Archive: 8 posts

---

It had been said before the LBP2 release that the record audio function of the magic mouth was a low bit-rate/quality and limited max duration to help keep the file size of levels low enough for players to download them in a reasonable amount of time. It was estimated that this would all result in being able to record about 10 minutes worth. I have a tutorial level I'm working on where I am using audio recording and I've recorded about 4 minutes worth, spread out over a few different tutorials, and I've already run out of thermometer for the entire level. I have built a few different rooms/environments to put the tutorials in so it is not as much as a regular level but still the entire thermometer is gone which surprised me. Has anyone else run into this issue and if so about how much record time did you do to get to that point? I'm wondering if there is much of a difference between long/complex levels and empty levels in terms of the amount of audio that can be recorded. And even further is if different recording inputs result in better/worse compression allowing more or less overall audio to be recorded. (I'm skeptical about this last part.) Recording audio is obviously costly to the thermometer but I was thinking that one of the 'mini thermometers' would record that data giving me more indication of when the recorded audio thermometer was full. I believe that when a thermometer for indices, or sackbots or music becomes full it can borrow some more of the over-all thermometer to continue. If there was a way of knowing when the audio themo is full then I could try and balance the different elements in the level more effectively. As it stands in this tutorial level there is about 3.5 bars free on the overall thermometer. But if I record about 15 - 20 seconds of more audio then it will fill up the rest of the thermo and overheat the level. Whereas when I first started recording I could record quite a lot of audio without the thermometer moving much at all.

2011-02-07 22:22:00 Author: Trindall Posts: 297

---

Recording audio is obviously costly to the thermometer but I was thinking that one of the 'mini thermometers' would record that data giving me more indication of when the recorded audio thermometer was full. LBP2 doesn't use the maximum of the "mini thermometers" system to determine the overall thermo like LBP1, but rather all the "mini thermometers" are now cumulative into the overall thermo, so, yes, there will be "a difference between long/complex levels and empty levels in terms of the amount of audio that can be recorded."

2011-02-07 23:09:00 Author: Aya042 Posts: 2870

---

LBP2 doesn't use the maximum of the "mini thermometers" system to determine the overall thermo like LBP1, but rather all the "mini thermometers" are now cumulative into the overall thermo, so, yes, there will be "a difference between long/complex levels and empty levels in terms of the amount of audio that can be recorded." Somehow I was thinking that some of the mini-thermometers had a cap on how big they could grow in LBP2 before thermometer resource reallocation started happening. (Maybe it was the clips from MM people showing off the pre-retail builds when the music sequencer was opened up and the thermometer turned blue and they said that that thermometer was not the total thermometer but just for the music sequencer that I got the incorrect assumption from.) It sounds like the mini thermometers are just counters of different resources that can expand or contract until the sum of all thermometers fill the entire level thermometer based on the construction of the level rather than having fixed pool limits. I wonder if the pre-retail-approximation-guess of about 10 minutes of recorded audio comes from an empty level except for a checkpoint and a row of magic mouths or if possibly I'm doing something inefficient when I'm getting about 4 minutes in a sparsely populated tutorial level?

2011-02-07 23:22:00 Author: Trindall Posts: 297

---

I wonder if the pre-retail-approximation-guess of about 10 minutes of recorded audio comes from an empty level except for a checkpoint and a row of magic mouths or if possibly I'm doing something inefficient when I'm getting about 4 minutes in a sparsely populated tutorial level? Well, you could always try creating a copy of the level, then deleting everything except the magic mouths, and see what happens to the thermo. Realistically, you're not gonna be able to get the maximum amount of audio in a single level if you intend for it to have anything else in it, so you may have to look at splitting into sublevels.

2011-02-07 23:28:00 Author: Aya042 Posts: 2870

---

If your tutorials are already compartmentalized into different rooms in the level, maybe splitting the level up using level links would alleviate the issue? Also, have you tried downloading your level to see if takes longer than typical? Because, as you said, this voice thermometer is probably primarily designed to limit download times, so maxing it out might lead to worst-case loading pains for the players. (Much more so than maxing most of the other mini thermos, which are set up to throttle complexity for engine performance, and don't necessarily imply massive file sizes.)

2011-02-07 23:38:00 Author: LittleBigDave Posts: 324

---

Well, you could always try creating a copy of the level, then deleting everything except the magic mouths, and see what happens to the thermo. Realistically, you're not gonna be able to get the maximum amount of audio in a single level if you intend for it to have anything else in it, so you may have to look at splitting into sublevels. I agree with this completely. As I get only a little time each day to create in LBP after the kids go to sleep I thought I would check to see if anyone has already crunched all of the numbers and theorycrafted this to exhaustion before I start doing limit testing. Also since I had only my data I couldn't be sure if this was typical or abnormal performance. But realistically I'm working on how to break up the level into sub-levels without breaking up the user flow/experience too much. I think the multimedia approach will help a wider audience get something out of the tutorial rather than trying to fit five different tutorials into a single level and using only notes and signs. Also, have you tried downloading your level to see if takes longer than typical? Because, as you said, this voice thermometer is probably primarily designed to limit download times, so maxing it out might lead to worst-case loading pains for the players. (Much more so than maxing most of the other mini thermos, which are set up to throttle complexity for engine performance, and don't necessarily imply massive file sizes.) You have a very good point with the file size difference in the thermo calculations of audio versus vertices or textures. I do have the different tutorials compartmentalized and they can be easily broken apart but the change from a single level with a menu selector into a more hub specific flow has some changes in philosophy that I'd need to build into the level design. Not a problem to do at all just time, duct tape and elbow grease. Thanks for the replies.

2011-02-07 23:40:00 Author: Trindall Posts: 297

---

I thought I would check to see if anyone has already crunched all of the numbers and theorycrafted this to exhaustion before I start doing limit testing. If you glean anything from testing, please do post your conclusions. I'm looking to add some voice recording myself, and was also wondering what the ballpark amount I can add is going to be. I'm further wondering if there is a practical limit to how many short sound clips I can have. I don't know a lot about audio compression, but I thought most compression algorithms become less efficient as the payload size gets closer to zero. So even if I could have one four minute clip, maybe I can't have anything close to 24 ten second clips?

2011-02-08 00:07:00 Author: LittleBigDave Posts: 324

---

I don't know a lot about audio compression, but I thought most compression algorithms become less efficient as the payload size gets closer to zero. So even if I could have one four minute clip, maybe I can't have anything close to 24 ten second clips? Typical contemporary audio compression algorithms are based on Fourier analysis, and typically subdivide the sample into small chunks and compress each chunk separately by calculating a superposition of cosine functions required to recreate the audio in the chunk, so the overall sample length to filesize scales fairly linearly. The JPEG picture format uses a similar method, but use the waves to refer to pixel intensities across luma and chroma planes. What tends to affect the filesize most is the relative complexity of the sample, based on how many discrete cosine waves are required to recreate the chunk, so obviously clips with large amounts of silence will compress better. It's possible they didn't bother taking this into account when determining the thermo requirements of a particular sample, but simply based it on the total length of all samples, and ignored the complexity.

2011-02-08 00:44:00 Author: Aya042 Posts: 2870

Posts: 1077139    Threads: 69970    Members: 9661    Archive-Date: 2019-01-19

Aus dem Archiv wurden alle persönlichen Daten wie Name, Anschrift, Email etc. - aber auch sämtliche Inhalte wie z.B. persönliche Nachrichten - entfernt.
Die Nutzung dieser Webseite erfolgt ohne Speicherung personenbezogener Daten. Es werden keinerlei Cookies, Logs, 3rd-Party-Plugins etc. verwendet.