Making a reel is hard. Don't make it harder by working blind. Tune in to the best reel critique show around, and while you're at it, here are a few things you might learn from it.Read More
As mentioned here, I recently purchased a $16 Floureon off the Amazon Deal of the Day. Is it a good mic? Oh no, nononononono, but it's fun, as evidenced by how many people are playing with it. And it's gold! What's not to love? Well, okay, the build quality, and the sound, but other than that...
For a bit of fun, I dragged the Floureon into the studio at work and threw it on a stand (that cost more) and plugged it in (with a cable that cost more) and did a quick voice shootout. It's below. Enjoy!
P.S. Aaaaand beaten to the punch by the amazing @mikedelgaudio
I was broken and reforged in the crucible of GDC, as always.Read More
Is a $16 microphone the key to great recordings? Oh god no. But is it fun?Read More
Constraints can be useful. Sure drawing inside the lines is no way to get ahead, but if you're the one drawing the lines...Read More
It's good to have a hobby that doesn't involve work. An outlet for your creativity, but make sure you have an outlet for your baked goods.Read More
Twine is a great tool for prototyping, don't let it's complete lack of tools, audio, and visuals scare you off!Read More
Somebody asked me at GDC, "So you bake, huh? What's the best thing you've ever baked?" No hesitation. No pause to consider. "Blueberry boy bait, first runner up in the 1954 Pillsbury Bake off (Juniors category). It's amazing." Here's the recipe from Smitten Kitchen. Go, make this today, make it right goddamn now. It's magical.
Even better, if you're looking to take your baking game to the next level, this is a great recipe to start with. It's no more complicated than a boxed cake recipe, and by experimenting with the sugars, you can take this from super soft to almost crunchy, with a sweet sugar layer on top and bottom. Adjusting the brown to white sugar ratio, or substituting coconut sugar, yield different, but still excellent, textures. Fresh and frozen blueberries both work great, and demonstrate the subtle, and sometimes not so subtle, differences in fruit selection. In fact, a little birdy told me that dried cherries soaked in brandy and drained work well in this. Of course, that's getting in to cranberry/craisin territory, which is a whole other thing.
Make this recipe. It's great, it's worth your time, and it goes over great at the office.
I'm not much of a video editor. You need your short film cut? Yeaaaahhhh.... I'm not the guy. You need 10,000 video clips trimmed by exactly 1 second, then transcoded into 3 different formats, all automatically in the middle of the night? That, that I can do. Between Adobe Media Encoder (expensive, heavy, but effective) and a cron job watching a folder on the video server, most of my video editing is of the, "I'll kick it when I'm back at my desk" variety. This month though, things have been a bit different.
My current challenge is I'm working on a project that involves about 400 hours of interviews, all shot in the Doomsphere at the ICT. The interviews are cut into hundreds of individual responses, and that's where the problems come in. 400 hour long R3D clips isn't so bad, but 10,000+ individual clips requires some work. Suddenly, small choices matter, and sometimes, you accidentally torpedo the project schedule. Each clip's in/out points were delivered in an Excel sheet in HH:MM:SS. No fractional seconds, no frames. This was a problem. Thankfully, I had two tools at my disposal that have put us back on track.
First, know when to automate
To get those I/Os from Excel to a format our NLE could read was our first use of automation. I wrote a Python script that pulls in the Excel file, grabs the In, Out, and Clip ID, converts the HH:MM:SS to HH:MM:SS:FF, inserts an extra frame to the out to make sure the clip is at least 1 frame long, then drops the result in a text file ready for import to Vegas. Every time the timecode rolled over, split the output to a new file. Piece of cake, and saved hours of manually slicing up Excel sheets.
This left us with 400 hundred tab delimited text files ready to be dragged into Vegas, but somebody still had to go through and adjust every single cut. We budgeted 2 interns for 3 months. We didn't think that'd be enough. We needed a better plan.
Second, know when to hire a professional
We posted a job for a video editor, a simple AE type job, meh pay, few perks, but all the coffee you could drink. We figured they could get it done faster than the interns, and would have a better eye for when to cut than a total novice. So we made a hire.
It took her just over a week to finish the first 60 hours. She got bored waiting for the farm to render more proxies and went back and reworked the first sessions. Then she labeled all of the audio glitches and started learning RX. That's why you hire someone who knows what they're doing. 3 weeks ahead of schedule, now we had to figure out how to get the new I/O's out of Vegas and onto our render farm.
Go with what works
The task was simple, open the Vegas project, open the edit details tab, select all of the regions, then copy them into a text file. My first thought was a custom Vegas script, or maybe using the SDK to launch the project and extract the data, but the SDK is a bit uh... thin on documentation. I was about to put the interns on it when somebody mentioned AutoHotKey on #gameaudio. So today, I exported the regions to text. Well, my computer did. I listened to Beyoncé. My mouse danced over the Vegas interface, spawned Notepad++, named files, and waited patiently while I occasionally checked the output.
Okay, so how?
For the Python bits, I cribbed big parts of the technique from Al Sweigart's excellent Automate the Boring Stuff with Python, specifically the chapter on Openpyxl, which I've used for some other things. There's a Python module, Timecode, that is vital for this kind of work. As for the AHK script, it isn't perfect, but you can hit me on Twitter if you want a copy.
Yesterday, one of the researchers I work with asked me about transcoding some video for ELAN. I'd setup their ELAN workflow a few years ago, but since then a few staff had moved around, ELAN had seen a few releases, and they'd upgraded their camcorders. This meant the old script I'd written (cough cough batch file with YOURFILEHERE scrawled all over it cough) had been lost, mangled, or was otherwise indisposed. No big deal, how hard could it be to concatenate a few MTS's and then transcode to something ELAN would like?
I'll spare you the details, but this is how I spent my 4 o'clock hour on Friday.
ffmpeg -i 'concat:c:\foo\barA.mts|c:\foo\barB.mts" -c copy barAB.mts
or, if you want to do it in one step...
ffmpeg -i "concat:c:\foo\barA.mts|c:\foo\barB.mts" -c:v msmpeg4 -c:a libmp3lame barAB.avi
Probably, second one is from memory, and you'll want to do some bit rate and resolution adjusting. We annotate 720p proxies (great for turn taking, head nods, posture analysis) when we don't need high detail (eye blinks, gears turning in subjects' heads, AU markup) but do what works for you.
Now, if you're looking at that and going, 'Wait, why are you using the wrong concatenation setup?" that's what took so long. All of the video is hosted remotely, and that server wasn't playing nice with FFmpeg. Probably permissions between Windows and the Linux server. Throw in some issues with escaping the paths (no quotes! Windows build handles that for you.) On top of that, the researcher wanted a single line, no external files, and definitely no Python dependencies solution for her interns. On top of that, ELAN leans on Quicktime and WMP for video playback, so good luck picking a codec. So is it pretty? Oh no. Does it work? Mostly. That's why I love FFmpeg. When it works, it's great, and when it doesn't, you'll probably learn something.