blenderblog

Ton has posted some details about how SIGGRAPH turned out for the Blender people. Seems like the SIGGRAPH organizers are finally starting to take notice of Blender.

The fact we applied for three BOF sessions (birds of a feather) attracted two members of the siggraph committee to visit the first BOF and make a statement about plans to include open source and Blender in the official program of next year; especially for courses and the educators program.

And Blender performed well in the The One Million Vertex Challenge.

Having access on Siggraph to fast machines and the latest tools, we did a quick test how the "professional" tools behave on larger datasets. It was also asked to be demonstrated on the booths, but we took time to verify it ourselves too.

The challenge we did is simple; add a cube mesh, and subdivide it in single steps until it reaches 1 million vertices. Then save the file and load it back.
Here's the results on XEON 3.4 GHz with 2 GB mem:

Lightwave:
Doing the subidivides became too slow at 100k vertices, but duplicating the mesh then still worked OK.
Save the file took 4 seconds, loading it back 20 seconds.
Softimage:
Doing the subdivides went reasonable OK, but the last step to make more than 1 Million vertices always crashed.
Saving a file with 500k vertices went reasonable fast, about 1 second, reading it back was 10 seconds
Maya:
Asking it on the booth itself crashed Maya. When trying ourselves we didn't manage to create a huge mesh (it uses modifier stacks or subsurf instead), but a file of 900k we managed to create already was quite slow to read back.
Blender:
Subdividing the last steps took some time, but not slower than the others. Save file was 0.5 seconds, loading it back was instanteously. :)

Nice one! Good to see Blender prove itself and gain a little respect against the commercial products.