Pages: [1]
|
|
|
Author
|
Topic: Ways to improve the game.. (Read 3015 times)
|
fluke
Zebranky food
Offline
Posts: 10
|
I've been watching this project for a while, and its good to see theres finally a solid, playable version of this classic.
So far, the UQMasters team have done an amazing job. Not only has SC2 been revived, its actually better then ever! The new music and resource gathering interface are absolutly excellent.
Still, there are two things that I feel severly need to be updated. Yet these are both massive undertakings.
First off, the game could use a facelift. Now, I dont want redesigns or 'artists' interpretations, but very litteral translations of the sprites into high rez high color versions really would not hurt.
Second, the voices. UGH. I played the PC version as a kid, never seeing or hearing the 3dO version.. and I have got to say,. much of the voice acting is just terrible. It simply does not do the brilliant dialog justice! Now I know there are thousands of lines, but there has to be some way to undertake this. I even happily donate my own voice for a character or two as I am sure I could do a much better job then the 3dO team did.
|
|
|
Logged
|
All hail Toys for Bob! Long live the Spathi!
|
|
|
|
Neonlare
*Smell* controller
Offline
Gender:
Posts: 277
Nut Case for Star Control
|
I've been watching this project for a while, and its good to see theres finally a solid, playable version of this classic.
So far, the UQMasters team have done an amazing job. Not only has SC2 been revived, its actually better then ever! The new music and resource gathering interface are absolutly excellent.
Still, there are two things that I feel severly need to be updated. Yet these are both massive undertakings.
First off, the game could use a facelift. Now, I dont want redesigns or 'artists' interpretations, but very litteral translations of the sprites into high rez high color versions really would not hurt.
Second, the voices. UGH. I played the PC version as a kid, never seeing or hearing the 3dO version.. and I have got to say,. much of the voice acting is just terrible. It simply does not do the brilliant dialog justice! Now I know there are thousands of lines, but there has to be some way to undertake this. I even happily donate my own voice for a character or two as I am sure I could do a much better job then the 3dO team did.
A Question, to everyone...
I didn't notice anything bad with much of the voice acting, yet the Utwig and Chmmr did annoy me, what voices make people slate this so bad?
And, I think the animation for the com-links should be more "interesting" and smooth, so it looks like a proffesional animation!
|
|
|
Logged
|
"would newton's law theory actually work if a Chmmr Avatar did a backwards pelvic thrust towards a planet and would this constitute an X=Y-0 in the part it ran straight into a Supox Blade and lasted long enough to survive?" - Elerium (as Valaggar)
|
|
|
Svip
Zebranky food
Offline
Posts: 10
|
Well, I'd love to rerecord a separated species. If at all possible, some may start a subproject for this purpose.
As Novus said, it may be a bit large to undertake.
I have considered rerecording the Orz speech myself. I was a bit disappointed, because I thought they would speak faster and have longer pauses between sentence - which would acompany the music in the background - so it would make them sound more wacky.
It is not that hard, just need a mic, read the text, give yourself a bunch of hours/days. Then make the ogg files, and edit the files that control when the new line is displayed (in milliseconds, a second being 1000) and them zip them into the uqm file.
Sounds not that hard, only thing that make take time would be all the recording, given you might record something wrong.
|
|
|
Logged
|
|
|
|
|
Novus
Enlightened
Offline
Gender:
Posts: 1938
Fot or not?
|
What we need is a friendly programmer to automate the voice recording process!
If we could get a recording interface that works straight of the game database, prompting lines and recording OGG's, modifying the database however nessisary and archiving to UQMs, this would make it simple enough for anybody to take a stab at it. With this, we could easily pool the entire fan populous for the best vocal interpritations, and end up with a kickass product.
The way I see it, getting the text and putting the voices in UQM is not the problem; after all, the text is in easily-accessible plain text files (load into your favourite text editor; problem solved) and to put the files in UQM all you have to do is save the files in Ogg Vorbis format to the right directory (a plain save or export operation in most current sound editors).
The hard and time-consuming parts are, as I see it (my estimates of time use):
- Reading the text and planning how to say every line (emotion, emphasis, et.c.); maybe a minute per row.
- Planning and testing effects to use; an hour or two per race, maybe more.
- Rehearsing the lines so they sound as intended (no unnecessary hesitation, mistakes, sounds convincing et.c.); another minute per row, at least, I guess.
- Recording (about a minute per row for a few takes).
- Post-processing (trimming, noise removal, volume normalisation, applying effects); another minute per row at the least.
For a race with 74 to 880 (Slylandro Probes to Starbase (post Ilwrath incident)) lines (median 344; Spathi Home)*, this amounts to something like 1.5 hours+4 min/row = 6 to 60 hours (median 24) per race, or 1 to 10 (median 6) workdays assuming 6 hours effective work per day. Total for 27 races and 9944 rows is 703 hours or roughly 5 man-months (as an aside, how well does this match the workload for the 3DO voices?). And that's the non-automatable artistic work.
Saving with the right filename, on the other hand, takes all of ten seconds per line, if your audio editor supports Vorbis. Similarly, opening a text editor on the side with the text is equally easy (and you only have to do it once per session). Writing a program to automate these operations seems kind of useless, especially since you'd have to integrate it with whatever sound editor people want to use to gain anything at all. Downloading and installing the program would probably eat your gains up immediately.
So how do we get a programmer hyped on writing this out? Well, the first step would be to convince him that your idea would help; preferably a lot.
* Rows of text per race counted using bash script:
for a in content/comm/*/*.txt; do (echo -n $a; (grep -v "^ *$" $a | grep -v "^#") | wc) ; done | sort -n -k2
Output (file, rows, words, characters):
content/comm/slyland/slyland.txt 74 496 3161 content/comm/rebel/rebel.txt 151 1649 9459 content/comm/shofixt/shofixt.txt 196 1539 8646 content/comm/mycon/mycon.txt 199 1619 9188 content/comm/chmmr/chmmr.txt 209 2154 12340 content/comm/supox/supox.txt 212 2179 12383 content/comm/yehat/yehat.txt 214 2273 12286 content/comm/comandr/comandr.txt 224 2319 13123 content/comm/urquan/urquan.txt 242 2308 13215 content/comm/blackur/blackur.txt 243 2037 11660 content/comm/umgah/umgah.txt 247 2091 12222 content/comm/vux/vux.txt 317 3454 19543 content/comm/arilou/arilou.txt 318 3527 20011 content/comm/spahome/spahome.txt 344 3429 19396 content/comm/druuge/druuge.txt 352 3525 20103 content/comm/slyhome/slyhome.txt 385 3574 20805 content/comm/orz/orz.txt 388 3637 19669 content/comm/talkpet/talkpet.txt 402 4243 23579 content/comm/spathi/spathi.txt 409 4552 25649 content/comm/ilwrath/ilwrath.txt 415 4395 24899 content/comm/syreen/syreen.txt 417 4337 24284 content/comm/utwig/utwig.txt 485 5484 31361 content/comm/zoqfot/zoqfot.txt 545 3458 19647 content/comm/thradd/thradd.txt 606 5855 33655 content/comm/pkunk/pkunk.txt 643 6891 38030 content/comm/melnorm/melnorm.txt 827 7831 46378 content/comm/starbas/starbas.txt 880 9132 53156
|
|
|
Logged
|
|
|
|
Svip
Zebranky food
Offline
Posts: 10
|
I do not think it is possible to actually write a program to make it "faster".
On another note, some of the voices in the game are good, some are not so good. I don't think our first concern is to change all of them. Just locate those we agree are a bit rough and old or just plain bad and needs improvement, and start with those first.
Besides, there should be one person per race, I do not think that a single person should do more than one race, or the other way around.
|
|
|
Logged
|
|
|
|
kilojava
Guest
|
The Precursors take plenty of time because they're project is a massive undertaking too. Someone just needs to make a website where everyone can send the races that they have done. And eventually all will be taken.
|
|
|
Logged
|
|
|
|
Culture20
Enlightened
Offline
Posts: 917
Thraddash Flower Child
|
There are several ways a program could help automate this process. A simple text interface could ask you to read a line, and your voice would be repeated back to you until you give it an "okay". This is stored in a statically-named file where you could manipulate it with such programs as audacity. Once completely done, run the UQM-über-voice-acting-helper-program (UÜVAHM©®) to have all of the voice files scanned for length, then store the time info in the appropriate text files.
As an aside: automation is rarely for time gains (even in manufacturing; improvements on automation are for time gains); as a sysadmin, I often take more time writing a script that will automate something for me (and others) than the whole saved time. However, it often achieves what is otherwise difficult to find (especially when working with others): consistancy/accuracy. Less things to fix is always a good thing.
|
|
« Last Edit: April 28, 2006, 03:02:41 am by Culture20 »
|
Logged
|
|
|
|
Svip
Zebranky food
Offline
Posts: 10
|
Remember, that it is not each line that is stored in ogg files, it is entire paragraphs. This means that these lines have to connect, and not sound like they do not. It is hardly unlikely that any program could automate this progress well enough for it to be useful.
I do not think you can save any time - anywhere. It is likely that the result will be much better without any automated progress. And perhaps you won't get too annoyed because the program at points does not do as you do.
I cannot see the problem with having a recording application open and a terminal open which have the .txt files of the speech open. You read - then you record. When you are done, you check the sound you have recorded and see how long each line takes, because the entire sound file covers a whole paragraph. And all in all, you are basically done.
I can only say this honestly; you must be very lazy if you want an automated progress. Perhaps too lazy to even get a considerable result out of this. Mmmm?
|
|
|
Logged
|
|
|
|
Novus
Enlightened
Offline
Gender:
Posts: 1938
Fot or not?
|
There are several ways a program could help automate this process. A simple text interface could ask you to read a line, and your voice would be repeated back to you until you give it an "okay". This is stored in a statically-named file where you could manipulate it with such programs as audacity. Once completely done, run the UQM-über-voice-acting-helper-program (UÜVAHM©®) to have all of the voice files scanned for length, then store the time info in the appropriate text files. Before we get any more confused, I'd better clarify the terminology I've been using:
- "row" = newline-terminated string of text (a line in the text file); the amount of text shown on screen at once
- "line" = a single Vorbis file of speech (a line in the theatrical sense)
Recording has to be done a line at a time, not a row at a time (as Svip points out in different terminology), so breaks between rows have to be marked manually. Culture20 seems to suggest saving each row separately and then using a custom program to mark timestamps for each row of text before combining the files. In practice, this would probably involve cutting the recorded sample into a number of pieces manually in the sound editor. With a little practise, it should be easier and quicker to use the existing timestamping program to listen to the speech files and timestamp them. Furthermore, this offloads work from the voice artists onto the general contributing fan population, just like the 3DO timestamping, as the timestamping can be done separately.
As an aside: automation is rarely for time gains (even in manufacturing; improvements on automation are for time gains); as a sysadmin, I often take more time writing a script that will automate something for me (and others) than the whole saved time. However, it often achieves what is otherwise difficult to find (especially when working with others): consistancy/accuracy. Less things to fix is always a good thing. Putting on my software quality assurance hat, I'd like to point out that automation is not a guarantee for consistency and accuracy, either; a badly-designed automated system can cause a lot of trouble in unexpected special cases or if it malfunctions. The illusion of correctness bad automation creates also encourages lax quality checking. As a rule of thumb, I'd say that less manual work corresponds to less chance of something going wrong due to human error. Building complicated automated systems for simple one-off problems is therefore a good way to waste time and decrease quality.
Besides, in this case, consistency is unlikely to be a problem as you will have hordes of fans like us checking every single line you record twice (and analysing every single nuance of your pronunciation of "Ultimate EVIL!!!"). If you reverse two lines, people are bound to notice.
I agree with kilojava that this project needs a web page or suchlike for co-ordination (a wiki page in the Ultronomicon, perhaps?). This would also make it easier to compare different voice actors and different voice processing for the same races.
|
|
|
Logged
|
|
|
|
Pages: [1]
|
|
|
|
|