What's up this week? - 11/5/2012

That's an interesting idea. My tenets of localization are:

1. Don't put text in images if you can help it
2. Don't put words in the game if you can help it
3. Don't localize until the game is completely done, definitely for sure

But if you have potentially infinite localization power, maybe you could set up a web site with all your strings in English, and let people enter translations for each string in each language, or vote for an existing one. Unless a large part of the community makes a concerted effort to screw it up, you might come out with a free, nearly real-time localization service that you can use from day 1.

I think this is to big for UBER to control every day.Nice idea but too uncontrolable.


Question for the last Planetary Annihilation update(livestream 31.October)
What the hack does "give us a little spiel on ot" mean? specialy the word "spiel"?
Does he mean the german word "spiel".
On 0:12:18

spiel (spl, shpl) Informal
n.
A lengthy or extravagant speech or argument usually intended to persuade.
intr. & tr.v. spieled, spiel·ing, spiels
To talk or say (something) at length or extravagantly.
[German, play, or Yiddish shpil, both from Middle High German spil, from Old High German.]

Yay, map editor! I want to test it and I hope we can find easy ways how we can combine the procedural generation of maps with custom art assets. I really enjoyed using custom textures, decals and skyboxes in SupCom/FA.

Ok thanks, I got it now.

I like the way facebook does it, allowing users to enter translations and asking people if the translation was good, or which of the translations seems best.

But I think for the scale of this game, having them translated by an office might be cheaper than paying a developer to set up a website for decent community - based translation...

One of the biggest challenges in localizing games has always been context. While context is kind of slippery in English, in other languages, it often has huge impacts on the grammatical structure and specific word use. Even more so when you get into Asian languages.

I'm doing an ongoing investigation into the best localization techniques, but where I'm leaning is toward first pass localization by professionals. That gets you about 80% of the way there. But even when you spend a lot of time to provide the context for each string, without the translator sitting there, playing the game, there's a good chance they'll get at least a few wrong. The lack of the context of the game running makes it hard, and I've never worked with an outsourced localization team (or even in-sourced much of the time) that can or is willing to run the game while they translate - they do too many products, and too much translation to try and learn a new piece of software every time they want to do a project.

Where using the crowd can come in is after that first pass. Taking all the existing string tables, presenting them to a community, and letting natural language speakers who are also playing the game have a chance to vote on, or even provide better alternatives, for translations.

The next obvious step after that is - what about the languages that don't make economic sense to make, but people still want? Say there's a call for Portugese, but we aren't sure about spending tens of thousands on a Portugese translation. That's where providing crowdsourced language packs can come in. If we find that the Portugese community pack is hugely popular, and actually had 200,000 people using it, well.. for the next project, DLC, or whatever, it becomes a pretty big no-brainer to say - we'll add it to the list of first pass translation services.

Localization has always been one of the biggest x-factors in game development that isn't a key part of the development process. What languages to support, how to cover as many people as possible without spending an inordinate amount of your budget on translating to languages that no one will use anyway, and how to get translations that are actually contextually accurate.

And obviously, for a game like this, localization will be pretty important, as will be covering a wide variety of languages. RTS games are actually as popular, if not more so, outside the US, and so it's easy to imagine the percentage of non-English speakers be a plurality of the gaming population. EFIGS (English, French, Italian, German, Spainish) is the default group of translations most studios do. But I can see, at a minimum, Russian, Portugese, Korean, and likely Norwegian and Finnish also being important. And quite possibly quite a few other languages as well!

And once you add in Cyrillic languages and double byte character asian languages, you add technical challenges to the existing pure contextual challenges.. I could go on..

As I worked on an internation game myself we used XML files which stored all text. The community was asked to do translations. The players only had to copy the XML to a folder to enable a language (even imaginary languages like pirate speak :mrgreen: ).

I think this could work for PA as well.. store every word, even on buttons and such, in a XML file and let the community do their part..

Yeah, can you try to get that in before the beta start plox? I have a feeling that those special characters are resposible for lots of strange issues if you don't support them from the very beginning. Special characters in the Windows profile paths - always fun...

Absolutely is the plan. I actually left off an important one.. Arabic. THat one is a much bigger technical pain to support than any other script because, as William reminded me, the multiple forms each character can take, depending on its location in the text, on top of the right to left nature of the text. WHEE! Though cool, as I've never worked on a title that could even consider supporting Arabic script.

My hope is that we'll have some of the early stages of this in by alpha, or at least _before_ beta, as even though this won't be a text heavy game, it'd be awesome if, at launch, it had partial to full translations of every language that has a few million + speakers (about 20 or so different languages I think). Optimistic goal, for sure, but I'm hopeful we can at least get partway there.

Outsourcing translations to a team, then getting feedback from actual players - that sounds simple and sound I'm certainly in to help if you want to do a Dutch version!

And good luck with the Arabic. I'm glad I never had to deal with that in my software projects. Being an RTL language, it's not just different alignment, it's sometimes also different positioning of UI elements that may be needed... Yuck.

So there are two levels to the editor. The procedural generation part which generates a whole bunch of output stuff and then all of the data that actually defines the procedural generation itself. Since the procedural generation phase is separate from the actual final geometry generation the maps can technically be edited offline by modders after they've been initially generated. Of course the intent is to have the front-end be scriptable so people could potentially extend our editor with more functionality as well. In addition people will be able to add their own geometry, textures etc. and definitions for new planet archetypes and terrain biomes as all of this is driven by json text files.

Any pictures yet? :roll: (I find myself having a hard time remaining patient)

I'm fascinated by the level of user accommodation you guys are putting into this game. You are really throwing everything you have got to make the game as user friendly as possible!

That sounds so sweet that I might need a new filling! =P

Thanks neutrino!

I hope for some moving visual stuff. Videos are better than pictures because you can see it realy happening.

Sounds interesting . Its really nice to see how a game is forming from the start, and gives much more understanding what effort is being put into a game.

Something i like to know is. It looks like you guys having"deep" moding as an option for the players right from the start on your mind.
Is it rather "difficult" to keep that in mind or does it actually makes your life easier later on.
Because it always bothers me when i see some games were you can make the game into something completly different and bring it to a new "level", or a modder is very limited in what he can do.
Is it a question of extra time needed to make it moddable or is it normaly a publisher descision on how much you can mod the final product?

And thanks for the updates you give us

One thing that would be great to have is to be able to edit the generated map and then distribute the map by just storing the generation parameters as well as only those pieces of the map that have been manually altered.

It would allow easy distribution of maps that can still be partially hand-made.

This is what Valve does with Steam, they actually launched a separate website dedicated to a translating project ran by the community. http://translation.steampowered.com

I think it's a sound idea, at least with a userbase of over 40 million.

First post here Just wanted to point you to this talk by Twitter's i18n team on how they do community driven localization. While the talk itself is a bit rails-centric, I think there are a lot of good ideas that can be applied to PA (how they provide context, etc).

OK awesome. I was starting to get worried there for a minute that we'd only have the Civ style "randomize me up a map yo" terrain generator.

And I am sure a lot of us are perfectly comfortable fiddling around in a json file full of triangles and surface normals, but I bet a lot of people would appreciate some sort of WYSIWYG planetary map editor.