We need it yesterday. We always need it yesterday. A common theme in post production is that high-end creative must be produced on a tight deadline. Few things are more frustrating than waiting on an After Effects render while the clock is ticking loudly in the background. Like many post-production facilities, DotLot generally has a limited amount of time to deliver beautiful videos augmented with amazing graphics. Even though our computers are fairly powerful, we use of dozens of video layers and plugins at once, which taxes our poor little 8-core processors to the edge of their capabilities. By putting our heads together with our amazing IT tech Aziz, we were able to construct a solid render farm at a fraction of the cost of most high-end professional solutions.

For the uninitiated, a render farm is a group of networked computers that help to speed up render time by sharing the workload. For instance, a 30-second commercial has 900 frames. If each frame takes 2 minutes to render, then a single computer would require 1800 minutes, or 30 hours to complete the job. With multiple computers working together, each one can tackle a single frame at a time and then take the next frame in queue when it is done. If you have a five-node farm, that very same render would be completed in one-fifth of the time, or just 6 hours. In a world of looming deadlines, this makes a huge difference.

Our render farm journey started at Ikea. Sort of. A while back, Aziz sent me a link to a blog post about a guy who had built a multi-core Linux cluster using DIY parts and an Ikea Helmer cabinet. (To be fair, Obscured Clarity actually credits this guy for coming up with the original idea.) Prior to seeing this post, we had assembled a five-node render farm using old workstations that were left behind when systems were upgraded. It was a decent solution, but having five big Dell towers in a small edit bay made a lot of noise and heat and wasn’t very practical. This Helmer idea seemed to be a great alternative - we could build cheap, bare-bones systems; connect to them through our network; and they would take up less space than a single tower. Good deal!

Our plan was to split up the work - Aziz would set up the computers while I modded the case. His first task was to order all the parts for the computers. Since this project was somewhat experimental, we decided to start small with three render nodes, figuring we would add more later. Here was our buy list from NewEgg.

We originally ordered 3.5″ drives - DON’T DO THIS!  The Helmer is only big enough to fit 2.5″ laptop drives. We returned the 3.5″ drives and ordered up some spiffy Western Digital Scorpio Black 160GB 2.5″ hard drives.

With all the parts in hand, Aziz set out to build the master drive, from which we would clone all the others. He installed Windows XP and the After Effects render module and then added all of our plugins and fonts. (We primarily use Red Giant’s Trapcode Suite and Knoll Light Factory.)

While he worked on that, my first task was to head out to Ikea to buy the Helmer. It’s available in a few colors (silver, white and red) but since we were building a render farm I decided it would be appropriate to go with the barn-red version.

Like most Ikea stuff, the basic build was pretty easy. The hardest part was cutting holes in the back of each drawer for the PSU and cooling fan.

Word to the wise - if you’re using a Dremel, be sure to use an appropriate routing bit for metal. I burned through a few wood routing bits before I finally realized that there was a bit made specifically for metal routing.

Once the rear panels are cut and the screw holes are drilled, you can mount your PSU and fan.

All that’s left to do after that is to slit the label holder on the front panel of the drawer so that you can fold it back to allow for a little extra airflow over the CPU.

The next step was to build the three systems into their drawers. Aziz used Clonezilla to make identical copies of the master drive.

In order to avoid issues with the bottom of the motherboards touching the metal surface of the shelf, we cut insulating sheets out of the tops of the motherboard boxes. High tech, I know!

With a handful of zip ties we were able to get the cables under control without having to cut any of the extra ones away.

We added a switch so that we could connect all of the systems to our network.

And our Pièce de résistance is our patent-pending screwdriver power switch. We had planned to order some lighted power switches that we were going to add to the front of each drawer, but there was a problem with the order and we needed to build the farm, so we went super low-tech and decided to forego any physical switch at all. We literally said “screw it” and picked up a nearby phillips head driver instead.

In this shot you can see the CPU fan is finally up and running…

We assigned each of the three render nodes a unique, static IP, and using the Windows Remote Desktop Connection feature, we were able to connect to each one from our primary edit system.

It worked like a charm!

Now all of our renders are three times as fast as they used to be. In the near future, we plan to expand our farm to a full six systems. Eventually, we might even add additional Helmers to make our rendering lives even better.

–Mike

We are a studio. We have gear. Gear breaks.

I’m all for sending pro gear back to the manufacturer to be serviced, but in this economy, I also like to save where I can. When one of the four ballast boards on my Kino Ve 4Bank Ballast took a permanent nap, I decided to fix it myself.

With a quick search on the Kino site, I was able to identify the part number I needed. It took a few days to get the board (Kino made me order it from Samy’s) but once it came in, it only took about 5 minutes to swap.

I took some pictures…

Here’s the main ballast unit…

That had a dead ballast board, as evidenced by the non-op tube…

Start by removing 10 screws from the case cover. 3 on each side and 4 on top.

Here’s what is looks like inside. You can see in this picture (and the next) that I have already removed the white connectors on either end and the smaller red connector at one side on the second board up from the bottom.

To remove the board, just undo the three hex nuts. Two are on one side of the board (I’m pointing to one of them) and one is one the opposite end. DO NOT remove the small phillips head screw - it’s part of the board.

That’s all. Now, back to making movies…

- Mike

Recently I’ve been exploring and learning about Twitter. What is this thing? How can Twitter benefit me? Like most people, I had signed up at one point but hadn’t done anything with it after that.  What’s its purpose and why should I spend any more of my valuable time on it when I already have email, a smart cell phone and FaceBook?

So I started by learning a little more about it in plain English. Thanks to CommonCraft,  I got a simple and clear understanding of Twitter’s function and uses.

I love the action hands with the cutout pictures. It makes the message so clear as it’s not distracted with typical heavy graphics and digital sprinkles of eye candy. From this video I started to understand its basic uses. I began to follow people of interest from a variety of areas: video production, social media, friends & family (of course), public servants, and some celebrities, why not?

I’ve come to see that little messages lead to interesting and hard to find information. For instance, @mashable Reading: “10 Golden Rules of Social Media” - http://bit.ly/4by3vD. Or other entertaining pages: “Three Wolf Moon!”, fun Amazon reviews : http://bit.ly/s8gVL.

So now the question is, how can I take advantage of this tool for DotLot and myself? One way I’ve begun is to search the chatter on Twitter for what other production companies are talking about or sharing.  For example: @nilsbreckoff: Cranking up for another fine day of video production, or @Xcelus: We are putting two new marketing scripts into video production. A JR Golf video and ecommerce video.

Lastly I’ve found little nuggets of information that is very useful that has been right under my nose. For instance check out the Speech Search from Adobe.

Moving forward I can see myself using Twitter to research companies we’re doing business with or are targeting. This little bit of information will give me more insight to the target audience, the users and the advocates, and see what messaging is getting a good response and what’s not working. Our job as media professionals is to help our clients connect with their audience in the best way possible.

I’ve leave you with this last bit I found on Twitter from @carloguillot. I think all media production people can relate to this video:

You can find me @NBHaus.

We recently received a project to capture past webinars and create FLVs so the company could play them on the web. Our first thought was to just bring the presentation into an NLE system for editing and encoding. However, in looking at the data delivered, we found these webinars had all sorts of programming that made it pretty impossible for NLE’s to capture straight out of the box

Ideas were shared on the best ways to gather the data into a format that was malleable for encoding.  One thought was to have one of the in-house programmers sort through the code and export the audio and images into a video codec we could use. However in looking through the code our programmer gave me the angry look that translated a “No Go”.

Our second thought was to do a screen capture of the material. Well, we’ve used a couple different screen recording software packages before. All of them have been ok in their own ways.  We decided to re-try an old go-to software everyone has used in one project or another, Camtasia Studio from TechSmith. This software has done some growing up in the past several years. Now you can screen record you presentation and export it to many more presets than I remember; blog, CD, DVD-ready, iPhone, iPod, Web and YouTube, or if you prefer formats: WMV, MOV, AVI, M4V, MP3, RM, GIF (animated file). So yeah … there is quite a lot of versatility with this program.

The recording of the screen is pretty straight forward with a simple interface. Hey, even an intern could do it, right Rob? Moving on … you set your window to record and hit the red button. Nothing’s going to blow up but for the most part that is all you need to do. Do check your audio levels however.

Once the recording is finished, you then get the option to edit the piece. Again with a simple, iMovie-ish user interface, cutting the piece is just a shortcut “S” away. Trim, save your project and “Produce video as” – ah … this part is up to you. Your end delivery is moments away.

The process is very time consuming and there needs to be a lot attention to the details. One miss steps could cause you time and the potential to recapture the entire project. Watch your settings and capture area very closely. Remember we’re all professionals.

Overall, I’ve come to appreciate Camtasia for what it is – a really inexpensive screen recording software that can get the job done. Now, there are many other functions I didn’t go into that this particular software can do (i.e. 3D tilt effect) so go and revisit this old friend and maybe it can help you out again someday.

What do you use to screen record?

I’m not an engineer.

I don’t have a degree in electrical engineering, or any other form of engineering. I’ve never sat on an ISO, IEEE or any other standards panel. I’m not a computer genius. Everything I know about cameras, color space and NLEs I either picked up in film school or learned on the job along the way. I do, however, work with cameras and post-production workflows on a daily basis. I do a lot of shooting, editing and keying and I’m always trying to grasp the concept of different color spaces across the SD and HD spectrum. 

If you’re reading this post, then that probably closely describes you, too.

What is 4:2:0 HDV color space? How does 4:2:0 color space work? 

As I have done research over the years, the one thing I have learned is that color space is a deep, complicated topic. To make things worse, there doesn’t seem to be a single, easy-to-understand post on the web that breaks the concepts of color space down into layman’s terms. This post won’t be all-encompassing, but I do aim to share with you the concept of 4:2:0 color space in clear english to the best of my ability.

I highly encourage conversation on this topic. If you read this post and think I’ve gotten something wrong, please post a comment. If you can help to further clarify a point, please join in. As I continue my research and learn more about color space I’ll write about it in future posts.

Okay… here are the oversimplified basics of color space…

4:X:X refers to the number of pixels being sampled for each component of a three color image. “4″ is the maximum number and “1″ is technically the minimum number. 

For digital images, the whole ratio is generally referred to as YCbCr, where “Y” represents luminance, “Cb” represents the blue channel (as a difference between that and the luminance channel) and “Cr” represents the red channel (also as a difference from luminance). In the analog world, “YPbPr” means basically the same thing.

The first number refers to the number of luminance pixels that are sampled for each block of 4 pixels. In almost all cases, the first number will be 4. The second number refers to the number of blue pixels that are sampled for each block of 4 pixels. The third number refers to the number of red pixels that are sampled for each block of 4 pixels. Thus, 4:4:4 color space is the best possible option, since every single pixel is sampled for luminance, blue and red information.

Enter digital video and HDV. In order to save bandwidth and bitrate during DV compression, most manufacturers made their standard def DV cameras work on a 4:1:1 system. This meant that for each block of 4 pixels, all four were sampled for luminance, 1 was sampled for blue and 1was sampled for red. This means that DV doesn’t take up much bandwidth, but it does lack a significant amount of color information, which means it’s not ideal for pulling keys.

With the 4:2:0 color space of HDV the color information issue gets just a bit better. HDV works by sampling every luminance pixel as well as 2 of every 4 blue and red pixels. But what’s that crazy “0″ all about? Apparently, when the third number is 0, it means that every other line is skipped. This means that we now have 2 blue and 2 red pixels of information out of every 4, but only on alternate scan lines. This effectively gives us 4:2:2 color space, but at half the true resolution. This is better than 4:1:1, but still not great, since it leads to issues when trying to key out fine edges, moving objects, fingers, etc.

I hope that this helps to clear up some of your basic questions about 4:2:0 color space. I’ll be updating this post and adding new ones as I learn more.

– Mike

Google has now entered into the browser world with its deployment of Chrome. At first it takes a little getting used to as the toolbars are now incorporated into the whole interface. But, what I think Google does best is how they use videos to walk you through how the browser works and what to look forward to. This form of explanation through illustration is ideal for all us visual types. Check it out: How To’s

There are several interesting features you might want to take a look at. For instance, dynamic tabs allow you to rearrange your tab’s order, move it to create its own window, and drop it back in to consolidate windows. It reminds me of Adobe’s new interface where you can create new windows for bins or tools as well as place them in the same window for tab navigation. It makes a cleaner interface and over crowding with toolbars and other nonsense we all download.

Another interesting feature is the “Incognito Mode”. It allows you to browser the web without storing a record on your computer. Now there’s an interesting use of browser technology. Now I can check the scores with peace of mind.

Google set out to create a browser to run on today’s complex web application better. Being a participant in this digital media space, one has to wonder how this new browser is going to handle all the other pitfalls of creating web pages with video, Flash, Flex, Air, Papervision, etc.  Overall, I’m very excited of this new entrant and I’ll be testing its limits to see if it retains my default browser award. Has anyone been using Chrome and want to share their experiences with us?

For a recent project, we were charged with producing a video that was going to be displayed on a plasma screen at a tradeshow booth and later in the company’s lobby. We knew we needed to produce the video presentation for 16:9, but we had many rounds of debate over which pixel aspect ratio to use and what our final delivery format needed to be. In order to find the answer we needed to take a step back and discuss what the difference is between a square and rectangular pixel in the first place.

Aspect ratio specifies the ratio of the width of a picture or pixel to its height. Photographic pixels are square with a pixel ratio of 1:1 which means that the width and height are equal, or square. NTSC TV uses a pixel aspect ratio of 0.9:1, which means that the height of each pixel is greater than the width, or rectangular. Widescreen TV (16:9) uses a pixel aspect ratio of 1.2:1, indicating the width is greater than its height, also rectangular. Luckily, HDTV at vertical resolutions of 720 and 1080 has an aspect ratio of 16:9 and uses square pixels, as standardized by the ATSC, so we chose a square pixel format for our project. Unfortunately, the challenges of delivering to an unknown monitor didn’t end there.

Plasma screens are made to display 16:9 formatted presentations but can also display 4:3. If you do decide to go with a 4:3 format you need to be aware that the viewer may not see a true 4:3 image. While there is usually a 4:3 pillarbox format option available on plasma displays, several manufacturer’s engineers have figured out their own unique ways to display a 4:3 on a 16:9 screen. In some cases, they do this by preserving the image in the center and stretching the edges towards the sides of the screen. In our case we created our presentation at 16:9 to allow the presentation to fill the screen. But the quest for the best display of a production didn’t end there, either.

Take a closer look at Plasma display resolutions and you’ll see that all screens are not created equal. Examining 7 plasma manufactures and their screens, we discovered that very few have a display of 1920 x 1080. In fact, of the 16 models we reviewed, only 4 had a native 1920×1080 display. This means that you really need to find out which model and manufacturer your client is using before you start production. If you don’t know, you might want to create your video to allow for the best presentation on the majority of plasma displays. We try to follow the rule of always working in the best quality (1080) and scaling down when necessary to meet the demands of the presentation, of course it’s always better if you can avoid resizing.

One last thought to consider is how to actually get your well produced production to the plasma display. One way would be to have the plasma connected to a hard drive where you can call up your presentation – in native HD, if possible. Another option would be to author a Blu-Ray disk, but this only works if your client has or plans to rent a Blu-Ray player. Finally, you could always author to DVD ,which is the path we had to take for this project since the client was on the east coast and they weren’t able to provide us enough details about their display setup. We exported our final comp as 16:9 with square pixels and brought it into our DVD authoring tool. From there we exported using DVD standards for a widescreen output. Did we get the greatest quality output? Maybe not the best possible, but by using the right settings throughout our post production workflow we produced the best possible DVD and it was certainly the most appropriate output for this project.

What’s your experience in working with Plasma screens? Leave a comment here with some of your thoughts and experiences on the subject.

Using User Presets as an Efficient Workflow Tool:

One of the most common buzzwords in the industry is “workflow”. You will see this term used more often than any other whenever people are discussing post-production. What they are really talking about is efficiency. The faster and more effectively one is able to perform tasks, the more they can use their time to improve the final output through creativity and polishing. In other words, the less time wasted on menial details, the more time can be put towards creating better projects.

In that vein, I would like to share a useful tip I have been using in Adobe After Effects CS3 and Adobe Premiere that has saved more time than I can imagine. Most AE users are familiar with effect presets and the ability to create user presets. What I am going to demonstrate is very fundamental, but is scaleable and applicable to numerous uses. I am going to demonstrate how to use user presets in conjunction with an organizational process to streamline tasks such as keying.

For this process to work best it actually starts before the camera rolls. The first step is to have a shotlist, with shots labeled in a standard numbering and lettering format. Obviously, you may not have control over this, but if you do, make sure that you stick to this list. During the shoot, make sure to slate every shot clearly and correctly. Once the footage is brought into Premiere, label it accordingly. This is common practice, and should be a part of any post-production process. If there wasn’t a shotlist, create one at this point, making sure to use a numeric system (see below).

EDIT

You’ll have a timeline that looks something like this:

Notice that all the clips were labeled accurately and the same shot was used multiple times while cutting together this dialogue scene between two characters. Fine-tune your edit until you are completely satisfied and ready to move onto the VFX/Motions Graphix.

Import the premiere project into AE. When this is done, all the data from Premiere is brought in, including the sequences, clips, and naming conventions. You might have to do a little organization, depending on how your tracks were setup in Premiere, and how you want your AE project layed out. In a future article I will discuss some practices and methods for streamling AE projects. But for now, let’s assume we’re neat and tidy and ready to key. Start at the beginning and move forward. Everyone has their own keying process, so do what works best for you. Once again, in the future I will discuss my methods a little more in depth. However, almost all techniques require the use of multiple effects.

Once you have your key set up, highlight all your effects and select the make new preset icon in the bottom right corner of Effects and Presets box. It will open a dialogue box to save the preset. Here is where the system kicks into high gear. At this point, what you are doing is creating a database of effects (in this case, keys). Since the same shot is used multiple times in this project, the effect will need to be used multiple times, with little to no changes. One might ask, “How does one keep track of all these effects?” It’s easy. Your naming convention has already been set up; it’s your shotlist. As long as you labeled your clips correctly in Premiere, each instance of a shot will be its labeled accordingly as well.

When you go to save, create a folder structure in your preset library, preferably numeric and labels. Create a folder for keys, and save the preset in the folder labeled by the following three things: Folder number (100 in my case), shot, and description. By using all three, it will be much easier to search for the desired preset quickly by simply typing in the keywords or a shot label in the search box.

As you move through your project, you will be able to quickly locate the correct key and efficiently. By using the shot label to search, a few keystrokes will bring up the correct key. This is much easier than any ctl C/V method I have ever used previously. The purpose of the numeric labeling is more clearly seen when working with artistic effects that may be used in any number of shots or projects. By creating preset animations/effects, and labeling them by number, you can quickly pull up any effect quickly and effectively, without a lot of time spent searching or recreating. For instance, all lens flares animations can be labeled 51* so by simply typing 51 in the search box, all presets will be at your fingertips instantly.

(footage courtesy of iTwinkle.tv)

How to Identify and Solve Web Video Production Problems Before They Occur

By Mike Kirsch, Director of DotLot, The Buddy Group

An important trait of any successful video producer is the ability to troubleshoot problems quickly and effectively. To the end client, efficiency and quality of final delivery is just as important as killer front-end creative. For the online space, “quality of delivery” expands to include visual quality of an online video, speed of download or buffer time and smoothness of playback. When problems arise, quick solutions win the day, but in many cases problems can be avoided altogether if they are anticipated before they occur. Recognizing certain “Red Flags” of web video production is the first step towards making sure they never cause problems at all.

Picking the right combination of settings to maximize the playability of a video means juggling a slew of settings while prepping that clip for final export. It’s no accident that the number of visitors to a site is referred to as “traffic”. The internet is not unlike a freeway. The speed at which a given piece of video will play is limited by the bandwidth of a server (the number of lanes) and the total number of current users (the number of cars on the road). There are other factors, too. “Bitrate” refers to the amount of data encoded in a video file for each second of runtime. Higher bitrates mean better image quality, but the higher the bitrate, the more data there is that must stream before the video can play. In the freeway analogy, this might correlate to the steepness of a hill. As hills get steeper (higher bitrate), it takes more power (bandwidth) to climb at a given speed. Without sufficient bandwidth, you’ll either need to reduce the bitrate of the video or risk sliding backwards down the hill, causing an accident, and wrecking your car (client relationship). Believe it or not there are even more factors than those mentioned above! Resolution, length, frame rate and even visual content all factor in, but the freeway analogy gets more complicated from here, so let’s just talk in terms of general considerations.

When planning a project that is going to involve video, the first thing that needs to be determined is the intended audience. Who is going to watch it? What type of connection are they expected to have? How many users will potentially be accessing the site at one time? Understanding the intended use and distribution of a website and its videos will help dictate the parameters that should be used to prepare the final files. Someone with experience producing and encoding videos for the web should always be consulted when planning a video project, but the following list can serve as some basic Red Flags to help identify issues that will require further investigation.

• Bitrates over 400kbps
  • Some videos look fine at bitrates of 400 or less. In instances where there are a lot of smaller textual elements or a lot of quick motion, higher bitrates may be necessary in order to achieve acceptable visual quality. Higher bitrate, though, means longer download times and more bandwidth costs.
• Frame Rates above 15fps
  • Video on television typically plays at either 24 or 30 fps (frames per second). For the web, the standard is to cut the native frame rate in half, which reduces file size and bandwidth needs. Tests have shown that higher frame rates place more load on a user’s CPU than higher bitrates, which means that from a user-end perspective, lower frame rates will increase performance dramatically. The drawback is that video looks choppier as frame rate is reduced. In videos with a lot of fast motion, pans or small details, this can lead to strobing and other issues. If a client requests video at 30fps, take note. This could cause a problem later.
• Resolution
  • 320×240 used to be the industry standard for web video. As broadband connections get more common, the capability for producers to go live with larger videos increases. Be aware that larger videos mean longer load times and higher bandwidth needs. Always compare requested video resolution to the other involved factors to determine if a video will work at a given size.
• Running Time
  • A 10 second video at 900kbps running a 30fps may not pose a huge problem. A 2 minute video with those same parameters will likely not play properly on most computers. If a client is planning a site with a series of videos longer than 15-20 seconds, then length becomes an issue that will have ramifications on video quality.
• Screen Content
  • If the creative call for a lot of on-screen textual information, charts, closed captioning or other small details, then the videos will need to be played larger and at higher bitrates in order to be clear and legible.
• Transparency
  • Interactive videos that are intended to be laid over other web content using alpha channels are always a cause for concern. Even though the alpha channel information is invisible, it still draws bandwidth and CPU power, not to mention the bandwidth required by all the stuff we see through the transparent layer.

By Dylan Highsmith, DotLot Editor, The Buddy Group

Ok, so we know how best to optimize content for mobile video, but how does the average person get their content distributed? Currently, when you think of mobile video, one of the first names that comes to mind is Verizon’s V-Cast. V-CAST is a mobile network for Verizon Wireless phones that delivers media to its users for a subscription fee. V-CAST typically streams between 400-700kb/sec and streams can be saved to a phone or memory card.

Initially, when V-CAST was launched, it only offered video clips from major TV networks such as NBC, soon after smaller cable networks started to follow. What is most interesting, is the new addition of the V-CAST channel ‘Net’s Best’. This channel provides content from some of the top video hosting sites on the web such as YouTube, Maxim, Break.com and Revver.

YouTube’s deal with V-CAST is currently a limited-time only deal and the content consists of videos from YouTube’s Most Popular tab from its website, so to get on V-CAST through YouTube, your video would have to appear on the Most Popular tab of YouTubes site. Revver however, has a slightly different approach.

Revver, like YouTube is a video hosting site. What’s unique about Revver is that it shares all ad revenue roughly 50/50 with you – it’s content creators. With V-CAST, as mobile video doesn’t include advertising – content creaters share the subscription fees that Verizon pays for access to the content.

What is unique about Revver’s approach to V-CAST is that their content that appears on V-CAST is not simply taken from their most popular tab, but rather picked out of submissions.

In order to be considered for V-CAST, when uploading video to Revver you must opt to participate in the Revver Mobile Distribution opportunities. From there, fifty new videos are chosen by V-CAST each week. Revver’s website has a few basic guidelines of what kind of videos they are looking for:

Videos should be between 30 seconds and two minutes long.

• No pornography
• No copyrighted music (not even in the background)
• No depictions of crime or drugs
• No advertising or prominent brand logos
• Only videos in English can be accepted.
• Videos shot at a distance won’t be accepted — it needs to be decipherable on a small screen!
• Watch out for text — that screen makes small text really really tiny.
• Please avoid mentioning religious or political views.

One thing of particular note is that Revver and V-CAST don’t seem to like branding. They are looking for media that is purely entertainment. As the content on V-CAST is paid for by users through a subscription fee, this makes sense – and is an extra thing to keep in mind when creating content that you may want to be distributed through mobile media.