This project is read-only.

Progress Update?

Oct 10, 2007 at 12:34 PM
Hi Joel,

I've read through the XNA Creators Club forums where you discuss some of the fixes that you still have listed here in the Roadmap.

Could you give us an update if there is any?

Oct 13, 2007 at 3:11 PM
apologies for the delay in response, I thought these forums sent me an email when a new post was posted :-) Yes, here is the status. I basically have two things that I want to do for the next release, even though I'd obviously check in and make the code available if I succeeded in one or the other.

First one would be the easiest, I need to get in there and refactor the pipeline API that I've created. The gist of the problem is that I'm pre-loading everything in the Importer, and by the time I get to the writer, the computer could potentially have no more memory. What I'm going to do is refactor it so that the importer merely gathers information about the source of data. The content type that it returns to the processor (and then writer) would then be responsible for reading the data from disk on an as-needed basis. At that point, the gen 1 GC should clean up any memory that was allocated for the processing of that frame.

That alone should solve most or all of the pipeline issues. The second thing I want to get working is the storage of the frame data using DXT1. I was successful in making it write (which resulting in a far smaller disk footprint), however, I couldn't figure out (yet) how to read the dxt1 data back into a texture. for a bit of background, I store the TextureContent in binary form in one big .xnb file ... and at runtime, I merely step through it, reading the frame info as needed in a file stream from disk.

If I can figure out that second piece, then the component should be a lot nicer to the hard drive as it would take advantage of dxt1 compression :-)
Oct 15, 2007 at 7:02 AM
Thanks for the update Joel.

How are you planning on getting audio sync'd with the video?
Oct 15, 2007 at 2:39 PM
The first milestone will be a very simple sync mechanism that will require you to extract the audio separately and pass in a reference to the sound bank. The Video class will then start the cue and perhaps pause it at the same time. I'd also include a utility to extract the audio for this purpose. Simplistic I know, but it would get the job done for now