I'm going to ignore the nasty defects and say,,,,: 

Look at those very cool and very large crystal grain boundaries!   Each one of those polygon shapes are individual crystals.

I suspect some pretty rapid cooling,,,  did they have a test fixture for those fillets?  Something with a powerful heat-sink?

Ok ok,,,,  Nobody asked but I think I can tell you why those defects are they way they are...  From long experience :)

Tungsten angle on those samples were very exaggerated to a push angle.

What happens with titanium and this kind of angle is the following:   The end of the filler metal wire gets a small VERY STICKY ball at the end,,, (any Ti welder will attest to this)....   The large sticky ball then touches the base metal in front of the weld pool and and sticks,, and I mean really stricks, like liquid glue.. Then the operator panics....  When the operator panics, they turn up the heat and shove the torch closer and weave it to wash it into the horizontal piece of base metal  (this is why the fillet is unbalanced and there are voids in unusual locations).

Having a little extra long stickout and a special gas shield will help to give a good torch angle that is 45 to the axis of the weld with about a 5-10 degree "push" angle max...  When this configuration is in place, the filler metal will dip into and out of the leading edge of the puddle without the "stickiness" problems that caused the defects we see above.

Thus endeth the lesson.