Got some testing time tonight for this diode.
Using the same lens 405-g-2 the red diode has 2.55 times the divergence of the 445nm diodes we've come to love. The 'beam' does look like a flashlight coming from the lens, in fact I have flashlights that are more collimatedcorrection tests to follow..
update: Here are the results using a few lenses I had in stock:
405-g-2 (broadband coat, ~4mmfl): 400mA 239mW; 70mm wide beam at 26'
aixiz red lens (red coat, ~8mm fl): 400mA 176mW; 35mm beam at 26'
340mW @ 700mA (should be about 490mW)
laser66 'red' lens (red coat, ~8.x mm fl): 400mA 173mW; 34mm beam at 26'
244mW at 525mA
The losses, and the better divergence with the second two lenses are because of the longer focal length which also clips the fast axis.
The combination of one of the longer focal length lenses and the cylindrical correction optics in the group buy/kit will give a good beam, but clipping from the collimating lens will mean 30-35% losses will be necessary.
Given these results, it appears that PBS combining two 150mW 642's will cost just a little more, have much better divergence, and a little less visible brightness.
For beams, the combination of a 6mmFl lens and the cyl lens set from the group buy and in the module kits will extract most of the optical power from the diode and provide about 1.2mrad divergence, which isn't too bad.
In conclusion: with the right lenses these make good beamshow diodes however the 150mW 642nm diodes are the right choice for graphics and precision.