submission;
   
   • Make sure it's content appropriate (there's been some bizarre entries in the past).
   • Make sure it's quality appropriate (no low quality images etc).
   • Do a reverse image search (https://images.google.com) and do some investigation into the image.  This might give you various bits of information such as where it came from (and if we're allowed to use it) or who the image is of (which can be good for figuring out what gallery to put it into or what keywords to use).
   • Make sure it's mainly a face (when something with a face) or otherwise cropped appropriately.
   • Make sure it's 100 x 100 pixels.  Depending on your whim you might want to edit it down to the correct size.
4. Optimise the image (while retaining quality), save as jpeg.
   
5. Upload to the appropriate gallery section (which sometimes won't be what the submitter suggests).
   
6. Update the portrait database with the new submission.
   
7. Add the search data for the new portrait so people can find the new portrait by keywords (rather than browsing).
   
8. Update the FAQ (/help/?t=faqs&page=portraitsfaq) with links to artists when required.

Currently the submissions are handled once a month and there's a limit of one per member, but that's something that you can dictate.  The only think I'd ask is something consistent.  You can handle submissions as soon as they're received and not have any limit, but for your own sanity I'd suggest not!  People used to send in massive portrait zips.

As mentioned in the other thread I think there's been around 60 portraits added the gallery each month.  Not sure how many unsuccessful that entails!

I've put as much information in as possible as I don't want to sugar-coat it and you get a horrible surprise about how much work is involved when you do actually go to do it!  If you (or anyone else) are still interested then we can chat more via rMail.

tx

I recently have started experimenting with the Mechanical Turk, not going anything serious with it or anything like that but sometimes I have 10 minutes to kill so why not make 50 cents or whatever?

Anyway the point of the Mechanical Turk is it takes small tasks which are very difficult to automate such as rewriting a sentence or writing descriptions for images and breaks it into a large batch of many micro tasks that don't take more than a few minutes to do.  In this way it simulates automation of the non-automatable (is that a word, it is now I guess).

Knowing absolutely nothing about the back end of RPoL, would it be a viable solution to this problem to cut the portrait submissions up into lots of 3-5 and crowdsource the tasl?  Surely the email submissions can be collected and uploaded automatically.  You'd still need a reviewer/moderator, but the task would be to verify rather than to go through dozens or hundreds of submissions.  In fact the review itself could be in the microwork format.

I lack the technical skills to describe exactly how one would do this, but I have to solve problems with processes and people not having the resources to process workloads pretty often and part of that process is brainstorming alternative approaches.

That all sounds good, thanks for the info. I would definitely be interested in taking on the task. Hopefully I can get a basic idea of what makes an image something we are allowed to use before I start. I'm thinking the work of an artist requires permission, but I am unsure about photos. The few images I added to the list were clips from old paintings, which I always assumed were public domain. That kind of information there...

Anyway, sign me up if that's acceptable. The bigger a catalogue of pics the better, I've always thought.

drewalt - I'm not sure what the pricing structure is for the Mechanical Turk, but I can't see requesters being willing to pay to have their submitted picture vetted, much less if it turns out it can't be uploaded.  Nor is RPoL able to process such payments yet (outside of the donations to cover server costs).

- resize the file to 5K -- it helps if you've got software that can do this

- edit the image for clarity at 100x100 pixels - a lot of the time this can involve playing with contrast, brightness, and also smoothing edges for drawn portraits as these do not resize smoothly or well.

-keep a detailed list of numbering.  If you screw up the numbering of the portraits when you name, it will screw up the gallery.

-pics must be converted to *.jpg, and the extension must be in lower case or it won't show up

-be aware of copyright issues as possible.  For instance, Disney does not allow the use of its images on other websites (no Disney pics in the gallery) and some artists specifically state that they do not allow the use of their images (Larry Elmore, the guy who does 'Order of the Stick')

-keep in mind that you're not only risking problems for rpol, you're risking problems for the volunteer hosts of the portraits

-uploading the portraits will require software or an app for FTP

-remember rpol's requirements - no text on pics other than watermarks, no nudity, no duplicates, no altered pics of artist's work

-also be aware that you're going to get stuff sent to the portrait emails
that has nothing to do with portrait submissions -- some of it not very pleasant

-maintain a copy of the portrait gallery on your local computer.  The portraits are hosted on volunteer servers, and you never know when you might need to reload, either because the host changes servers or there is an issue at the host that causes data loss.   Not  to mention what happens when you make mistakes yourself ...

-keep in mind that changing a currently existing portrait may cause problems for users of that portrait.  Don't do it unless necessary, and if you do, figure out a way to minimize the effect for anyone using the portrait
This message was last edited by the user at 13:28, Mon 26 Dec 2016.

Well I wasn't proposing the site pay rather asking if there was a way to cut this task into small jobs so site users could do most of the work rather than drown one person in it.  Honestly it sounds like too much for any one person to do.

I'm on board with Piestar and drewalt. Is there some way that we could break this up and make it easier? Maybe have one person who does portraits for category A and another for category B? Or are there any other suggestions for possibly doling this out amongst a group of volunteers? Shannara's been amazing at handling it all, herself, but it might be more attractive to future volunteers if there were a "portraits (men)" editor and a "portraits (women)" editor, or one for fantasy/historical and another for modern/sci-fi. Perhaps there's some background reason why that won't work, but otherwise, it could help spread the load.

Either way, I'm still interested in helping. I'm used to tweaking my portraits prior to submission, anyway, in an effort to both maintain the specific image I want and to minimize the work for Shannara; so, I know the editing software and the gallery rules (and boy, am I good at keeping lists!). About the only thing I don't already have a strong familiarity with is the FTP software, but I at least understand the basics. If there's any way I can help, I'm in.

Still have the problem of control - what guarantee do you have that every single person of the crowdsourced mob has exercised the appropriate diligence ?  The more there are the less confidence you can have in the outcome.

But yes, a small dedicated group as Morgan Coldsoul mentions could certainly work - spreads the work out, even if only a little, but gives the required level of assurance/control.  :>

In reply to Skald (msg # 9):

To be honest; given we're going into a new person doing it regardless, about as much guarantee as you'll have that your one single person will exercise that due diligence. I'd share your concerns a bit more if this was someone who'd been doing it for a while who was looking for someone new to help out; but given it's someone new regardless; this might make it more likely that you can get people to help out.

You don't think that a few people who volunteer for what's a fair amount of work/effort will be more reliable than a LOT of people down for a half hour here or there ?

That's fair enough - I'm certainly no expert on human psychology/motivation/etc - but personally I'd rather rely on the dedicated and committed few than the enthusiastic and well-meaning many.  :>

But you don't know -who- is going to be the dedicated few until you start.

In reply to willvr (msg # 10):

The flipside of this argument is, if you only have one person responsible, and they don't cover the job with due diligence, you know exactly who's responsible and where to address any concerns over shortcomings in the work.  If you've got a small committee, the responsibility is shared around, so you could still wind up playing 'Monkey in the Middle' trying to find out who cleared a copyrighted image or something of the sort, but the options are still pretty limited.  When you crowd-source it, you open the door to a whole lot of "Okay, so who dropped the ball here?" without any clear answers.

I suspect, rather than having multiple people processing each round of requests (because part of the problem is resolving the question of where, exactly, each portrait would be best located), it would be more efficient, in some ways, to have a small corps of portrait processors, who rotate the duty of handling submissions.  That way, nobody is stuck having to tackle the whole mess all the time...it's only every other month, or every third, or however you spread them out.  That option could also enable the possibility of doing submission processing more often, so the list doesn't get so big...say, you have four people, and they rotate through processing all the requests at the end of a week, so nobody does it more than once a month, and instead of having to vet sixty portraits and size and sort them, they only have 10-15 or so.

Not saying it's the way to go, for certain...but it's an option that may be worth considering.  Keeps the pool of responsible parties small, but breaks the job down into a more easily-manageable size and helps prevent any one person from becoming overburdened with the task.

I waited to reply, as I have great interest in the gallery but wanted to see what developed here. I too am very willing to pitch in and help with the gallery. I have time to devote to the effort, though storing the gallery might be problematic for me; I would have to see how large the entire gallery is ... thirteen thousand images could take a sizable chunk of a hard drive. Of course if it does not have to be "live" I have an external drive I can lend to the cause.

So count me in as a volunteer as well. I think a small team, just taking images as they drop into the submission basket, as they have time, and then releasing them to the leader of the task, who is ultimately in charge of vetting the whole operation would be the best idea. Then there is a consistent quality to the whole operation. Whereas dividing up the tasks would lessen each participants work load, BUT would not guarantee that all the tasks are done in a similar manner.

For everybody wanting to help, you can join Portrait Posse. I hope that doesn't count as getting people to join a game. I did it for awhile and searched for duplicates. 

One thing I think might help is if portraits of real people/celebrities are labled with their names. It might cut down on duplicates and similarities in the future stock. And how many pics of Sarah Michelle Gellar and Michelle Tractenburg do we need? :)

In reply to Grimmond (msg # 14):

They're under 5k each, so under 100 megabytes storage.


Still thinking about the best way to organise this.  facemaker329 is right, we have to take care we don't lose quality control plus there's the concern of consistency.  As Eggy suggested, I think hijacking the Portrait Posse discussion board would be the most appropriate (or at least; the best) place to throw around ideas.

I cleaned out the cast list of the old Portrait Posse board for use and opened up a thread there.

Everyone who has volunteered, please request access there, and I, jase, or one of the other mods can add you to the board.