During the following series of articles I will analyze certain aspects of my Squattertown project. Created as trans-media project, Squattertown was a huge test run and experimental lab for my crew and me. We wanted to see how far we can go. Can we actually pull it off? And what lessons we can take away for the projects to come. In this respect, Squattertown was a full success. The amount of information collected is immense and versatile. However, as a living project over the past 14 months it adopted to its project web channels as those kept changing themselves. I will offer a transparent and honest analysis on every main channel we used. However, I cannot reveal all details as some of my crew members make a living with this sort of internet business.

I will separate my analysis into a number of sub-topics and focus sole on the time period of August 8, 2010 (the day I launched the Indiegogo campaign) to October 11, 2011. If you look at the numbers they won’t be that impressing nor overwhelming compared to other pages or film project. But I think we achieved quite a lot as small indie HK flick within the extreme narrow budget of HK$20K. On top, you most definitely won’t find a more transparent film project analysis within the Greater China region. So here we go:




Facebook was one of the project’s main channels but also the channel that went through the most severe changed during the past 14 months. Primarily, I used the Facebbok page as the central platform to post all updates on the project throughout the entire production period. It was supposed to be a collection of all activities on the web that dealt with Squattertown.

LIKES (or Fans as it was called when we started)

As of October 12, 2011 we got 168 likes. Now the official FB page insights get a bit confusing since they list a total of 177 likes since the beginning. However, as there are 32 people who unliked or better unsubscribed to the project page throughout the course of the project the actual number of people who subscribed to the page was somewhere around 200. That is low compared to my previous UCP (https://www.facebook.com/ucp2009) project back in 2009. We reached a number of 180 likes within just 2 weeks. The peak here was somewhere around 240. Even after 2 years online the page still got 192.

Now, what happened within these 2 years? Back in 2009 FB pages were something new, something different to the groups who were about to start dying. And obviously FB is just about to cut down all open groups with low interactions. (Closed groups seem to remain untouched for now.) So inviting someone to a page and engaging him/her to like it was quite easy and we actually went over 100 within 2 days on the UCP. But during fall 2010 things had changed, FB was flooded with pages, most monetized. People’s notification inboxes were crammed with page and game suggestions from friends one barely knew. So driving someone on your project page and making them engage became a tricky thing.

The like numbers gradually went up, steady but slow. However, there was a significant difference within those likes compared to the UCP. The community that was build with Squattertown was smaller but far more active on the page. The number of interactions was higher. They were actually that high, that I could figure out a posting strategy when and what to post with the maximum impact.

FB on the other hand, since pages became such a great monetary value to them, refined the stats options and analytic tools. This all helped throughout the project tremendously but I ended up changing the strategy every couple of weeks which is quite time consuming.



When we now look at the “active users” graph you can clearly see two big peaks. The first one starts mid of November 2010, roughly around the time when my second campaign on MySherpas got off the ground due to the great press work and help by the German team of this crowd-sponsoring platform. The second and far more distinct peak was the premiere event end of July 2011. In all FB insight graphs you will find an enormous peak at this time spot. So let me explain a bit more in detail what happened here:

On July 31, 2011 we had our official (physical) premiere in Hong Kong at the venue of Videotage. For this event we teamed up with a new Austrian start-up company, RealLifeConnect (http://blog.reallifeconnect.com/), that provides RFID card check-in solutions for events. Guests of the premiere received a special Squattertown RFID card with which they were able to check into the event or liking the event. Whenever someone swiped a card at a venue’s scanners it would result in a small message post on his/her FB profile, telling their friends what they are doing right now. A special was our photo booth where guests could take on one of the series’ costumes and by swiping their RFID cards such photos would also be posted to their FB profiles. (photo booth pictures: https://www.facebook.com/Squattertown?sk=app_167168409988996)

Now I cannot go into details on all the numbers and interactions that were aggregated that night since it is RealLifeConnect’s daily bread and butter but the concept has proven most viable and useful for a on-site film promotion. We were able to connect the event with the internet in real time, having friends of guests interacting on our internet channels. The peak you see in the graphs results from people who cycle back from their friends profiles to the project page. However, it does not reflect how many impressions or posts we had due to the check-ins. Those were far higher.


Now, we did gain a good number of new “followers” to the page on that night as well as during the following days. However, about one month after the premiere a significant change in “like” (fan) numbers started to emerge and still continues. Unlikes are taking place, constantly almost on a daily basis. So here is my theory:

This phenomenon became apparent after the launch of Google+ and FB’s subsequent radical change of key features on its site. For example, the site now offers friend lists (equivalent to the G+ circles) and one can subscribe to a strangers wall without the need of being friend with him. The individual profile sees a tremendous boost, which is correlating with the recent trend of real-identity on the web. G+ is criticized for demanding its members to reveal real profile names, but that is where the money will be in the web of the future. Hence, we will see on FB a significant shift in actual online value from the page toward the profile.

Since the war of believes emerged when G+ launched (Google fans against FB fans) a huge number of Facebookers are starting to update and clean out their profile pages. This is indeed a necessary step as their profiles will eventually be openly accessible blogs. Let’s face it. FB’s original principles of being exclusive, establishing and communicating with your own circle of friends only, are long gone. The site will open up your data step by step, no matter what privacy settings you tick.

Hence, during all those profile clean-ups small projects like Squattertown get kicked out by people. I experienced this behavior quite a lot among my own FB-friends. The day they unlike the page, they added their old school, past working employments and what not. Most unlikes are indeed people from Hong Kong. G+ does not play a big role in this city, but FB became like a business card, a second me. And you better keep that neat and tidy!

So what I am currently experiencing is a shift in fans. Since the start Hong Kong fans were dominant. Now they begin to drop. New likes coming primarily from Western countries. And there is actually a slight trend towards South America.


Since we started off the project under the objective to create a niche sub-genre within the Western it was clear female fans would be rather scarce. When we look  at the 14 months graph we got 31% female and 68% male fans, by having the strongest audience between 25-34. The latter one was a success as we specifically targeted this group with our trans-media concept and online presence. However, most the time during the project, as well as on other channel’s stats we had a female-male ration of 25% to 75%. At the beginning of the project even higher, around 15% to 85%. So I would say we were able to get more female fans into the project than expected, even though I was hoping for a number around 40%.

The four top countries are no actual surprise as those were the ones we conducted our fund-raising campaign at. Ultimately a FB page is a good tool to connect to a very international fan base in a short time. But by having fans from different countries and cultures raises the question of language. In which language should updates be posted? Should there be a bilingual or even multiple translation on each information/content? How do you engage a great number of fans to interact with your content via certain languages?

The Squattertown page fans come from 20 different countries and I usually posted updates in English, occasionally in German. There have been a couple of Portuguese comments but interactions were mostly in English. Recently, FB implemented a Bing translation tool,but I still got no viable answers to my questions above.

I think there could have been more interactions when I would have posted all updates in English and Chinese simultaneously, considering our main target was Hong Kong. Thus, I believe a more specifically narrowed targeting on one country alone can give you higher interaction rate and maybe even more fans. But ultimately Squattertown was supposed to be international and I was constantly trying to figure out where our limitations were.

Cycling back to the unlike-issue, to a greater extend I believe the abandoning of this quite local HK project by local people is also rooted within the language issue. Looking at other film related local FB pages it becomes evident that those with big like numbers and interactions are sole in Chinese. So I do believe to a certain degree I focused not enough on the project’s local aspects, needs and potential. there was a lot left on the table…


I think the graphs are self-explanatory here. Page views were relatively constant. Usually one can say each peak was one project update. During the fund-raising period views were always between 100-200. during the production and post-production it went down a bit as I was more focusing on the actual project rather than writing updates. So the lesson her would be, employ someone who gives updates and maintains the page. But then again, I wanted people to know when there is an update that they could rely on it came from me, personally.

Now again, the premiere event blows the chart out of proportion. But I think more important are referrals here. Since the project had a broad presence on the web it was a bit hard to determine who or what drives views/clicks/people from where to where. But you will see my personal web page was significant in terms of driving people to the page, as was Google/YouTube. there are also other project channels linking here as well as several blogs and pages I teamed up with during the production period.

However, don’t underestimate the impact a simple link in your email signature can have, still, after all those years and social media! However, driving people to the FB page was just the strategy for the first couple of months, during fund-raising. By December 2010/January 2011 I started primarily driving clicks to the main project page http://squattertown.com (from which all project channels are being linked with) as well as the YouTube channel. Now, one of my biggest mistakes here was not to track the URL links I sent out or simply limit to a set of links. Thus, I cannot provide further detail on this issue.

When it comes to media consumption, one thing is for certain, biggest interaction driver are photos. Videos are only relevant when they are short and one can tag a large number of people in them. Un-tagged videos have a far lesser chance of being watched. We do not have any audio items on the page so there is no analysis on this aspect. In my previous experience with UCP, audio was quite a hassle as one needs to apply for the app to be activated on the page by FB. Last time this took about 3 months or so as they are checking for potential copyright infringements.



For the analysis of the YouTube channels as one of our three video platforms I will mainly focus on the teaser since view numbers and stats for the four webisodes are still rising steady. I suspect to evaluate an online video run would need about 6-9 months until there is enough material for a thorough evaluation.


The teaser was launched on March 19th, 2011. Primarily spread throughout the project’s social media channels, it was then further shared by fans and friends. As you can see on the graph, there were 2 minor spikes in April and May. Views were mainly driven by the YouTube channel or embedded views which, to a great extend, involved Facebook, my own website and other fan/crew pages/blogs. Then there is this immense spike in beginning of June. This is when CNNgo published an article on Squattertown and embedded the teaser to the top of the page. You can also see how diversified the view sources became. Other pages and blogs started to link to the teaser. The aftershock of the article went on for about 2-3 weeks until end of July, when the premiere event took place. You will also recognize, that the CNNgo article turned the majority of the view sources from YouTube to embedded views. After the premiere, YouTube views became almost redundant, leaving only the actual channel views inside the stats.

This showed me how important actual press work is to a project. One cannot just rely on social media. Most young filmmakers believe social media is a great promotion tool for their film but in the end it is what it always was, conversation and networking. Promotion and advertising are killing those two characteristics. Hence, one has to find methods and ways to create a conversation online that eventually leads to your project. Pointing with the finger at something will alienate most.



YouTube offers a so called Hot Spots tool within its stats. It basically tells which parts of the video are attractive to viewers and subsequently how “hot” it is. This tool won’t be available for videos below 1K view numbers, apparently because there is not enough data. I checked this graph a number of times and it keeps indeed changing quite a lot. I have seen this green graph starting high and having a deep valley toward the end.

For a filmmaker this tool is questionable. Somehow nonetheless, I hold it quite interesting as it links up (presumable) audience attention/interest directly with the images. That way one can check if certain shots have the intended reaction on the viewers. However, this should be taken as reference only and not as absolute.

Btw, YouTube recently published a site manual in which they argue content creators should focus on the first 15secs to make them as enticing  and engaging as possible. Apparently, according to this graph the teaser fails in this respect, at least for now.



Even though it is too early for an in-depth look on the YouTube numbers I believe there is already a lesson to learn when we compare the demographics between teaser and webisode #1 as this will correspond to the FB analysis above.

First, gender and age. The teaser got a ratio of 15% female and 85% male viewers. That is a much heavier imbalance to the Facebook numbers. Webisode #1 is just a tiny bit deviating, 13% to 87%. Now what strikes me is the huge difference in age. While webisode #1 shows a similar sectioning in age as the FB fans, with 25-34 year olds as major group, the teaser is surprisingly even all the way through from 25 to 64 year olds.

Now, of course, this numbers can change but it basically tells me that there are two different viewer groups active. This definitely is a result of changing my promotion strategy during the web launch in August. I took the lessons learned from the teaser and targeted the series more specifically to niche web channels. Hence, I would argue “traditional” web press channels, like e.g. CNNgo with a vast regular reader base give you a great broad platform but barely real interactions nor sustainable push within other project channels. This is simply because the trans-media aspects are not utilized by their older demographics. This might also apply to print media, but I did not test them on this particular project. However, I must say the CNNgo article brought other advantages and talents to the project. I met a number of new people who then started contributing to Squattertown and are still working with me on new projects.

When we now look at the world maps to see which country got the most views it is clear that all the aimed markets have been reacting to the project as planned. Primary targets were US, Germany, Hong Kong, Austria and Portugal. That Asia got a quite even number of views is surprising. But you have to consider one mistake within the YouTube stats: China as a whole should not be marked. There have been no Chinese views. Also, I can’t say much on the views among the various US states. California seems a steady hot spot, but that’s rather usual I would say.


In the next post I will analyze: VIMEO , GOOGLE+, EGGUP and our biggest mistake!


The following are thoughts, memories and background information on the shooting period of SQUATTERTOWN. I try to structure them by using each shooting day as one blog entry, but the stories might deviate a little. So when you think the text makes no sense please feel free to comment/contact me. This production log is another piece to our large SQUATTERTOWN universe and I hope you like it. But be aware, the text holds several spoilers!

Day 7

The seventh and last day of principal photography was more than a week after our Tai Kok Tsui adventure and a rather spontaneous set-up. The only scene still open from the script was the Mahjong game sequence in webisode #2. Originally planned to take place at the Tai Kok Tsui location too, I decided to settle for a less risky choice due to the amount of cast, crew and props involved. Also the sequence turned out being a bit more complex than expected and needed some more preparation on my side. Hence, the final location choice fell on a small fisher village named Lei Yue Mun. Almost matching the squatter roof landscape perfectly this village offered a variety of different spots suitable for this project. However, none was really on a rooftop.

[natural lighting test on designated location – day before shoot]

A couple of days after the previous shooting I went over to the village to check out which spot would be best. I knew the village from previous projects. We shot our first Dim Sum Western attempt Due Parole, Tre Bugie (http://vimeo.com/7887640) here, along the village’s coastline back in fall 2009. Now, however, I was looking for something more hidden, something that can easily cover as rooftop setting. The day before the actual shooting I visited the village again, together with my DoP Diogo and ShelB, one of our production assistants. We narrowed down the spots I had chooses before and finally settled for a small canopied intersection of  two back alleys within the village’s backside. Residents used it as a place to dry their laundry and it actually seemed to be the front yard of a house entrance that had been blocked for years. We tried to get permission from the residents, but couldn’t find the man to whom this place belonged to. In hope he would approve we left a note.

[location at back alley intersection]

On  arrival at the next day, the man was at home and first did not wanted to talk to us. He actually did not care much about the whole situation. We brought some fruits as token of our appreciation, but he did not care about that either. Eventually he came out, grabbed his drying laundry, cleared the place and left. We took that as approval. To set up the scenery we found a bunch of old, rotten stools and borrowed a small table from another neighbor.

[set set-up & choreographing the “attack” scene with actors]

One thing we learned from previous projects was that the actor’s faces, costumes and props had to look not only dirty but also used and worn out. Our special squatter make-up this time was charcoal and  incense ashes. Every actor and every piece of clothes got a short rub-over with it. As we wrapped this short 4hrs shooting day principal photography was completed. Now post-production began from end of January 2011 till end of May 2011.  Several minor shoots for empty shots, pick-ups and ground material for the visual effects works all over the city took place during that time, but none with actors anymore.

Along this post-production period sound was probably the most important aspect, simply due to the fact that we had little on-set sound recording. Since we were shooting with DSRL cameras external sound recording was necessary. However, the locations usually had all kinds of sounds that were not suitable for the story. Sounds like car noises, construction sites, people talking on the streets and so on. All those would destroy the post-apocalyptic atmosphere of what was intended. So we recorded only the dialogs directly on the set, trying to get them as clean as possible. All other sounds were specifically mixed, separately recorded and tailored during the 7-8 weeks of sound editing. Between April and May we did a series of on-location foley recordings. For the most important of these sessions we had to return to this very village.

[narrow back alleys of Lei Yue Mun]

During the editing process I realized none of the sound libraries we had access to was holding sufficient sounds on footsteps. I needed gritty squatter footsteps and most material I had at hand was simply too conventional. Hence, I ended up reenacting all footstep sounds of SQUATTERTOWN. Mark, our sound recordist was following me with a mic through the back alleys of Li Yue Mun for about 3hrs. Biggest problem we faced that day were cicadas. Back in January there were none whatsoever. Now in May their sounds were penetrating, all over. Additionally, barking dogs, crying babies and taking residents gave us a hard time getting clean footstep sounds. But why were we recording at this troubled place. Simply because it was a place that matched the conditions of rooftop floors. The grounds and stairs were of concrete with gravel and occasional plants. The alleys had perfect reverbs and there were no car noises due to the remote location of the village. Especially the latter one is hard to find within Hong Kong.

The following are thoughts, memories and background information on the shooting period of SQUATTERTOWN. I try to structure them by using each shooting day as one blog entry, but the stories might deviate a little. So when you think the text makes no sense please feel free to comment/contact me. This production log is another piece to our large SQUATTERTOWN universe and I hope you like it. But be aware, the text holds several spoilers!

Day 6

The sixth day was a tough one and we had quite some trouble during pre-production and even location scouting. Our designated shooting location was an old housing block in Tai Kok Tsui. One of those houses who are about to be removed due to the prestigious and scandal shaken high speed train railway between Hong Kong and Guangzhou.

However, the block itself gained quite some popularity as one of the landmark squatter roof villages within Hong Kong. The book Portraits From Above by Rufina Wu and Stefan Canham (http://peperoni-books.de/portraits_from_above_en.html) as well as the short documentary Once Upon A Rooftop by Sybil Wendler (http://www.youtube.com/watch?v=xY7LeNOhCHk) both visited this very location when the residential community still was intact. The time we started our project only about three families were still living in this block. Everyone else had moved out already. We had a hard time actually locating the exact spot and were trying to get photos from a higher angle on top of surrounding buildings. On our first visit it was not possible to get access to the building as all doors were closed, which is very unusual for this kind of building. Usually there were no problems gaining access to other houses next door.

[Tai Kok Tsui roof location]

But we were lucky. Our location assistant Amy went to the location a couple of days later and met one of the remaining residents. He told us that he is not constantly living in this house anymore, it has become too dangerous. We set a meeting date with him so he could give us access to the house. The morning we arrived he was not available however and seemed not to be in the house at all. With slight frustration we walked around the block and tried again every door. By accident one of the doors was unlocked and we stepped inside. On the way up we encountered two floors with completely openly accessible apartments. All apartments were empty and abandoned but most had still furniture, clothes, CDs and loads of other things inside. It looked like residents had to flee their apartments in a rush and left everything behind they couldn’t carry away. The scenery was quite scary, it felt like a war zone.

[inside abandoned apartment]

[a calendar on the wall indicates when apartment was abandoned]

Arrived on the roof we faced the same situation. Everything was abandoned, apartment walls ripped apart, toilets shattered. During my research I saw a number of photos from this squatter roof. It looked like a small village with an intact infrastructure and community. What we saw now was rather the leftover of something that was swept away by a typhoon. To our surprise two families were still living on this very roof. One told us about incidents of accidental fires in staircases and how burglaries have increased in the past months. Of course none of this can be traced back to the property developer who wants the residents to sell their apartments asap. However the huge red propaganda banners outside on the opposite building’s facades indicate their intentions clearly and belong to the company’s public campaign to “drive residents out”. Christopher DeWolf just recently wrote a very good article on this issue: http://www.urbanphoto.net/blog/2010/10/20/say-goodbye-to-old-hong-kong/

[red banner of property developer across the street]

As depressing and dangerous this location was, it was most perfect for SQUATTERTOWN. It was simply the essence of what I had envisioned and there was no way around, we had to shoot here. But I had to be careful and choose wisely what kind of scene and for how long. After all, at the time, we still had to deal with the patrol guards. Applying the same strategy as in Kwun Tong, a precise timetable was set. Everything went well till the Friday, six days before our scheduled shoot. We got the news that a big fire had stricken the very building. A metal workshop located on the ground floor had caught fire and the blaze was so strong that it jumped up onto the roof. Most shocking aspect, one man who lived inside the workshop died during this incident. (Some links to news articles from the incident can be found at the end of this post)

[before/after picture]

After long discussions with the team I decided to visit the location again to see what the actual situation was like. So close to the shooting dates there was no way we could find an alternative location. After all I planned three major scenes in and on this building. We might could have managed finding another roof but definitely not another house with abandoned and free accessible apartments. I expected we encounter investigating police and more security guards now. However, as we arrived the building was even more empty than before. The property developer withdrew all patrol guards one day before the fire incident and police already wrapped investigation after 24hrs, labeling it as accident. Half the roof was now tinted black and the rest looked even more devastated than before as the fire men had pulled out every single wall. It was also obvious that the remaining residents still don’t want to leave and adapted to the situation by taking measures into their own hands. All over the roof we could find big buckets filled with water, a simple safety precaution.

[water buckets]

The risk was still high, even though we seemed to have no disturbance at all for our shoot. After even more discussions and a bunch of grey hair on my head we settled for giving it a shot, trying to shoot as much as we could with an even more reduced crew. Only remaining issue now was how we can enter the building on shooting day morning. We made a deal with the resident who let us enter before, but since he was not always available and reliable it was still a gamble. Also we were told not to wait with the entire team in from of the house, or even enter in large numbers. Word was, neighbors across the street were watching the house and reporting every suspicious situation to the police.

[team during location scouting]

So I arrived at 6:45am at the location with some of the equipment and two crew members. We waited till 7am when our contact came out of the building, on his way to work. The plan worked perfectly. Once we were in others could follow in single file. First, our cameraman Diogo set up some LED light panels inside the apartments to start shooting with the “ghost” shots that you can see in webisode #2. Our second cameraman that day was Mark, who was straight going up to the roof doing some empty and atmosphere shots. As the actors arrived we could finish all planned scenes smoothly and without problems. As we climbed the fragile, decayed tin roofs, getting the last shots of sunbathing cats, the atmosphere on this roof was extremely peaceful and quiet. I enjoyed being here, but time was running out and we had to wrap, moving equipment and team over to Kwun Tong again.

[shooting in roof courtyard ]

After lunch we set up camp at a new roof location that was located at the far end of the housing block we shot at two days before. It was an idyllic corner residence with four shanty huts, a small garden like area and a courtyard in the middle. The residence belonged to a family who lived here since the 1960s and was into the business of sub-renting apartment space. Christoper DeWolfe also wrote an article about this family here: http://www.urbanphoto.net/blog/2010/04/23/hong-kong-rooftops-condemned/. This setting was our “Church” scene that occupies most of webisode #3. As we came to this location the first time during our scouting period three members of the family where still living here. The time we came for the shooting three months later only one remained. When we revisited the residence two months after the shoot everything was evicted and cleared out. The family had left entirely.

[before/after images – left: residence in October 2010 / right: residence in March 2011]

In previous posts I was explaining the situation we had to deal with concerning location shooting permits and how we went in circles from gov authority to property developer to residents to gov authority and so on. While we shot on this roof that afternoon, a team from the very authority visited the remaining resident. They were doing their weekly check-in with the remaining occupants, persuading them to leave. The authority team saw us when they arrived, a bit of an awkward situation. Then they moved inside one of the huts with the resident. We could hear from outside how a heated discussion broke loose.


Link collection on Tai Kok Tsui:

1) on the fire incident:

2) documentary video from roof:

The following are thoughts, memories and background information on the shooting period of SQUATTERTOWN. I try to structure them by using each shooting day as one blog entry, but the stories might deviate a little. So when you think the text makes no sense please feel free to comment/contact me. This production log is another piece to our large SQUATTERTOWN universe and I hope you like it. But be aware, the text holds several spoilers!

Day 5

The second day of principal photography started early again. There was only sufficient daylight from 8:30am to 5pm and we had 3 different locations to cover that day. First up, an industrial building roof in To Kwa Wan. I was told one of the INFERNAL AFFAIRS films had been shot on that roof too, so it seems quite prominent. The building is actually not very high, compared to the other locations but got a stunning view on the old Kai Tak Airport area as well as the skyline of East Kowloon. However, as great as this setting looks in the camera it created particular light changes difficulties for the cameramen. We had to balance everything from a sunrise to a foggy morning. Hence, I was trying to shoot all scenes as chronological as possible. This roof also got another stunning visual aspect, a huge advertising billboard standing right in the middle. Somehow the setting evoked memories of the legendary billboard fight showdown in HIGHLANDER.

Since we were shooting the showdown scene of webisode #4 a great deal of practical blood effects as well as a jumping/running stunts had to be managed in time. Ken Law, the actor of the BAD is a professional martial artist and stunt man. He created a complex stunt scene of the jump on the spot as what I had in mind turn out looking boring on camera. The scene was the first confrontation of the GOOD and the BAD. The GOOD was supposed to advance with his knife, trying to stab the BAD. The BAD dodges and jumps over his knife. Now this sounds simple and I thought it would be, but it was not. We ended up shooting Ken jumping from a 2m high position of the giant billboard onto the hard concrete floor with a roll. I had loads of concerns at that moment, fearing Ken would injure himself. This whole situation felt so Hong Kong film like: tough stunts by the actor himself, no safety, no double, no special effect. Eventually Ken jumped four times and the footage looked very good. It actually reminded me of the flying swordsman jumps in the old Shaw Brother films.

Next scene before lunch was the back alley setting, which was about one block from our morning location. Here, the script also demanded for a short fighting scene. A scene that again sounded simple but required specific choreography. Once more, Ken came up with a quick solution that would give good physical action as well as suitable angles for the cameras to capture it. However, this back alley location was on street level and we were not shooting in secret on some roof anymore. After all we were about to shoot a scene were two guys try to stab, kick and hit each other. So that draws certain attention and I knew that there were a couple of policemen patrolling the street leading to this alley. Hence, we set up a lookout post with one from the team who would signal us whenever someone comes along.

Everything turned out well without any disturbance, only we overran our schedule, for the first time. The alley scenes turned out more time consuming than I expected. After a quick lunch in one of the nearby restaurants we drove to the last location of the day, the 3 housing blocks in Shek Kip Mei we went before on Day 1 back in December 2010. This day was an extremely windy one, giving us a hard time with ever changing light situations. We had times of full bright sun light as well as dark cloudy spots. This situation is most apparent during the GARDENER scene in webisode #4. Due to logistic and personal schedules of the actors we could not shoot chronological in Shek Kip Mei. Hence you will clearly see a big difference from shot to shot within this scene.

As we started setting up the GARDENER scene that involved the longest dialog scene within all four webisodes the two owners of the roof garden we wanted to shoot at showed up. First we feared they shoo us away and we needed to improvise but it turned out they were extremely friendly and excited about our project. While we went on shooting they even brought up from their apartment a plate with fruits and nuts, just for us. It was amazing. As we wrapped this shooting day it was close to 5pm. Another resident came up on the roof and brought his Husky dog out to play. Some crew member fell in love with it 😉