[Mark Suszko]

Does it need to be a real, existing wall? Or can you cut to a prop wall or even a sub-scale miniature?

[Mark Suszko]

I think it’s typical that the live captioners add a modest upcharge to convert their transcriptions to VTT.

If you’re really hurting for budget, a trick I use for shorter projects is to upload the video to an unlisted youtube and let their AI take the first crack at it. From their editing interface (when you have a creator studio account) you can edit and correct the mistakes the AI made and then download an SRT or VTT file. The quality of the transcription depends a lot on the clearness of the audio and a lack of accents or dialect. One good clear voice at a time, in a clean room, well-recorded, comes out of the box about 90 percent correct. Off-mic, echo-ey rooms and people speaking too fast or with strong accents – you get 50 t0 60 percent accurate, and have to clean the rest up manually. Youtube’s AI doesn’t do punctuation or capitalization either, nor will it assign name headers as each new voice chimes in on the track… But hey, it’s a start and better than nothing. Those extra features are why live human transcriptionists are still worth what they want to charge.

There’s some other free transcription systems out there, I don’t think they do any better of a job though, and they may not come with an easy editing interface like Youtube has.

[Mark Suszko]

The broadcasting school in Salt Lake City churns out hundreds of such newsfolk annually, suggest you try the coms dept. at Brigham Young U

[Mark Suszko]

“I learned it from YOU, dad!!!”

[Mark Suszko]

What would you think about using the fresnels targeted at bounce cards, instead of going direct thru diffusion? I had some good luck putting an HMI keylight behind the talent and bouncing it off a foam core panel on the camera side, the HMI in that position had just enough spill to act as rim/backlight AND keylight in one instrument.

[Mark Suszko]

The reverse side of a roll of vinyl sheet flooring, primed with Kilz Latex and then painted, makes an excellent backdrop that’s cheap yet also durable.

[Mark Suszko]

I can’t help you much with the camera side, but you can learn a lot of the basics about how to light, without owning any lights, or needing any large studio space, by modeling a scene in a 3-d animation app like Poser. You don’t have to be an animator: the program comes with pre-made people and props, and there’s many additional ones available online for free or at low cost.

But the point here is to learn what a frame looks like with different lens settings ( these are changeable in Poser), camera placement, and lighting. The program gives you spot lights, softlights, linear lights like florescent and neon tubing – all with the ability to modify color, intensity, beam spread, falloff… and the settings can be keyframed and animated to change over time… you can spend a LOT of time playing with all this and rendering out stills to show what you can accomplish.

The camera and lens can be animated to do trucking and jib moves as well as focus pulls, too.

One can also do this in Apple Motion or Adobe AfterEffects, but I think Poser is the better place to do this. One can often find Poser or Daz 3-d on sale online; you don’t need the newest version, any of them will do for lighting training, as long as it can run on your computer.

[Mark Suszko]

Greg’s description is pretty much the standard. There might be variations regarding stereo and surround audio tracks that could complicate matters but personally I think it’s been years since we were asked to do this on a deliverable, since they are played off servers now. In fact, on the last round of broadcast spots for handling by the client’s network, strict requirements for our spots that there was to be only 2 frames of black before the first actual frame, and only ten frames of black after.

But it always defaults to whatever the station specifically wants. They may need to be able to see it in pre-roll to help cue a live take to air. Or its the only way they can calibrate. That someone still DOES calibrate anything for playback in this business warms my heart.

[Mark Suszko]

https://www.openculture.com/2015/01/bound-by-law-free-comic-book-explains-how-copyright-complicates-art.html

[Mark Suszko]

Two guys with very sensitive shotgun mics, one on each side, could handle this, each operator covering one half of the set. Not perfect, but better than a fixed camera-mounted shotgun.

This would be a good scenario for boundary mics, which can cover 2-3 people each with good results, but you need a place to lay them and your set doesn’t have that. if you had little tiny cocktail tables interspersed between the folks, one table for each three people, the mics could rest on those. They lay flat and are not very visible.

I think how I’d approach this, if I couldn’t mic each person, is I’d give the host their own lav mic, then hang some omni- directional mics from overhead, just out of shot, one mic for each close “cluster” of 4 individuals, then you could keep each of those mics on a separate, recorded track, or feed them to a live person to mix in real time, or to an auto-mixer.

The auto-mixer is something that’s evolved from phone conferencing technology: it listens to something like 12 or more mic inputs simultaneously, and when it hears one of those start to have more signal, because it’s nearest the one talking, it ducks the other mics and brings up the talker’s mic in a space of a millisecond or two, then drops the talker’s mic level once it no longer detects a steady level of speech. We use one called “gator”. It’s not perfect, but it does the job.

Since you have two or more rows of people talking, it could be possible, instead of hanging mics from above, to hanging omni lavs on the backs of some of the people in the front row, to pick up 2-3 people immediately behind that person.

Part of your calculation of expense has to consider the savings in getting the live audio more or less right in real time, live, versus the longer time it takes to mix in post. If your post time is essentially free, you can afford to do it there, sure. But it can take three hours to mix one single hour, perhaps longer, if the conversations are rapid and there’s problems to deal with here and there. It’s not always immediately apparent, which path ends up really being cheaper.

As to retroactively improving the existing shows, and helping fill in gaps in the coverage of your future shows, I’d strongly suggest investing in the audio app/ plug-in called Izotope. A little pricy, but the results are very worth it. It can take out the hollowness of off-mic sources and restore problematic tracks.