Usage about Wavemesh

[Seven377]

Usage about Wavemesh

[valette]

As explained in the Readme file:

execute wavemesh without arguments to see the available options.

Also, feel free to submit pull requests if you want to add more documentation!

Thanks

[Seven377]

I have successfully run Wavemesh, I have two questions:

1. when I set the coordinates quantization to 16 bits, there have a error :
   " terminate called after throwing an instance of 'std::length_error'
   what(): vector::_M_fill_insert "
2. I don't know how to compute compression rates (bpv)，including connectivity compression rates and geometry compression rates.

[valette]

when I set the coordinates quantization to 16 bits, there have a error :
" terminate called after throwing an instance of 'std::length_error'
what(): vector::_M_fill_insert "

I think you cannot use 16 bits, 15 bits should work however

I don't know how to compute compression rates (bpv)，including connectivity compression rates and geometry compression rates

during decompression, a file "report.txt" should give you all these infos

[Seven377]

I find it , thank you very much.

[Seven377]

I have understood the meaning of each value in the file "report.txt" , however, I want to know that if the "total data" contains the compression rates of connectivity and geometry ? Can I obtain the compression rates of geometry at each detail of level ?

[valette]

OK, the data is not completely clear:

Let's see:
Level 25: 11547f, 5826v, valence entropy= 2.2838, total data: 4.51972 bits/v (connectivity: 25757bits, 3.73387 bits/vertex for this level)

at level 25, the total data weighs 4.51972 b/v i.e. 4.51972 *5826 = 26331 bits. As written, connectivity took 25757 bits, so geometry took the remaining part.

[Seven377]

I have tried to compute geometry compression rates in this way, however, the size of total data sometimes is less than connectivity. for example, when I used Bunny model ,
Level 16: 5618f, 2811v, valence entropy= 2.31131, total data: 3.19518 bits/v (connectivity: 13092bits, 3.92571 bits/vertex for this level)
the total data weighs 3.19518 bits/v i.e. 3.19518*2811=8982 bits, but connectivity took 13092bits.
In fact, this problem exists at level 1 to level 16.

[valette]

oops, sorry, I made a mistake in my explanation on bunny:
Filename: out.ddd
Quantization : 12 bits
No lifting
No Wavelet Geometrical Criterion
Total execution time : 0.75816 seconds : 91604.7 faces/s
Level 0 : 26f, 19v, total data: 1456 bits (connectivity: 422bits)
Level 1: 27f, 20v, valence entropy= 2.22821, total data: 0.050066 bits/v (connectivity: 606bits, 184 bits/vertex for this level)
[...cut...]
Level 25: 25587f, 12874v, valence entropy= 2.0304, total data: 8.37412 bits/v (connectivity: 48430bits, 3.29857 bits/vertex for this level)
Level 26: 69451f, 34834v, valence entropy= 1.19835, total data: 16.2191 bits/v (connectivity: 96366bits, 2.18288 bits/vertex for this level)
Global coding: 16.2191 bits/vertex, connectivity : 2.76644 bits/vertex, geometry : 13.4527bits/vertex
File size: 70622bytes

for level 25:
the total data weighs 8.37412 bits/v i.e. 8.37412 * 34834=291704 bits. for this model, you should always multiply the bitrate (in bits/v) by the original number of vertices : 34834 in the bunny case.

Sorry for the mixup is that better now?

[Seven377]

I have two questions:

1. For Bunny model, at each level, compression rates of geometry=(total data* 34834- connectivity)/34834 , right ? or I should divide the current number of vertices at this level (12874, take level 25 for example )?
2. Should I use the original number of vertices for all model ? Because this problem also exists when I used Sphere model.

[valette]

1 - correct.
2 - Yes, you should always use the original number of vertices. This is a common use for progressive compression (all papers report this in the same way)

[Seven377]

I have learned is far from enough for mesh compression, I will continue in my efforts.Thank you for your patient explanation.

[valette]

you're welcome!