Tuesday, June 9, 2009

Hunter testing methodology

I've been playing some of my hunters again recently with a view to updating my Hunter Pet Reviews for the current state of play.

My highest hunter is at 56 or 57 now and will probably comfortably level into Outlands during this process. Most of the other hunter alts are still languishing back in the mid 20's or early 30's so are much the same as they were when the original tests were done last year.

A hunter and pet in the 20's is a vastly different experience to a hunter in the higher levels. The hunter's own talents are only just developing and the pets are barely started. I have kept a very similar talent set for each hunter and pet for these tests. Hunters spend the first 15 or so points in Marksman to get the Go For The Throat talent, then switch to Beastmaster for their remaining talents. Pet talents are first a movement option: Dash/dive or charge, then points in cobra reflexes. With so few talent points there is little difference between talent trees.

On my higher level hunter I have roughly followed a MM/BM hybrid levelling spec, somewhat similar to one that Pike posted over on her blog. For the pet talents, I have continued a mostly dps-focussed build on Cunning and Ferocity families and a tanking/survivability build on Tenacity families. Where possible I have grabbed the Thunderstomp talent.

All talent choices for hunter and pet are loosely based around benefiting a "levelling" play style. I hate downtime. I hate micromanagement. Talents such as bloodthirsty are essential to this play style.

Cower is removed from autocasting. I wish I could just unlearn it completely! Growl, the family focus dump and special ability are all on autocast. Prowl is the notable exception; it's off autocast and is manually used when I remember to do so.

I use recount to track damage. I have Omen running to help manage aggro but am not carefully watching these while fighting. My testing is purely for PvE levelling. A typical test will start with a tame of a fresh pet. I will spam mend pet and/or feed the pet until it is happy then set its talents. Once ready, I clear the recount meters then go farm some mobs.

Testing starts with attacking single mobs that are around or just above the pet's new level. If the pet is the full 5 levels below me, I'll also ease back on my hunter's DPS. I repeat single mob fights until around 10,000 damage has been done (50k for the higher level hunter - its about 20 kills for each stage of the test). During this time I evaluate subjectively how well the pet holds aggro and watch for the effects of its family skills. Hunter shot rotation is some variant of serpent sting as opener, then arcane/auto with multishot where possible. I try to avoid aggroing multiple mobs during this stage but it does not invalidate the test if adds join in. Mend pet is used whenever needed. Aspect of the hawk is up for as long as possible, then aspect of the viper is used to regenerate mana. Hunter and pet keep moving and killing almost continually, fighting one mob at a time without much downtime.

Once the 10,000 damage threshold is reached, I pause for a bit and take a look at recount. If the pet's family skill is a DPS based one, I note the percentage of the pet's DPS this represents. Overall, for a BM/MM hybrid, the pet gets 30-55% of total dps for this type of fighting, probably averaging out at about 40%. Some family special skills get about 10-15% of the pet's damage.

After a brief rest and refit, I will switch to grinding on multiple targets. The pet will be sent in to attack a mob. Serpent sting will be applied, then I'll target another mob and send in the pet again. Dash/Dive is great for this as it allows the pet to pick up small groups of targets very quickly. Targets are again chosen based on the pet's level rather than the hunter. If the targets are lower than the pet, I'll grab bunches of 5 or so. Higher level targets are gathered in smaller groups, or staggered in waves but the pet is taking on at least 2 at a time. Mend pet is used where needed, often when the pet still has higher health than in single target fights. Shot rotation for the hunter is mostly about keeping serpent sting up on the multiple targets and burning down the most-damaged target quickly. Pet families with an AOE aggro skill are awesome for this kind of fight, but some pet target swapping is often needed to manage aggro on groups. Careful use of traps and AOE shots are added in where possible. Aspect of the viper is sometimes used to reduce hunter threat output, though it has to be balanced against the massive drop in DPS it causes.

Again, a further 10,000 damage is recorded in recount. The counters are not reset between these parts of the test. It is rare for me to lose aggro in fights of less than 4 targets but a string of crits and some healing can cause problems with larger groups. Few pets can survive groups of 6 or more at once unless the targets are quite low level.

After both sets of testing, a BM/MM build usually has the pet doing around 50% of total damage. This is usually higher than after the single target tests, mostly due to use of Viper to manage aggro. Similar percentages seem to apply even after volley is added into the shot rotation on the higher level hunter, though this is one of the things I'm testing in this current round.

A third round of testing is performed for those pets with non-dps based family skills. Those pets with snare-type abilities are usually tested against humanoid targets that flee. The snare is manually triggered, not autocast. Caster-debuff skills are tested against caster mobs. Hard to manage family skills like dust cloud are tested manually against suitable targets (groups of melee in this case). Recount is usually left running but it is hard to evaluate the effectiveness of the skill in terms of damage done.

Sometimes I will do some "extreme" testing on particularly promising pets, where they try taking on multiple mobs significantly above their level, or we will take on elites of similar level. Very few pets should take on elites more than a level or two above them in a solo hunter/pet scenario.

Pet looks, animations and sounds are observed throughout this process. These evaluations are going to be entirely subjective and will be mentioned only where they are particularly noticeable.

At this stage, only the same pet families as evaluated last time will be done. I still do not have the 51-point BM talent to tame exotics and most of the outland only pets will still be out of reach by the end of this test. I intend to test all other families with a lower level and higher level hunter.

No comments: