I have no love lost for XML but you can use it and not end up with a bloated file. HLS uses it for the show format but not all the way down to the byte level. That way his files are totally manageable.
Here is an example from part of my MONSTER MASH.hlsdata test file. As you can see for XML it quite compact. HLS does not use variable timing as LOR does but looking at the XML I could see ways to add it and not add much data.
C:\HLS\MONSTER MASH.PCM
3845
50
0
0
1
1
USB_DMX
0
0
0
0
Entec_Open
AH00NYGG
16
Pumpkin 1 - Outline, Normal
1
0-00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,
50-00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,
100-00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,
150-00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,02,19,33,4C,66,7F,99,B2,CC,E5,FE,FE,FE,FE,FE,FE,FE,
.
.
.
.
3800-09,07,04,02,02,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,
smeighan wrote:Dont use IE.Firefox, Chrome, Opera or Safari work. Kurt built the gallery with ajaz and it get a "please wait" when using IE.
Actually I'm using Safari, but I think the issue was with www in the hostname. I realize IE is messy sometimes, but in general ajax should work fine in IE too.
timon wrote:I have no love lost for XML but you can use it and not end up with a bloated file. HLS uses it for the show format but not all the way down to the byte level. That way his files are totally manageable.Here is an example from part of my MONSTER MASH.hlsdata test file. As you can see for XML it quite compact. HLS does not use variable timing as LOR does but looking at the XML I could see ways to add it and not add much data.
C:\HLS\MONSTER MASH.PCM
3845
50
0
0
11
USB_DMX
0
0
0
0
Entec_Open
AH00NYGG
16Pumpkin 1 - Outline, Normal
10-00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,
50-00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,
100-00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,
150-00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,02,19,33,4C,66,7F,99,B2,CC,E5,FE,FE,FE,FE,FE,FE,FE,
.
.
.
.
3800-09,07,04,02,02,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,
I agree this would be smaller. The secret is to NOT have xml tags on every byte.
vixen vix files are xml but when they come to the actual data for your song it switches to base64. 4 ascii bytes to store 3 binary bytes. that is probably the most efficient. In the form you list it takes 3 ascii bytes for each binary byte "FF," for example.
vixen 2.1 is the current champ at reading in the data. my 12,100 channels of data x 240 seconds with 100ms frame timing = 29 million cells of data. This vix file reads in 3 seconds ready to press play. Since i wasnt actually doing any editing in vixen, i cant speak to how easy it it to create effects in it.
HLS pixel data is efficent
21345 123453 432161
34134 43234 134123
Each word is a 24bit rgb value, each value represents the next frame,
vixen vir, vix, hlsnc, nutcracker nc and the hls xml file all assume fixed frame timing. this has worked just fine for my show. if you add variable timing the file size will grow again.
If you have a few changes in a song the LOR xml file format is more efficient. It does things like for the next 400 values, make them ass FF
If you are doing the butterfly effect where every single pixel is different than its neighbor, the LOR xml is the least efficient.
As we get to 15-25K channels the file formats are going to be noticed for massiv rgb channel counts.
That's good and also seem to me very interesting. I will defiantly try it.
smeighan wrote:4 ascii bytes to store 3 binary bytes. that is probably the most efficient. In the form you list it takes 3 ascii bytes for each binary byte "FF," for example.
In terms of storage space, any type of data compression would likely be more efficient than bare Base64 encoding because there tends to be a lot of empty or similar channel values within a sequence.
Compression can be used in conjunction with XML - just apply Zlib (or other favorite compression algorithm), then Base64 encode the results (like what Vixen 2.x currently does with raw channel values). The compressed data could be one entire block, but it's probably safer or better playback performance to just compress one frame at a time and then store those separately so you only need to uncompress one at a time.
don

