The file format the FPP reads at this time is the conductor format. This is a format that LSP creates that was intended to run on DLA's conductor. This format is very limiting because the channel count is fixed at 16384 channels. We would have to load 4 of these files to output 128K channels. Also the file was created to output data to 4 channels of RS-485 (interleaved data). So we are proposing a new file format. All other formats (LOR, Vixen, Conductor (merging of multiple files), etc would be converted to this file format.
Please look over file format and give feed back. I am open to using an existing format but it needs to be straight binary.
Why does it need to be straight binary? Some people will prefer some compression unless you provide a converter. I know Sean (of Nutcracker fame) has a lot of stats about filesizes, load time, etc.
He seems to like the Vixen 2.x format with base-64 encoding of the data. It's pretty small, yet loads quickly. I think there are a couple smaller formats, but take a long time to load.
I think that your proposed format will be too large of files. With your example size of 16384 channels (which is a lot) will produce 480k of data every second. A 3 minute song would be about 86.4Mb. If somebody had more than 10 songs, by the time they have the audio and sequence files, it would be more than a gigabyte to backup/transfer/etc.
dpitts wrote:The file format the FPP reads at this time is the conductor format. This is a format that LSP creates that was intended to run on DLA's conductor. This format is very limiting because the channel count is fixed at 16384 channels. We would have to load 4 of these files to output 128K channels. Also the file was created to output data to 4 channels of RS-485 (interleaved data). So we are proposing a new file format. All other formats (LOR, Vixen, Conductor (merging of multiple files), etc would be converted to this file format.Please look over file format and give feed back. I am open to using an existing format but it needs to be straight binary.
Have you given any consideration to allowing data sizes larger than one byte? Apart from allowing smoother ramps as hardware supports it, it might make it easier to embed audio channels in one of the file 'channels'. Also, I think that having a file format identifier (including version identifier) at the start of the file is always useful for when you want to update/change the file format.
Well the idea was to have the Pi just play the file and not try to do to much processing of the file. For the playing i am just doing a sequential reading of the file and outputting. This is how i am able to output 128 universes of data. So do you suggest some sort of converter that runs on the PI to convert file after it is transferred?
The format identifier is a good addition. As far as greater data sizes go I am open to that. Would it be good to keep it a multiple of 8 bits. So if you sent 12-Bit data it would need two bytes for each.
