I did the Rolltables method. It took a little setting up, and one player who'd used the API method grumbled a bit because it wasn't as flashy. At first it took a little time to read the dice, but by games 2 onward you could look at the dice (just like you'd do at a real table) and very quickly work out the summary. Steps: 1) Go to the online EotE roller and use windows snipping tool to grab images of the dice. 2) Upload the images to Roll20 and build the rolltables for each dice. Use the table in the book to set probabilities (you don't have to upload the same image more than once). 3) Create a macro to roll the rolltables. I preferred one that popped up with text boxes asking for the amount of each dice. This way once you learned the order you could quickly mash out the entries on a numpad. 1 enter 4 enter 3 enter etc. My advice here is put the GM determined dice last (boost, setback, difficulty, challenge) so the player can be imputing his dice (ability, proficiency) while you decide the others. I found this to be a very respectable free option. The ruleset itself worked, but didn't grip me as much as I'd hoped. After a while coming up with unique advantage / disadvantages felt clunky, there was too much bogging down to work out "Ok, you failed but with 4 advantage, so... whats that mean?" The dozenth time (hyperbole) your character misses but miraculously hits a fuel barrel - which is arguably better than having hit in the first place - it starts to feel... strained. Add to this the wishy'washy'ness. Whats the difference between 2 advantage and 3? You could say nothing, but for the player that rolled it he's going to be disappointed to learn better rolls aren't actually better. I also felt the class system was a little bland. I thought the grid idea was fine, and those grids being freely distributable (the book says you're free to share them) worked well with Roll20. I could get them to people who didn't have the book. However there were a lot of Traits that just weren't very fun. Maybe they were "good", but being "good" and being fun aren't the same. Take the ones that reduce the number of setback dice. I found myself feeling obligated to add setback to tasks that wouldn't have had them just so players could feel those traits were useful. Or the ones that halved the time to perform a task. What if time doesn't matter? Which, for something like hacking, could be a lot of the time. I had a fun and memorable campaign with Edge, and it was an intriguing experience using the narrative dice. However I don't feel any urge to go back to it for another campaign. I've also had a lot more cinematic moments come out of the Cypher System (Numenera, the Strange) than Edge... but that's a different topic.