Archive for the ‘workflow’ Category

Using Avid Unity Media Network on OS X 10.8+

May 24th, 2015

The last supported version of Mac OS X is 10.7.x with MN Client v5.5.5 (Mac only — the PC version 5.5.4 still works on Win7-64bit) — because 10.7.x is the last OS X version to support 32-bit extensions. You think! With the respective hack I already described for (now deprecated) versions of the ISIS client with OS X 10.9, you can take the equivalent steps to connect your Unity on OS X 10.8.x (haven’t tested with 10.9+, maybe this works, maybe it doesn’t), just pay attention that the Unity kexts and other files have slightly different names.

And another freebie.. even though Avid recommends using ATTO FC cards, this works with the original Mac FC cards (by LSI) too. The only drawback is the “limitation” to a 2Gb/s throughput, but that’s sufficient for most HD projects.

Scope

August 16th, 2014

Every decision we are about to make depends on our scope relevant to it. Which is why you get a variety of answers/solutions to exactly the same question/problem depending on whom you ask. Some answers/solutions encompass others, some are completely different, but none of them are wrong in their own scope.
This is overlooked very often, because it’s easier to assume a mutual implicit understanding and worldview, but that’s rarely the case.
Hence we must define the scope on which a certain decision will be judged, because if there is a divergence in the scopes of decision and evaluation, the oddities of judging any decision as wrong or bad go way up, which leads to wrong conclusions about the competence or even personality of the decision-maker.

How to delight an engineer

July 29th, 2014

If you happen to work with engineers and they’re mostly unhappy with the way they’re being asked about potential projects, or if you’re an engineer in this very situation and your colleagues don’t seem to get it, here’s a simple how-to:

Step 1: Describe what the product or service is like when it’s finished, as precisely as you can, from A to Z. Put in the effort to sketch out the interface (if there is one), who does what in which particular order, who needs to know when something has happened, etc. Imagine how people will use it and what their expectations might be. Try to put yourself in every party’s shoes and walk through the whole thing. Think of legal constraints. Write everything down. (Use flowcharts whenever possible. Engineers love flowcharts.) Yep, that’s a lot, but believe me, it’s worth it.
Step 2: Define process boundaries. What is supposed to happen, what must not happen under any circumstance?
Step 3: Define development constraints. What is the budget? How much time is there to get this done?
Step 4: Have a meeting with the engineer(s) and give them all of what you have worked out so far, and they should be able to tell you what you need to know (can we do this in the quality proposed, considering all boundaries and constraints?) on the spot.

Bonus: How to delight engineers and non-engineers alike
Say Thank You, even if the answer you get is not what you were expecting.

Connecting to Avid ISIS 5000 on OS X 10.9

July 3rd, 2014

Though Avid doesn’t officially support this, it is entirely possible to connect your ISIS5000 < 4.6 using the respective ISIS client to your Mac on OS X 10.9, which currently only supports ISIS7000. Update: As of ISIS v4.6 this article is partially obsolete (see Avid KB).

To make it work, you need to have a Mac on which the ISIS client software that corresponds to your ISIS 5000 is installed. From there, copy the following items to a USB stick or network location, ideally you mimic the folder structure of your Mac HD in order to put everything in its appropriate place:

/Applications/AvidISIS (entire folder)
/Library/Application Support/Avid/AvidUnityISIS
/Library/StartupItems/AvidUnityISIS
/sbin/mount_AvidUnityISIS
(executable; use Cmd-Shift-G to go to folder)
/System/Library/Extensions/AvidUnityISIS.kext

Next, copy the files to your 10.9 machine’s system drive. You will need to enter admin credentials when modifying the /Library, /sbin, and /System folders.
Next, fire up Terminal to modify the permissions to certain files and folders, because those will very likely be “damaged” during the copying process off the original computer. You will need to enter your admin credentials when doing the following:

sudo chmod -R 0755 /System/Library/Extensions/AvidUnityISIS.kext
sudo chown -R root:wheel /System/Library/Extensions/AvidUnityISIS.kext

sudo chmod -R 0755 /sbin/mount_AvidUnityISIS
sudo chown -R root:wheel /sbin/mount_AvidUnityISIS

sudo chmod -R 0755 /Library/StartupItems/AvidUnityISIS
sudo chown -R root:wheel /Library/StartupItems/AvidUnityISIS

This will set the files and folders to the proper permissions. However, I strongly recommend verifying your permissions with Apple’s Disc Utility, because on one machine where I did this there was one admin user who couldn’t mount any of the ISIS workspaces (some error on the /Volumes folder), after repairing permissions everything was fine. And it is vital to perform these steps in this exact order, if you leave them out it won’t work.

Reboot your Mac, then you should be able to launch the ISIS client, and after properly setting up your connections you should be able to mount workspaces as usual.

Please keep in mind that this is a hack; Avid does not provide support for any problem that might occur because of this, neither do I. You’re doing this at your own risk. Make a backup copy before working on critical data or projects.

De-Interlacing roundtrip

August 30th, 2013

It’s quite amazing that hi-quality de-interlacing of video footage is still an issue not easy to be adressed, especially when it comes to encoding for web and mobile. The basic problem is that your footage inevitably loses sharpness, but it depends on the quality of your de-interlacer how much aliasing you will end up with. And just to make sure: for web and mobile (and Windows Media Player..), always de-interlace your video. TVs, in most cases, handle interlaced H.264-encoded MP4 files quite well.

In my recent case I had some interlaced 1080i footage in DNxHD 120 which was being edited in Avid Media Composer 5.5.x and should be encoded to 720p25 and 1080p25 for mobile devices, and of course PC desktop playback. As it turns out, this version does a horrible job when it comes to aliasing, because the algorithm seems to be line-averaging only, no matter if your export from interlaced to progressive or import the interlaced footage to a project with progressive video setting. Though it takes ages, the edges of graphics, the part that is always the hardest to process for the computer but the easiest to spot for the viewer, are always quite jagged. So this was a no-no.

Why would you re-import the “baked” edit anyway? Well, I tried this because Adobe’s Media Encoder CC, which has the benefit of being a blazing fast encoder, also doesn’t offer any user-selectable de-interlacing algorithms. It just does its line averaging thing too, with equally inferior results. No-no.

So I gave Apple’s Compressor (I still use version 3 from Final Cut Studio 2) a shot, well knowing that the “better (motion adaptive)” algorithm yields good results but can be horribly slow (more on de-interlacing with Compressor here). And it was. On my 2011 Mac mini (on steroids) the encoding of a 30 second clip to a 10Mbit/s H.264 took 2 solid hours. With 130 clips in the pipeline, this was no approach worth pursuing.

Hence… putting each tool only to the function it’s best at, a working and quite fast approach with good results looks like this:

  • Export interlaced edit from Avid MC in DNxHD 120 (in my case, using a QuickTime mov container). Works in about half-real-time.
  • Either manually or by folder action, use Compressor only to de-interlace with motion adaption, again to QuickTime with DNxHD 120. This works in real-time speed.
  • Using a watch folder, process the Compressor output with AME CC to get your MP4’s. With parallel coding of two files, this is about 1/10th real time when using a 2-pass setting with a reasonably high bit rate.

Which is why dedicated encoding solutions make sense, if you need high throughput. For occasional use, it’s still good to know how to work this out on your machine.

P.S. Yeah, I know. HandBrake. Let’s just say I have my reasons for not using it a lot. One day, maybe. Then I’ll post an update saying how wrong I was yada yada.

Projects are doomed when…

July 6th, 2013

… there is no desire to assign or take responsibility right from the start.

… there is no desire to make it happen.

… there is no desire to do it better than the last time.

… there is no desire to keep the staff involved as happy as your superior and/or your client.

… as a consequence of the four items above, there is a lack of planning.

… as a consequence thereof, each step of execution becomes an iteration of the whole process.

… the manual or protocol being followed is outdated and/or doesn’t match the requirements of this project.

 

Projects are not doomed when people are (being perceived as) stupid (read: either not smart or not knowledgeable).

Projects are doomed when people don’t care enough.

It’s a game, but different

January 25th, 2013

We all know them, the often motivationally intended comparisons of organisations, working teams, to sports teams. Or armies. But let’s stick with the sports.

The shortcoming with these comparisons: They do not reflect reality. And even if you knew that already (which I assume), I wonder if you’ve ever asked yourself why that is. And the answer that hit me the other day is that you are not only Team Red, you’re also Team Blue, and there are at least two balls in the game.

Just like every person is the biggest obstacle on the way to [success, happiness, self-actualisation …], so is an organisation. Because the dynamics work both ways: Not only do you have to move “the” ball from your side towards the other goal, but at the same time prevent “the” ball from hitting yours. The ball(s) you’re moving forward are your assignments, and the balls coming towards you are (or will become) problems. Not only linked to assignments, some are random, “outside-world factors”, nevertheless you have to deal with them.

It’s not enough to focus on the development of your “offensive game”, you must also -at the same time- prevent or solve problems quickly and effectively or your score will be diminished by the problems that hit you in the back. After all, what’s hitting a goal worth when you get hit twice at the same time?

So the next time you’re going for the team circle, you may want to keep this in mind. You’re both teams, and there’s many balls in the game. All need to be taken care of, otherwise you better not bet on yourself.

Video quicktip roundup

January 6th, 2013
  • If you need to install Final Cut Studio 2 (FCP6) on a Mac running OS X Lion (10.7) or Mountain Lion (10.8), you should either have a Snow Leopard Installer Disc at hand to install Rosetta first (from the “Optional Installs” folder) or you can use Terminal to do it without the Graphic installer (which is the reason why it doesn’t install without Rosetta, it’s still PPC code (not the FCS apps themselves, to be clear), whereas Final Cut Studio 3’s installer is Intel only).
    The magic Terminal command is (one line)
    sudo install -package /Volumes/Final\ Cut \ Studio/Installer/FinalCutStudio.mpkg -target /
    Use your respective source directory if you copied the installer to disk beforehand. Hat tip to Jeremy Johnstone for solving this.
  • External disks which are supposed to be accessible for read/write on both Mac and PC should be formatted to exFAT (available since Mac OS X 10.6.5 and Win XP, for which you need to install an update filed under KB955704). This circumvents the 4GB file barrier of FAT32.
  • GoPro Hero HD footage (no matter what model) is best to be converted with MPEG Streamclip, especially if you need to go to SD formats. It’s way faster than Compressor and can merge all your files into one (open source files sorted by date), plus it takes everything to a “standard” codec (which GoPro’s CineForm Studio doesn’t, they have their own, which is not bad either, but…). Note that on the Mac you have the benefit of merging several clips into one, which doesn’t work that well on a Windows machine because the single clips are not always in the right order. You may circumvent this by (copying and) renaming the Gopro files to the correct sequential order.

Taking on a complex task

November 13th, 2012

In this context, a complex task is defined as one with multiple steps over a long time period, and one where the one(s) in charge don’t necessarily know everything about what’s coming over time. A classical development project, if you will.

The most important action before even starting on the task is admitting that it is a complex one. Also, it is good to admit that you cannot foresee all of the challenges that will come up over time, but that you are 100% determined to pull it off.

Also, you must review the objective of the task to know when it’s completed. For instance, when are you finished “learning to play guitar”? This “goal” is so abstract you’ll never reach it. So you may want to “learn to play guitar as good as Steve Vai”. That’s a bit better, but unless Steve Vai is dead he’ll progress too. So you may end up wanting to “learn to play guitar so good I can play all Steve Vai songs”. See? Now you know when you’ve reached the goal.

Next, you need to find personal support. This means not only do you need a trainer or teacher, but also a sparring partner. The former should be an expert in the field of your project, the latter not necessarily. His or her job is to reel you in each time you’re about to abandon the whole thing. And believe me, sooner or later you will find yourself in a situation where you want to throw the thing out of the window. There is very little certainty you’ve done your best underway, though you may very well have. So it is mandatory to have personal support because self-motivation doesn’t always work. Never underestimate your inner saboteur. He’s very patient, just waiting for the right moment to strike. Cover your back.

Then, start breaking down the whole thing. What parts does it consist of? What are the milestones? And, something that is overlooked quite often, what small parts can you incorporate in your daily routine? In every complex task, there are exercises or other routines which, when complete, will have become your second nature. But they only will when you include them as early in the the whole process as you can.

Also, allocate specific times when you fully commit to work on this particular task. No distractions. Creating this kind of external pressure often has more long-term impact than devoting a large chunk of time every now and then. This is why it’s important to have your sparring partner, who’s to make sure you stick to your regular appointments with the task.

Regular review is also very important. Every time you reach a milestone, take a look back and check on how it’s been going so far. Keep track of all your moves, where you were wrong and how you came back on course again. Don’t forget to put little rewards at every milestone, because you can’t be certain there will be rewards from others. Quite the contrary. Sometimes, the better you do, the more external resistance you will find.

Brace yourself. Persist. Good luck.

Room for failure

November 10th, 2012

To make my point, I will distinguish two kinds of failure here: the one that happens in defined processes, and the one that happens in the course of exploration.

It’s obvious that in defined processes there should be no room for failure. If it still occurs, there’s either something wrong with the process design, or people have not been correctly instructed. Of course people make mistakes, but it’s again obligatory to design a process in a way that mistakes are being discovered and fixed underway.

On the other hand, when you’re setting out to do something you haven’t done before, there should be a lot of room for “failure”, beause in this case failure is to be defined different. It’s silly to expect that you will have figured out the new thing straight away and do everything right. Even if so, you’re then being confronted with the problem of figuring out possible mistakes and their remedying when you develop the process. No, in exploration there is only failure when you don’t learn from your mistakes.

Yet this is exactly the problem, because for many organisations failure is just that, not having met the goal, often not even knowing what the actual goal is. And these organisations are stuck, because nobody’s exploring anymore.