Robert Grzeszczuk (rg)
Tue, 13 Jul 1999 09:36:49 -0700 (PDT)
On Jul 13, 4:12pm, C I Knowles wrote:
> Subject: volumeOptimize?
> I'm having a few problems with the volumeOptimize routine......if anyone
> can explain I'll be very interested. I'd also be very interested to hear
> from anyone else who is using Volumizer with Inventor.
>
> I'm using a version of the Inventor examples provided with Volumizer and
> getting some good results. However I've had a few problems when using
> the volumeOptimize method. The simple Inventor example file
> voivAppearance.cxx calls the method in two places - at the end of the
> readVolume function, with the parameter 'PRE_TRANSPOSE' and in the
> doAction method with the parameter 'BEST_PERFORMANCE'.
>
> Chapter 3 of the Volumizer manual states that parameters can be one of;
> BEST_PERFORMANCE, INTERLEAVED or TEXTURE_OBJECTS
>
> However the Enums.h file gives;
> BEST_PERFORMANCE, TEXTURE_OBJECTS, INTERLEAVE_2WAY, TIGHTEN_MEMORY,
> PRE_TRANSPOSE, LEAST_MEMORY and O2
A clear slip in our documentation efforts.
volumeOptimize will take any of the enumerants listed in Enum.h, although some
of them may be no-ops as of release 1.1 (for example, O2, LEAST_MEMORY, and
TIGHTEN_MEMORY) and were only left for backward compatibility.
PRE_TRANSPOSE will create XZ and YZ stacks of 2D textires from the original
XY stack. It is a no-op for 3D volumes. It is not indepotent and can be
called repeatedly to propagate changes.
INTERLEAVE_2WAY is used on certain platforms to improve memory utilisation and
texture download performace. It takes two textures of matching size and
interleaves them by placing the first pixel of the first texture, followed by
the first pixel of the 2nd texture, then the seconds pixels, and so on. This
oprtation is indepotent: a brick pair is marked as interleaved and further
attempts to interleave it will leave it unchanged.
TEXTURE_OBJECTS will create an OpenGL texture object for each brick (or pair of
interleaved bricks). It should only be called once. Multiple calls will cause
a new set of texture objects (will all the memory overhead) to be created. The
data in the brick can be replaced by marking it as dirty (using setDirtyFlag()
method) or explicitly forcing a re-dpwnload with textureLoad() or
textureSubLoad actions.
Patch3696 adds a new option: SHARED_OBJECT which willcreate only a single
OpenGL texture object that is re-used by a ll the bricks. This is critical to
get good download performace on IR type systems
BEST_PERFORMANCE will select a combination of the above flags to get maximum
frame rate on a specific platform.
>
> -I can find no description of what these mean anywhere, nor a good
> description of what function 'volumeOptimize' performs in each case.
>
> However, back to my example. I took the Inventor example and have
> modified the code. My voxel data is in memory already, hence my 'reader'
> copies it into the bricks in the voivAppearance node. This works with no
> problem and displays the results well.
>
> The debug information shows that texture objects are being created:
> -------------------------------------
> Optimizing volume ... Making texture objects ... 1 2 3 4 5 6 7 8 9 10
> 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 done.
> Making texture objects ... 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47
> 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71
> 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95
> 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114
> 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132
> 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150
> 151 152 153 154 155 156 157 158 159 160 done.
> Making texture objects ... 161 162 163 164 165 166 167 168 169 170 171
> 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189
> 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207
> 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225
> 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243
> 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261
> 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279
> 280 281 282 283 284 285 286 287 288 done.
> done optimizing.
> Allocating Transient Geometry Cache:
> tetraCount 5 maxBrickCount 128 maxSamplesNumber 512
> Pp 32 ms Ld 49 ms DR 0.0 Mv/s Dw 149 ms Tl 230 ms fps 4.35
> ------------------------------------
> However, if I then change the voxel data and force the voivAppearance
> node to re-read the data I get:
>
> Optimizing volume ... Making texture objects ... 289 290 291 292 293
> 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311
> 312 313 314 315 316 317 318 319 320 done.
> Making texture objects ... 321 322 323 324 325 326 327 328 329 330 331
> 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349
> 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367
> 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385
> 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403
> 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421
> 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439
> 440 441 442 443 444 445 446 447 448 done.
> Making texture objects ... 449 450 451 452 453 454 455 456 457 458 459
> 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477
> 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495
> 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513
> 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531
> 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549
> 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567
> 568 569 570 571 572 573 574 575 576 done.
> done optimizing.
> Pp 2477 ms Ld 2 ms DR 0.0 Mv/s Dw 1244 ms Tl 3723 ms fps 0.27
>
> And if I repeatedly force a re-read of data after about 3 times I get a
> core dump because I'm out of memory.
As mentioned above, repeated calls to volumeOptimize() that result in new set
of texture objects to be created are likely to casue just that. many texture
manager implementations will reserve space for a copy of the original texel
data when a texture object is created. Thefore, you are best off not making
this call repeatedly (or al least explicitly destroy the previous set of
texobjects whose ids cab be obtained with getTexObjectID() method).
If what you want to do is replace the values of texels in the bricks, you can
simply change the values and amrk the brick as dirty:
for (brickNo = 0; brickNo < BrickCount; brickNo++) {
...
aBrick->getSizes(X,Y,Z); // NOTE: brick sizes may differ
memcpy(aBrick->getDataPtr(),newData,X*Y*Z*aBrick->getBytesPerVoxel());
aBrick->setDirtyFlag(VO_TRUE);
voAppearanceActions::textureBind(aBrick);
...
// draw all transient geometry for updated aBrick here
}
next call to textureBind() action will propagate the changes.
You can also explicitly re-download data for the current brick using
textureLoad() or textureSubload() actions. Make sure to still mark the brick
as dirty in case pre-interpolation lookup table (or some such thing) is being
used.
>
> So it looked to me as though the previous texture objects are not being
> deleted. Hence I added a call to volumeUnOptimize and deleted all the
> bricks before the data is re-read. I still get the same results.
I'll lookinto this. volumeUnoptimize should delete the previous set of texture
objects. Note, however, that re-creating them is a rather expensive process
and that the method suggested above is likely to produce substantially better
performance.
>
> But......if I change the PRE_TRANSPOSE parameter at the end of the
> readVolume function to BEST_PERFORMANCE everything works without any
> problems....and similarly if I omit the volumeOptimize call completely.
>
> Anyone know why??
Omitting the call will simply use un-optimized volume (no interleaving, texture
objects or anything else) and on many platforms will result in sub-optimal
performance. I am not sure why the first change would stop re-creation of
texture objects. it shouldn't.
Let me know if problems persist.
-rg
PS Patch3696 fixes some of the problems with volumeOptimize on selected
platforms, but noe of them seem related to what you're seeing.
>
> Thanks
>
> Chris
>-- End of excerpt from C I Knowles
This archive was generated by hypermail 2.0b2 on Mon Nov 01 1999 - 14:11:13 PST