Problem with writing a matrix to jxf File and reading it back to Jitter
I still did not get an answer to my question so i try to edit my original post for better understanding and reproducing my Problem to hopefully some hints to my issue:
I am writing an jitter external for Max 9.0.8 with max-sdk-8.2.0:
Inside the external i am filling a 2x2x3 matrix and storing the matrix to brightnessTest.jxf in the Filesystem. After reading back into a matrix object i want to get the stored Values back in a jitter Patch.
While storing the values to the matrix with this code:
for (x->currentframe = 0; x->currentframe < 3; x->currentframe++) {
for (i = 0; i < x->nrOfCells; i++) {
yCellPos = i / x->cols;
xCellPos = i % x->cols;
brightness = 25. * i;
post("i: %ld", i);
post("x->currentframe: %ld, xPos: %ld, yPos: %ld", x->currentframe, xCellPos, yCellPos);
post("brightness: %f", brightness);
brightnessp = brightness_bp
+ xCellPos * sizeof(float)
+ yCellPos * sizeof(float) * (&brightness_minfo)->dim[0]
+ x->currentframe * sizeof(float) * (&brightness_minfo)->dim[0] * (&brightness_minfo)->dim[1];
*brightnessp = brightness;
}
}The output to console looks fine (frames 0,1,2 are filled with values):
But when i try to read back the matrix values in the patch
it seems that only the first frame (frames 0) was read back correctly. At the second frame (frames 1) i only get 0. Values and the third frame (frames 2) i get no Values.
Hope anyone could give me a hint to solve my problem.
I have edited my original post from Aug 11 for better understanding my Problem. Hope that anyone can help me.
I still have problems with filling a jitter matrix inside an Max c-external with float32 values.
For testing reasons i have tried to use different matrix dimensions:
dimcount = 3, dim[0] = 2, dim[1] = 2, dim[2] = 3, planecount = 1, type = float32
dimcount = 3, dim[0] = 3, dim[1] = 2, dim[2] = 3, planecount = 1, type = float32
dimcount = 3, dim[0] = 4, dim[1] = 2, dim[2] = 3, planecount = 1, type = float32
I am initializing my Matrix with this code:
void initMatrix(t_testmatrixset* x) {
t_jit_matrix_info brightnessinfo;
if (x->brightnessmatrix == NULL) {
jit_matrix_info_default(&brightnessinfo);
brightnessinfo.dimcount = 3;
brightnessinfo.planecount = 1;
brightnessinfo.dim[0] = x->cols; //2, 3, 4
brightnessinfo.dim[1] = x->rows; //2
brightnessinfo.dim[2] = x->frames; //3
brightnessinfo.type = _jit_sym_float32;
x->brightnessmatrix = jit_object_new(_jit_sym_jit_matrix, &brightnessinfo);
x->nrOfCells = x->rows * x->cols;
}
}After initializing i am setting a ascending distinct float value to the jitter matrix while iterating from cell to cell:
t_jit_err testmatrixset_fillmatrix(t_testmatrixset* x) {
t_jit_matrix_info brightness_minfo;
float* brightness_bp, * brightnessp, brightness;
long i, yCellPos, xCellPos, index;
long lock;
float* fdata;
initMatrix(x);
x->currentframe = 0;
index = 0;
jit_object_method(x->brightnessmatrix, _jit_sym_getinfo, &brightness_minfo);
lock = jit_object_method(x->brightnessmatrix, _jit_sym_lock, 1);
jit_object_method(x->brightnessmatrix, _jit_sym_getdata, &brightness_bp);
for (x->currentframe = 0; x->currentframe < x->frames; x->currentframe++) {
for (yCellPos = 0; yCellPos < x->rows; yCellPos++) {
for (xCellPos = 0; xCellPos < x->cols; xCellPos++) {
brightness = (float) index;
post("xPos: %ld, yPos: %ld, x->currentframe: %ld, brightness: %f" ,xCellPos, yCellPos, x->currentframe,brightness);
long xOffset = xCellPos;
long yOffset = yCellPos * (&brightness_minfo)->dim[0];
long frameOffset = x->currentframe * (&brightness_minfo)->dim[0] * (&brightness_minfo)->dim[1];
long fullOffset = (xOffset + yOffset + frameOffset);
brightnessp = brightness_bp + fullOffset;
post("xOffset: %ld, yOffset: %ld, frameOffset: %ld, fullOffset: %ld",xOffset, yOffset,frameOffset, fullOffset);
post("dimstride[0]: %ld", (&brightness_minfo)->dimstride[0]);
post("dimstride[1]: %ld", (&brightness_minfo)->dimstride[1]);
post("dimstride[2]: %ld", (&brightness_minfo)->dimstride[2]);
post("dim[0]: %ld", (&brightness_minfo)->dim[0]);
post("dim[1]: %ld", (&brightness_minfo)->dim[1]);
post("dim[2]: %ld", (&brightness_minfo)->dim[2]);
post("planecount: %ld", (&brightness_minfo)->planecount);
*brightnessp = brightness;
index+=1;
}
}
}
jit_object_method(x->brightnessmatrix, _jit_sym_lock, lock);
post("matrix filled!");
return JIT_ERR_NONE;
}For better understanding i have added some post functions with debug informations while setting each value in the matrix. For example these are some debug Informations from a test with an 3x2x3 Matrix:
When i print the matrix i get these values:
jit.print: 0.000,1.000,2.000
jit.print: 4.000,5.000,6.000
jit.print: <DIM 2>
jit.print: 8.000,9.000,10.000
jit.print: 12.000,13.000,14.000
jit.print: <DIM 2>
jit.print: 16.000,17.000,0.000
jit.print: 0.000,0.000,0.000
So as you can See in the matrix Values are missing some Values: 3.0, 7.0, 11. and 15
When i try the same thing with an 4x2x3 Matrix I get following debug Informations:
When i print the 4x2x3 matrix all expected Values are filled in correctly:
jit.print: 0.000,1.000,2.000,3.000
jit.print: 4.000,5.000,6.000,7.000
jit.print: <DIM 2>
jit.print: 8.000,9.000,10.000,11.000
jit.print: 12.000,13.000,14.000,15.000
jit.print: <DIM 2>
jit.print: 16.000,17.000,18.000,19.000
jit.print: 20.000,21.000,22.000,23.000
so anybody has an idea what could be the problem in the first 3x2x3 Matrix. I figured out that the dimstride values in all matrices are:
dimstride[0]: 4
dimstride[1]: 16
dimstride[2]: 32
i guess that could be the problem for setting float32 values correctly in my 3x2x3 matrix and i have some troubles with my 2x2x3 float Matrix too. So i hope someone can help me with this.
you need to use the info dimstrides and a base char pointer to properly calculate offsets. it should look something more like:
char* bp;
jit_object_method(x->brightnessmatrix, _jit_sym_getdata, &bp);
...
brightnessp = (float*)(bp
+ xCellPos * info.dimstride[0]
+ yCellPos * info.dimstride[1]
+ x->currentframe * info.dimstride[2]);
*brightnessp = brightness;If this doesn't get you sorted, zip up your source file and I'll try and build them with the SDK.
alright, Thanks for your reply!
i have tried to change my code with your suggestion but it's still not working. So i have prepared 3 Patches so that you can reproduce my problems. In Each Patch you can fill the matrix within the external by bang #1 and fill another matrix within the Patch through setcell messages through bang #2:
It was not possible to attach the zip file directly in the Forum, so you can download it from DropBox: testexternalZip
Let me know if anything is unclear. Thanks!
Hello Rob!
Have you already had the opportunity to view my patches and c external?
Best Regards C.
ok. i have had a mistake in my zip File. I changed float* brightness_bp to char* brightness_bp as you suggested in your post. Here is the updated source: cexternalDropbox
but also when i use char pointer instead float i do not get the expected values in my matrix.
The matrix values wich are set in the external are:
but i would expect:
pretty sure this line is wrong due to C casting precedence rules:
brightnessp = (float*) brightness_bp + fullOffset;and should be
brightnessp = (float*)(brightness_bp + fullOffset);you want to apply the byte offset to the char pointer, then cast to a float pointer
Alright! Thank you very much! Now it's working fine.
Best Regards
C.