DdsT

Si je me souviens bien c'était la même chose au début de PoE. Tout ce qui était long texte était plutôt correct, mais l'interface c'était n'importe quoi et j'avais du passer en anglais pour bien comprendre les descriptions d'objets et de sorts. C'est un problème inhérent aux systèmes de traduction par placeholders, où plusieurs personnes vont traduire une liste de mots sans contexte (par exemple « hit »), et où d'autres traduisent des phrases à trou sans savoir quels mots vont être mis derrière (surtout si les phrases viennent de l'anglais, où parfois on peut se passer d'article et où la structure doit être intégralement modifiée pour ressembler à quelque chose dans la langue cible). Il ne faut pas jeter la pierre aux traducteurs mais bien au contrôle qualité inexistant (ou alors la décision de passer outre).

As indicated above, the maths behind are: Graze: -50% -> -1 ( because -0.5 is negative, and therefore the formula 1-1/(1-0.5) applies) Might: + 9% -> 0.09 Scaling: +15% -> 0.15 Pen.: +30% -> 0.3 SUM: 0.685 (because the sum -0,46 is negative and therefore formula 1/(1-0.46) applies) 20.1 * 0.685 = 13.77 (seems it is rounded down) It is unlogical and very confusing: it is very hard to evaluate the effect of a negative effect since the scale is different.

Would the good strategy be to get a high damage weapon for backstabs and swap to dual-wield when out of invisibility? Or are the recovery time not worth the effort?

It seems that no quality check has been performed regarding the correct implementation of the translated strings in-game. My guess is that the string tokens have been translated with no in-game context, the best example being the combat log. For dialogs, books, journal entries etc. it is not really a problem, as the person just have to translate the text and it will be displayed as it is. There might be some translations errors, depending on the translation skills, but most of the books and dialogs are OK (quality-wise, compared to what we find in the current game industry). The problematic part are the short strings, especially those with variables. Let's take for instance a typical combat log sentence: Eder hits Shade for 10 damage. This is what is displayed in-game. The game actually processes the following sentence (not 100% true but it looks almost the same): %character1% %action% %character2% for %damagenumber% %damagetype% The game picks the variable in a list of predefined strings based on the in-game events and calculations. Now what does a translator see if he does not play the game? Only this list of variable without context. For instance, he will have a list of %action% to translate: hits, grazes, misses... In english those words are twofolds: -the plural form of a noun -the third person singular of a verb (which is used in the combat log) You cannot do that in french. The translator has chosen the nouns instead the verbs, which does make the combat log sentence incorrect. Another possible cause whould be that the game uses the both meaning of the strings, for instance in the "converts grazes to hits" sentence. Then the translator cannot do anything and the correction shall occur on the developper side (by adding more variables). But that's just an assumption. Anyway, it is a pity that the end of the quality process have to be carried out by the community Most of those errors can be easely avoided if you gives enough time to someone in the translation team to see if the in-game implementation is correct.

The cipher specific ressource "Focus" has been translated "Concentration" and therefore shares the same name as the defensive stat against interrupts. It can leads to confusion, especially with the bonus given by the RES attribute. I will come back later with more information on the following subjects: In the combat log, hits/grazes/miss have directly been translated in their nominal form, but in french, the verb and the noun are different. The bonus from items and talents have the unecessary/missplaced preposition "de" in the tooltips.

I thought that the -1 MIG of the Orlan reduced the maximum Might to 17 (18 with the culture). If not then there is only a 2 might gap between the class, i.e. 4,7% more damage for the Aumaua compared to a 19 MIG Orlan. For accuracy>deflection, the bonus gets lower in comparison to the Aumaua not because of the lower chance of hit, but because of the lower chance of miss, which means that the MIG bonus from the Aumaua gets applied more often (whereas the hit conversion stay the same until a difference greater than 51). Lets assume you have an accuracy = 50 + deflection. You never miss, have 50% chance to hit and 50% chance to crit. The Orlan with 19 MIG and 30% hits to crit has then on average per attack a multiplier of (1+0.03*9)*((1+0.5*0.3)*0.5 + 1.5*0.5) = 1.68 The Aumaua with 21 MIG and 20% hits to crit has then on average per attack a multiplier of (1+0.03*11)*((1+0.5*0.2)*0.5 + 1.5*0.5) = 1.72 You can see that the Aumaua clearly wins in this case. Actually after calculation, even without base hit to crit conversion and with a 2 MIG gap, the Aumaua always wins the DPS race (lowest difference between the two factors is at accuracy=deflection, 0.889 for the Orlan, 0.898 for the Aumaua). The difference is quite negligible (1-2% increase for the Aumaua in most case) but it doesn't require you to attack the same target as an allie and it grants you an additionnal trait.

With a bit of theorycraft it is not so easy to choose: Let's assume we go 18 Might with the Orlan build 1 and 21 Might with the Aumaua build2. The +9% (3 Might difference) of build2 add a raw +7.2% on all damage (graze, hit and crit) compared to build1 without taking the Orlan trait into account. The +10% hit in crit conversion only applies on hits. Most of the time, the chance to perform a hit will be 50% (it can be lower if the accuracy is lower than the defense stat or at least 50 higher) and a crit adds 50% damage to your attack compared to a hit. Let's assume we have accuracy=defense, so we have the standard 15, 50 and 100 thresholds for miss, graze, hit and crit. The basic average damage formula (without the might bonus and other modifiers) for build2 would be 0.35*0.5+0.5*(1+x*0.5) with x being the current hit to crit rate. For build2, this would be 0.35*0.5+0.5*(1+(0.1+x)*0.5) In the proposed talent choice, x=0.2 most of the time. We then have: build1 average damage factor per attack: 0.75 build2 average damage factor per attack: 0.725 With the added talent from the Orlan, build1 do on average 3.4% more damage than build2 if not taking might into account. But this is the best case scenario. The following conditions lower this bonus: -no companion is attacking your target -accuracy<def: hits become more rare -accuracy>def: the added critical chance lower the overall weight of hits in the average damage calculation -higher x (coming from talent or buff) -lower crit damage bonus (firearms) It is safe to say than the Aumaua "max might" build does more damage and with a higher consistency. On top of that, you get the racial ability and the "brute" look of Aumaua fits a 3 INT build The disavantage is the loss of 3 points in RES and PER compared to the Orlan build (17/16 PER and 16/17 RES look still ok for dialog checks). Having more crit is also a bonus to interrupt. Please correct me if something is missing or wrong in my calculation.

Point 1: Does the racial and cultural bonuses allow going above 18? When I look at the thread about companion attributes, it seems so. If yes, regarding the Hearth Orlan Dual Wielding Rogue: - Why not lower the CON to 3 and rise PER or RES above 18? - Aumaua could then be able to reach 21 Might. Wouldn't it be more interessant DPSwise than a Orlan? If not, disregard those two questions Point 2: What would be a great culture and background for the above mentionned build? Does it have an impact on the dialog options? Point 3: Having 3 in INT in some games makes most of the dialog impossible. Is it the case here and do the high RES and PER make up for the loss of dialog option?