diff options
| -rw-r--r-- | converter/other/jpeg2000/libjasper/jpc/jpc_enc.c | 14 |
1 files changed, 3 insertions, 11 deletions
diff --git a/converter/other/jpeg2000/libjasper/jpc/jpc_enc.c b/converter/other/jpeg2000/libjasper/jpc/jpc_enc.c index debdeec0..1a32de96 100644 --- a/converter/other/jpeg2000/libjasper/jpc/jpc_enc.c +++ b/converter/other/jpeg2000/libjasper/jpc/jpc_enc.c @@ -1561,17 +1561,8 @@ quantizeBand(jpc_enc_band_t * const bandP, } bandP->stepsize = jpc_abstorelstepsize( bandP->absstepsize, prec + bandP->analgain); - /* I couldn't figure out what the calculation with tccp_numgbits and - stepsize does (or even what a step size is), but there is an - assertion elsewhere than the number here is at least at large as - the 'numbps' value for every code block, which means - 'actualnumbps'. In practice, we saw that not be true, so we added - the code to make 'actualnumbps' the floor here in hopes that would - fix the problem. 22.11.06 - */ - bandP->numbps = - MAX(actualnumbps, - tccp_numgbits + JPC_QCX_GETEXPN(bandP->stepsize) - 1); + + bandP->numbps = tccp_numgbits + JPC_QCX_GETEXPN(bandP->stepsize) - 1; if (!tileP->intmode && bandP->data) quantize(bandP->data, bandP->absstepsize); @@ -1688,6 +1679,7 @@ encodemainbody(jpc_enc_t *enc) { int numgbits; + pm_message("comp %u level %u band %u", (unsigned)cmptno, (unsigned)rlvlno, (unsigned)bandno); quantizeBand(band, tile, cp, cp->ccps[cmptno].prec, cp->tccp.numgbits, |