Friday, September 30, 2016

Daimonji-yama (大文字山), Kyoto

At the culmination of the Obon (お盆) festival, each year on August 16th in Kyoto, fires are lit on the hillsides surrounding the city. One of these fires is lit on Daimonji-yama (大文字山), which is right behind the main campus of Kyoto University. Unfortunately, I wasn't in Kyoto on August 16th but it should look like this:
Picture taken from wikipedia
In daytime, it looks like this:
Don't have my telephoto lens with me, this picture is borrowed off the internet
From the vicinity of my apartment building at dusk, it looks like this:
See the clearing up on the hillside?
It's actually rather close to Kyoto University. Here is my GPS tracklog:
red = bike ride, yellow = hike up to the 大, blue = hike up to the top
It's about an 8 min  bicycle ride (2.5 km) from Kyoto University to the Ginkaku-ji (銀閣寺), or Silver Pavilion. At the entrance, go left and you will find a short road that leads to the path up the Daimonji (大文字) mountain. I parked my bike at the end of the road:
My Bike Friday at the bike parking area
The wooden board in front shows the route:

It's a 20 min (1.25km) walk. There was a lot of water running down the side of the road because it'd rained a lot the last few days here:
Cascading water besides the road

You proceed up the dirt/cement road until you see steps to this bridge:
Take the steps, go over the bridge and turn left.
We turn off the road here but just ahead at the end of the road is this cable-driven platform for taking supplies up the mountain. Useful for festival time.

Leaving the road behind, we have a dirt path, sometimes with steps:

Nearing the clearing, we have stone steps:

At the top, we're at the center of the 大 character. There is a panoramic view of the city below:
Panorama taken at the center of the 大
The elevation here is about 331m. Note the small fire pits made from brick. Each one will be lit (and extinguished) by someone come festival time.

(BTW, modern cameras are amazing. I took the stitched picture here hand-held, no tripod, with a 35mm effective focal length lens, i.e. not a telephoto. And the camera is only a 16Mpixel model. A lot depends on atmospheric conditions and a tripod will eliminate a lot of blurring, but even so, the level of detail available from a casual setup on this slightly hazy afternoon is impressive. A 100% excerpt:
At 100%, Kyoto Tower is plainly visible 

The panorama at full resolution is available here: http://elmo.sbs.arizona.edu/sandiway/pics/P9290019-P9290026.jpg. 26MB.) I guess a calm day in winter with a 42Mpixel camera would be simply incredible.)

At this point you can climb up these steps to go on to the top of the mountain. It's another 0.85km, or about 15 mins. You're actually climbing the vertical part (of the 2nd stroke) of the 大 character initially.
Up the vertical of the 大
The clearing will disappear soon. And the path will go on up through the forest, past several saddle points, until you reach the top at 446m elevation. There is another panoramic view here:
Panorama from the top of the mountain
The top is marked here:
Daimonji-yama 3 corner point is at 466m
At the top there is another way down that goes to Nanzen-ji (南禅寺). Unfortunately, I can't go that way. I have to retrace my steps as I have my bicycle at the bottom behind Ginkaku-ji (銀閣寺). Another time then.

Sunday, September 25, 2016

A bike ride along Kyotoyawatakizu (京都八幡木津) Cycling Road and to Osaka castle

In the previous blogpost, I mentioned that my coffee run to Arashiyama (嵐山) along the Katsuragawa (桂川) used the Kyotoyawatakizu (京都八幡木津) Cycling Road. As the name indicates, it runs from Kyoto (京都), goes through Yawata (八幡), and finishes in Kizu (木津) about 45 km away. I was also told one can make use of this road to go all the way into the heart of Osaka (大阪); in fact, one can pop up next to Osaka castle (大阪城). How cool is that?

Well, it's such a cool concept that I had to ride it using the portable Bike Friday Pocket Rocket Pro that I recently brought to Japan in my checked luggage.
Bike Friday Pocket Rocket Pro all disassembled into its Samsonite suitcase

Reassembled in the apartment in Kyoto, this is what it looks like.

iPhone 6 Plus on the stem for Google Maps navigation. The aerobars are only used to hold accessories.
Here is the GPS tracklog of the route I took. I started from Kyoto University using the path along the Kamogawa (鴨川) to join up with the Kyotoyawatakizu Cycling Road at the confluence of the Kamogawa and the Katsuragawa. Then I rode the bike path to its terminus at Kizu. Turned around, and rode from Kizu to Osaka castle. Then turned around again and rode from Osaka back to Kyoto University. All in all, that adds up to a century, i.e. 161 km or 100 miles.

GPS tracklog recorded using my new Garmin Vivoactive HR

Notice that the bike paths are simply taking advantage of, and following, the path set by the local rivers (marked in yellow above). From Kyoto, it's along the Katsuragawa and Kamogawa. To Kizu, we simply follow the Kizugawa. To Osaka, and ultimately out to sea at Osaka Bay, it's the Yodogawa. The Cycling Road is nearly completely paved with tarmac and  built on paths along river banks or on top of flood embankments. Sometimes it seems like it splits into multiple paths that proceed in parallel. Then it's a question of spotting the next road, railway or utility bridge, and figuring which path stops and which path ducks under the bridge. Signposting exists, though not always where you need them for disambiguation. On a weekend, the route is full of cyclists, young and old. You will see groups of club cyclists in a paceline wearing their club jerseys, guys with gear in their baskets going fishing along the river, a line of kids in their little league uniforms cycling to riverside practice grounds, walkers, people pushing their bikes, mothers with babies, runners of all ages and styles, even Japanese men dressed in Arab robes - you get the idea. Along the river, you may see housing estates, golf courses, softball fields, industry, railways, allotments where hobbyists grow vegetables, farms, people camping in tents, ducks, naked men washing their clothes in the river, people doing tai-chi, couples sitting on the banks - sometimes sleeping there overnight, whole families enjoying a picnic in the afternoon, or ordinary people sitting on the embankment watching the river. In summary, it seems like a good way to see a fairly wide cross-section of Japanese society. Not many tourists though, unless we're talking about downtown Kyoto or Osaka. One thing I didn't see any of is watercraft until I got well into Osaka; there I spotted some tourist barges. But along the way, no boats, no barges, no jet skis, no kayaks, no paddleboards, and no swimmers (the rivers were flowing swiftly).

A route map along the Kyotoyawatakizu Cycling Road

Here is a signpost. On one side it says to Kizu (12.2km), on the other side to Arashiyama (32.8 km):



I have to say the end of the cycling route is rather disappointing. The start is near the beautiful Togetsukyō (渡月橋) in Arashiyama.  The route peters out at an ordinary road intersection. Since Kizu is near Nara, it would be nice there the bike path could be extended all the way into Nara.

Cycling Road terminus

So I decided to backtrack from Kizu northwards along the Kizugawa, and eventually south onto the Yodogawa (淀川) down into Osaka. This is near the confluence of the rivers in Yawata (八幡). The path on the Kamogawa is just a few hundred meters away over two back-to-back bridges.

Osaka Bay 37.0 km away

Arriving in Osaka, one can continue ahead to the high-rise district of Umeda (梅田) (in the far distance), or turn left along the channel to Osaka castle.

Following the water, it appears this is as far along the bike path as one can get. Notice you can make out Osaka castle 400m away. The board explain how a bicycle can get there taking two city streets and one bridge:

The end of the bike path along this section. Osaka castle is visible in the background.

Basically, at this point the Keihan railway line is in the way. Those stairs go down to the right and up under the line. Alternatively, as recommended on the board, ride down the street until you can duck under the line.

Entrance to Osaka Castle
Navigation came courtesy of Google Maps running on my iPhone 6 Plus velcroed to the stem. A iPhone Lightning to USB cable can be seen running from the iPhone down to the main tube and thereon down to the Profile Design bento box pressed into service to hold a 7000mAh Elecom USB battery. The battery has two USB outputs; the other output is actually connected to a WIMAX hotspot, also stored inside the bento box.

Bike Friday by the moat. I am not here to fish:

The moat at Osaka castle

Allow me to emphasize how cool this is: it'd be a bit like riding from Princeton, New Jersey along a bike path and somehow popping up next to the Rockefeller Center or Times Square in New York, having avoided all motorized traffic and traffic lights.

Unfortunately, the Yodogawa section of the route is not so friendly for an entirely different reason: there are frequently gates or barriers that are designed to basically force cyclists to dismount:

Along the Yodogawa
The one that can be seen in the background can be ridden through without dismounting if you're careful to unclip from the pedals and keep the cranks horizontal, i.e. at 3 o'clock and 9 o'clock. But many of them are specifically designed to prevent even that. The motivating idea must be to slow cyclists down in areas where kids may be playing. I found them so annoying I preferred sometimes to take the road when one exists alongside the path. Fortunately, it is only along the Yodogawa that this happens.

Anyway, it's about 53 km from Osaka back to Kyoto University. And at 100 miles and 3000 kcal total, overall I'm a pretty happy camper:

Saturday, September 24, 2016

Arashiyama (嵐山) Coffee Run

I finally managed it at the 4th attempt! I ran 14.2 miles, slightly over half marathon distance, from Kyoto University 's main campus to beautiful Arashiyama (嵐山), following the Kamo River (鴨川) downstream and then up the Katsura River (桂川), for a cup of coffee at % Arabica.

GPS track log of the run

Of course, Arashiyama is actually only about 6 miles away from Kyoto University as the crow flies (westwards). And one can do that on busy city streets but the idea here is to do the run using the bike/running paths along the rivers in Kyoto. The running paths along both sides of the Kamogawa through the city center, in particular, are iconic for me:

View of the Kamogawa looking northwards from the Shijō (四条) bridge
The Katsuragawa is also quite pretty and unspoilt in places; for example, especially as it approaches Arashiyama:

Katsuragawa near Arashiyama
This part of the course is also part of the Kyotoyawatakizu (京都八幡木津) Cycling Road. (More on this route in my next post.) The Cycling Road terminates in Arashiyama at the Togetsukyō (渡月橋):

Togetsukyō (渡月橋) in Arashiyama (嵐山)
The poetically-named bridge translates as "moon crossing bridge". (Emeritus Emperor Kameyama in the 14th century is supposed to be the one who came up with the name after observing it on a full moon.)

My destination is the inconspicuous white coffee shop, named % Arabica, on the other side of the bridge.

% Arabica is the white coffee shop on the other side of the bridge
It has a world-class view in Fall and Spring, plus look at that espresso machine from Slayer
The view from % Arabica

After coming across the shop during a visit to Japan in February 2016, the germ of an idea to run here for coffee began to form in my mind. Originally, I thought the Katsuragawa simply ran down into Kyoto past the university. As the GPS tracklog indicates, I have to switch rivers from the Kamogawa first, making the distance much longer than a straight line.

(You may wonder: how did I get back? For the record I took the number 93 bus (price: 230 yen). It basically goes straight east and I got off very close to the campus.)

It's a good coffee shop. Worth the run or bike ride in my humble opinion. My cappuccino:


Since I did this in July, the heat and humidity that is omnipresent in the basin that is Kyoto kicked my butt. (See the rise in HR near the end.) On the plus side, I got to burn 2000 kcal.
Finally, here is a 3min movie of the run I made from the GPS tracklog using Google Earth Pro. It's nice to see detailed 3D buildings, even in Kyoto:





Sunday, January 10, 2016

Table Tennis: The Reverse Penhold Backhand (RPB)

The traditional penhold grip uses the same side of the paddle for backhand and forehand strokes. The newer penhold style with the Reverse Penhold Backhand (RPB) stroke follows the more popular shakehands grips in using both sides of the paddle.
(Ma Lin, RPB, from ITTF website)

The shakehands grips dominates the professional game because it is possible to generate heavy topspin from both sides. With traditional penhold, only the forehand can generate significant topspin; the backhand is limited to the block or flat hit.

(Kim Taek Soo, backhand flat hit, from ITTF website)
(Moon Hyun Jung, backhand block, from ITTF website)

This, of course, limits the attacking possibilities for the traditional penholder. By standing further to the backhand side, additional footwork can be used to cover more of the playing angles with the forehand, but having to cover more ground with the forehand has become a relative disadvantage. Nevertheless, penhold has enjoyed considerable success historically; for example, the current head honcho in the Chinese national team, Liu Guoliang (刘国梁), was a penholder in his playing days (achieving both world singles champion and Olympics titles), but in recent years, it has become clear that, at the world class level at least, the traditional penhold grip is outdated and dying out. In the past, both South Korea and Japan have had penhold success, but there hasn't been a top penhold player in either team for a while, and there appear to be none emerging on the junior side. The last significant penholder from South Korea was Ryu Seungmin (유승민; hanja: 柳承敏), born 1982 and Olympic champion in 2004. On the female side, aside from a few former top Chinese players representing European countries, penhold can be effectively declared extinct.

The first player who was successful at adapting the shakehands style to the penhold grip was Wang Hao (王皓), born 1983 (now retired), who became world singles champion in 2009 (Yokohama).
(Wang Hao, Olympics 2008, from gettyimages)

He used the other side of the paddle to execute his signature backhand topspin; this stroke is known as the Reverse Penhold Backhand (RPB). Ma Lin (马琳), see first picture, born 1980 but more properly belonging to the traditional penhold era, was a penholder able to play both styles, became Olympic champion in 2008 (Beijing) but never world singles champion, also now retired.  In the modern game, there are only a few top penholders left, all of whom use the RPB on the backhand side. In my opinion, currently the best two penhold players on the world circuit are Xu Xin (许昕), born 1990, and Wong Chun Ting (黃鎮廷), born 1991.
(Xu Xin, from tabletennisdaily.co.uk)

Xu Xin, who probably has the strongest forehand in the world and incredible footwork, obviously favors his forehand side and uses the RPB only when forced to. On the other hand, Wong Chun Ting favors his RPB, and his forehand is considerably less well developed.
(Wong Chun Ting, from http://cdn.tabletennista.net)

For those wishing to emulate the RPB of the professionals, I recommend studying youtube videos of Wang Hao, Wong Chun Ting, and Xu Xin (in that order). In particular, it is my belief that Wang Hao's RPB has not yet been equalled.

As an amateur (purely at the club level), in recent years I've been learning the RPB. Since table tennis is a sport where quickness is perhaps the most important limiting factor, muscle memory invariably dominates. As professional coaches repeatedly point out to me, if one has to think through and consciously adjust the mechanics of a stroke during a game, one hasn't really properly internalized the stroke. Therefore, for penholders like myself who began with the traditional penhold game; switching over to the RPB is a constant struggle and source of frustration: the degree of difficulty of which should not be underestimated. Practically speaking, despite training exclusively with the RPB for the last few years, even now, under pressure at times I inadvertently revert back to jabbing the ball using the traditional penhold backhand stroke initially programmed into my muscle memory.

As I mentioned above, those who wish to emulate the relative perfection of the world class players would be well served by studying videos of Wang Hao, Wong Chun Ting, and Xu Xin. These talented athletes make very difficult shots look easy and natural. In addition, the amount of spin and power they achieve with the RPB can be quite astonishing. But I believe there is sometimes value and something to be gained in studying the flawed strokes of ungifted amateur play too. More precisely, one can spot technical errors (not present at the world class level), and see if (A) one has the same flaws, and (B) perhaps take corrective action. Pursuing the line of thinking outlined above, I humbly offer the following video clips of me training at the South China (colloquially known as 南华club (formally, the 世纪南华乒乓球俱乐部) in Luohu (罗湖) district, Shenzhen, just across the border from Hong Kong. The video clips were all taken on day 3 (my last day) of an all-too-brief trip in Jan 2016.

The coach is He Qiming (启明), the current manager of the club and a native of Hunan province. I've taken the liberty of setting the embedded video to play at a minimum quality level of vq=hd720 (the original uploaded content is at 60fps/1080p), and I've annotated the videos with comments(using iMovie). My Olympus OM-D E-M5 MkII camera that I used to shoot the videos unfortunately only has an on-board microphone, but for those who can understand Cantonese (unfortunately, I am not a Mandarin speaker), perhaps you can hear the coaching comments first hand.


In the above drill, I use three variants of RPB stroke for play at three different distances from the table: (1) over the table (the return of serve), (2) close to the table, and (3) at middle distance. To be a competent at RPB, I believe one needs to learn all three variations.

Here is a similar video but with a bit more instructional content:


The middle distance RPB



Also here:


The close-to-the-table RPB


In my own opinion, my forehand has considerably more significant flaws than my backhand now. I count that as a small but significant indicator of progress. Of course, I learned the RPB with significant professional coaching right from the time I began retooling my backhand, but my forehand loop is home-grown and mechanically ugly. However, it is reasonably effective at the club level, and, somewhat ironically wins me many more points than the coached RPB. The topic of retooling the forehand is left as a possible subject for future posts.

Wednesday, June 3, 2015

Emotiva XMC-1 and Dirac Live

Introduction

All normal rooms have modes that induce standing waves that interfere destructively with sound reproduction.  (See here and here for two introductory articles by Michel Leduc on the topic of Listening Room Acoustics and why it matters.) Loudspeakers interact with the room bouncing sound waves off walls, ceilings, and floors, possibly muddying the sound received at the listening position. It gets pretty complicated fast. Proper speaker and hearer placement (see relevant Floyd O'Toole articles here and here), room treatment, and digital signal processing methods (aka Room Correction) can mitigate many of these undesirable effects and clearly improve sound reproduction.
from http://www.hunecke.de/en/calculators/room-eigenmodes.html
The Emotiva XMC-1 processor has built-in Dirac Live room correction. However, it needs the assistance of a computer (and an internet connection) at calibration time. This means you run Dirac Live on a computer, measure the room at the listening position, and download wirelessly the computed filters to the XMC-1's single Dirac preset. If your music is digital, this is all done in the PCM domain without any extra analog/digital conversion steps. Once dialed in properly, the difference between an uncorrected and corrected room is huge and plainly obvious.

Dirac Live is a DSP technology that corrects for both frequency-domain and time-domain (impulse) response. Dataset's RS20i (see here) also incorporates Dirac Live but is priced 10x higher. Trinnov is another company that has its own frequency and time domain correction algorithms, integrating an Intel PC into the box, but again at a much higher price point. Some competing systems typically found in home theater receivers, e.g. Audyssey, Yamaha's YPAO, Pioneer's MCACC, and Anthem's ARC (I believe), perform frequency domain correction only.

Another option would be to have Dirac Live in its own box separate from the pre-amp/processor, like with MiniDSP's nanoAVR DL (see here). (Unfortunately, the nanoAVR DL requires Windows.) Computer audiophiles with a pre-ripped music library on a server or hard drive can purchase Dirac Live as a generic PC/Mac application that can integrate with various player apps.

The main point of the XMC-1 processor for me is having a DSP stage with Dirac capabilities integrated with D/A  conversion. (The XMC-1 uses fairly high-end Burr-Brown 24/192 DSD1796 DACs that are hopefully excellent.) The Oppo player (see picture below) can function as a pure transport in the sense that it can send digital audio from either physical media (e.g. cd, sacd, dvd-a, blu-ray) or from a USB-connected hard drive to the XMC-1 for decoding and D/A conversion prior to power amplification. (In other words, the Oppo doesn't perform D/A conversion in my system - although it could, using its own  Cirrus Logic CS4382A DAC.)

Dirac Live for the Emotiva XMC-1

Unfortunately for those of us who are OSX-based, the free Dirac LE that comes with the XMC-1 is not yet available. However you can pony up $100 and spring for the full Dirac Live (like I did). In either case, I believe Emotiva supplies an activation code. 
Once activated, there's a page with links to download the Dirac application for Mac or Windows:
This Dirac app is specific to Emotiva's XMC-1 because it contains Emotiva-specific code that communicates with (and controls) the XMC-1.

The Local Area Network (LAN)

Assuming you're able to have the XMC-1 and computer see each on the LAN, the procedure is quite straightforward.  (Unfortunately, DHCP on the XMC-1 doesn't play well with my Netgear router, I ended up having to turn DHCP off and use a static local IP.)
(If all is well, you should be able to ping and login remotely to the XMC-1.) 
Note also that the XMC-1 curiously does not have built-in wifi. However, there's an Ethernet port on the back:
You could plug a wifi adapter into the port. I've verified it works. But since the XMC-1 seems finicky,  for reliability I chose to run an Ethernet cable from the router directly to the XMC-1.

Dirac Sound Quality

Though one should optimize speaker and listening position as far as possible given particular room parameters before running Dirac, there always seems to be room for improvement. (This is where REW is very handy, see my previous blog post on using REW here.) 

Frequency-wise, the benefits of Dirac are very apparent; it cleans up resonances. Not sure if it really attempts to deal with suckouts though (because those can be caused by room interactions).  The soundstage is also massively expanded. However, I find Dirac compromises (pin-point) imaging to the point I prefer to listen to Preset 1 (no Dirac, distances and levels set manually) - unless one is very careful and consistent with the measurements. (I will attempt to document these steps here below.)

The USB Microphone

An EMM-1 mic ships with the XMC-1. Its calibration file is supposedly doctored to produce an Emotiva-approved house curve. 

However, I used an UMIK-1 USB mic, individually calibrated by Cross Spectrum Labs, which I bought for REW use. The mic plugs into your computer (not the XMC-1) and the Dirac application recognizes it. 

Running the Emotiva Dirac app on a laptop is particularly convenient since you can move it close to the XMC-1:

With a desktop, there may be cabling problems. In my case, the supplied USB cable was not long enough and I needed to use an extension. (I also need a HDMI extension cable in order to run REW. I use REW in conjunction with the UMIK-1 as an SPL meter to check levels.)

Running Dirac Live for Emotiva

OK, let's fire up the Dirac application on OS X. It should see the XMC-1 on the LAN. 
If it doesn''t, obviously you need to fix the network issues first, otherwise you can't proceed. Once you press OK, the XMC-1 is under the control of the application and it no longer responds to the remote or front panel buttons (except, I suppose, the big turn on/standby button when things get stuck). The panel will display "connected". 

The application itself is divided into thematic tabs shown on the left, and the user is taken step-by-step through them. The Sound system tab refers to the speaker configuration already set up on the XMC-1 device. (You need to do that before running Dirac.)
My subwoofer still hasn't arrived yet, so I'm using it in 5.0 mode: front left and right speakers, a center channel and left and right surrounds. 

(There might be a bug with respect to levels if you don't use a subwoofer, and level adjustment might be required after Dirac.) 

Next, you need to select your mic.
Notice here I'm using the UMIK-1 with the 90 degree narrow band calibration file supplied by Cross Spectrum Labs. (The XMC-1 ships with a EMM-1 mic that works fine except that it has a pre-programmed calibration file that reflects Emotiva's preferences for a house curve.) 

Next I need to set the input gain. There are two steps. Dial down the input gain slider until the levels are just below -24dB. Then for each output channel in turn, hit Test and adjust the Output volume slider until the level hits the middle of the -12dB green zone.

Because I'm using different amplification to run the center channel, the sensitivity of that channel needs to be dialed down separately using the Channel volume slider to stay within the green zone. 
I had to dial in some trim with the surrounds too. (I'm using different amplification for them too.)
Finally, we are ready to take measurements. Dirac Live takes 9 measurements, the first is at the sweet spot, i.e. the main listening position, and the other 8 are tightly clustered around the sweet spot. 

Having performed measurements multiple times, I find the results really disappointing with respect to proper imaging unless I proceed very carefully and keep the cluster of points fairly compact. (Note: Dirac warns that the points be not too compact, the downside being over-fitted or optimized "lifeless" sound.)

The UMIK-1 (and the EMM-1) mics seem extremely sensitive. I believe one has to be very careful with spot selection. Probably, 6 inches of deviation would be massive. For consistency, I mark off the horizontal points on the back of the sofa with bits of tape. (Here I've spaced them out about 7 inches apart. I dare not go any tighter. More than 12 inches apart, I find the imaging loses its tightness. Obviously, given the angles subtended, it'd depend on how far away you sit. I'm about 10' 7" from each of the front speakers, as measured by laser to the main listening position.)
The Dirac manual states:
"Avoid making measurements in a too small space. Even for the “Chair” listening environment, it is important to spread out the microphone positions in a sphere of at least 1 meter of diameter. A too small space will result in over-
compensation that will sound very dry and dull."
I found I needed to use three different tripods to achieve reasonable vertical consistency between the high and low points, front and rear. (It would be nice if someone came up with an adjustable multi-parallelogram mesh rig with 9 points for easy and accurate mic placement.)
Example: high front
Example: low rear

Example: high rear
Each time I do not go to all this trouble, I find I've wasted time. The Dirac imaging is horribly diffused, skewed and sloppy compared to Preset 1 (where I have manually adjusted levels and distances). 

As mentioned previously, you're prompted for in turn to set up 9 different microphone positions, beginning with the sweet spot.
It'll generate graphs of the frequency response curves for various speaker groups, e.g. front left and right, center, surrounds etc.  And an editable target curve:
Front left and right speakers
Here are the corresponding plots for the center and surround channels:
Center channel
Left and right surrounds
Computing the filters gratuitously requires your computer to have internet access to Dirac's servers, which is an attempt to keep licensing and use firmly under Dirac's control. The resulting filters will be downloaded to the XMC-1 by Dirac Live for Emotiva via the LAN. The filters are also verified after downloading:
Unfortunately, on Mavericks, this is where it often crashes on me. If successful, a confirmatory dialog box will be displayed:
On Mavericks the verification fails to complete about half the time:


So it pays to save the project before download to save re-measuring the room. On Yosemite, I've not had any crashes. After success, the application disconnects from the XMC-1, and the remote works again.

Adjusting Levels

For reasons that I'm still not entirely clear about (possibly related to the differing gain of the various amplifiers I'm using or that I'm calibrating without a subwoofer), post-Dirac I find I always need to go in and equalize the levels of each channel. (The XMC-1 has a built-in test tone generator at 65/75/85dB levels.) I'm really not sure why this is not done for me automatically. 

System architecture-wise, levels are adjusted in the analog domain and follow the DSP stage. 

For example, here are my current post-Dirac levels (properly adjusted):
And just for comparison, I have my Preset 1 levels manually adjusted as follows:
The front left and right speakers are driven by two channels from a stereo tube amplifier. The center channel is driven by a different tube amplifier, which happens to be a monoblock. 

The surrounds are driven by yet another stereo amplifier; this time a solid-state amp. All three amplifiers have different gain, so these differences are not surprising. 

Distances are also manually programmed into Preset 1. (I use a laser distance measurement tool and a piece of white cardboard as the target.)
Unfortunately, it's not possible to see what the computed distances are for the Dirac Preset.

Finally, I leave Preset 2 untouched. It has distances set at 0'0" and no gain is applied on any of the channels.

Conclusions

When I first bought the XMC-1, I thought I'd made a mistake. What's the use of room correction if it messes up the sound in some critical way? I guess I'd expected it to be plug-n-play. It's plainly not that kind of hardware or technology. Although Dirac Live did a great job with the frequency response (and presumably phase coherency between speakers), plus the soundstage is much larger; however, I was unwilling to pay the price of inferior pinpoint stereo imaging. The Speaker Preset button on the remote allows you to turn Dirac on/off with just a short (1 sec) delay. This make A/B comparisons easy and obvious. And the differences are huge. Careful positioning of the microphone is critical; but the reward is room correction with no significant downsides as far as I can tell. A purist might be horrified, but technology marches onwards and upwards...