Finishing in Premiere

Thought I'll post this here. I just tweaked a set of promos in Premiere. The process is absolutely lovely. This is a short summary of the processes.

- Extracted the clean VO track from the full mix with a phase inversion of the mix minus VO track. I needed to do this to edit and version the promos because the original edit was done in another country on the Avid and getting original assets from those guys was a pain in the behind. Doing this took about 10 minutes, while getting the original assets will take about a day and a half. Thank God I didn't sleep in audio class all those years ago.

- Did up the titles on the end page in After Effects by tracking the existing graphics to use as motion attributes on the additional titles. Used a couple of built in CC effects and some masking and feathering for a feathered wipe dissolve.

- Versioned the titles. The way Premiere works with titles meant I can swap out one title and have the change ripple across multiple titles, so I had separate titles for "Monday" and "Tonight", and another set of dups for the different timezones and I only needed to change one title instead of the whole bunch.

- Graded the promo in SG via Direct Link and adjustment layers in Premiere. I used adjustment layers because I needed to do some work on a few shots in After Effects so the SG color correction was done on adjustment layers.

- Witness protection masking and tracking in AE via Dynamic Link and the mosaic effect. The mask tracker feature makes this very easy and pretty fast. It's an area tracker, so it's much faster than point tracking. On one of the shots I had about 15 masks.

- Burnt in subtitles with the built in Title Designer. Loved the fact that I can draw a rectangular mask behind the text and adjust the opacity and see the final image while doing up the title. This is an often maligned feature, but very very useful. And unlike the default title tool in FCP, I can actually adjust the font style of a few words without having to create a separate timeline object.

- Audio mastering. Used the built in EQ effect on track level to enhance the VO and compress the VO track. And checked the levels with the built in Loudness Radar and mixed program loudness at -23.9 LKFS. Also used the free Melda Production Phase Scopes to check for phase inversion on the mix and check that the final master is stereo. It's crazy that I finally have these tools to ensure I meet broadcast specs without having to spend thousands on hardware meters.

- Rendering. I rendered preview files to ProRes in Premiere because I needed to deliver ProRes QTs. This also lets me QC the final render. Premiere remembers the preview files very well, so I only needed to render my grade once and copied and pasted the grade into the different versions. I checked "Use Previews" to smart render to my final output. Sent the whole lot to queue in AME and compressed each version to H.264 mp4s for the low resolution preview copy. This basically allows the parallel encoding function to kick in, so while it is exporting the QT, it is also simultaneously encoding the H.264 version of each file. Multiproc processing.

No need to work overtime. Can't wait for the next update of the Adobe tools to see what improvements there will be!



www.strypesinpost.com

Wow, you're really flying that thing hard Strypes! I've just spent three sessions teaching a large government media department transitioning from FCP7 to PPCC and my favourite bit was when I asked one of the guys who was apprehensive at the beginning if he was OK with the program at the end he said 'Not OK. I LOVE it!'

Right now we're cutting a series for Nat Geo, and have a new 4k series about to start. We're waiting for the new Mac Pros to drop here - would love to hear anyone's experiences with PP on the new MacPro. Our biggest issue right now is ray tracing in our AE graphics. I've asked the graphics guys to precomp a background so we don't have to deal with it but they're reluctant. Not sure why - possibly what I'm asking for is more difficult than I imagine.

Oh, also, I checked the clock differences between workstations at that job where the peak generation issue is happening and they are indeed set to different servers. I'll let you know if the problem goes away once I can get IT on to it.

Thanks for sharing that list - there's lots on there I didn't even realise could be done.

I won't dynamic link a comp with ray tracing. Ray tracing needs a fast CUDA GPU. Without it, it can be SLOW. If you're doing edits, there's no point dynamic linking to a ray traced comp, unless you intend to tweak the text constantly, but there will be a hit on render time.



www.strypesinpost.com

We're not dynamic linking, but it's still murder on several of the machines (clearly the ones without the GPU grunt) We have a set of series templates where we insert footage into provided boxes on a shifting 3d field, and the background is bent into a curve using ray trace. We have to turn it off to even work on some machines. But I'm thinking that the 3dness might be part of the reason it can't be pre-comped? Not an expert here by any means.

Nice Strypes!

Michael Horton
-------------------

Hi Jude, not sure what you mean by not dynamic linking, unless you are working in After Effects. But I really haven't used Ray Tracing much. I tried getting some 3D text done for an end page with Ray Tracer and because I didn't want to install Element 3D on the work machine.. to cut the story short, I didn't have CUDA and I didn't like the waiting time.

I like Premiere for editing, but the finishing tools and effects is where Premiere really holds its own.



www.strypesinpost.com

Yeah, we're in AE, not PP. Exporting final comps to PP. Sorry, should have clarified.

The clocks apparently were set to the same server, but they have different names (?) So.. not sure if that is still part of the problem. It actually hasn't happened for a few days. Maybe we got an update and I didn't notice.

How did you do the phase inversion extraction? That could come in really really handy.

Yea. Ray tracing is very GPU heavy so definitely you would like a good CUDA GPU especially for final renders. You can preview in draft mode while working. Alternatively you can use Video CoPilot's Element 3D or C4D. There are reasons why you would avoid pre-rendering certain layers as compositing sometimes requires effects (especially lights and shadows) to interact with all the layers.

Inverting the phase is very simple operation. When 2 identical tracks are played out at the same time with a phase inversion on one track, you get silence. A simple example is if you take a VO track, you duplicate it to the track beneath it. Drop an invert effect on one of the clips and hit play. You should get perfect silence. That's also the reason all broadcast mixes are required to be mono compatible and the full mix when summed to mono should be in phase.

So in my case, I had stereo full mixes, and a mono mix minus VO track, and I needed the VO track. So I dropped the stereo full mix into the timeline, made sure they play out as mono, dropped the mix minus VO track onto another track, and used the "invert" effect on it. The resulting output should be the clean VO track which I can then export and use as needed.



www.strypesinpost.com

Ohh, you had the M&E - that's brilliant.

Quote
- Rendering. I rendered preview files to ProRes in Premiere because I needed to deliver ProRes QTs. This also lets me QC the final render. Premiere remembers the preview files very well, so I only needed to render my grade once and copied and pasted the grade into the different versions. I checked "Use Previews" to smart render to my final output. Sent the whole lot to queue in AME and compressed each version to H.264 mp4s for the low resolution preview copy. This basically allows the parallel encoding function to kick in, so while it is exporting the QT, it is also simultaneously encoding the H.264 version of each file. Multiproc processing.

I don't typically render in premiere but I think maybe I should start?

I'm looking into a faster workflow for exporting RED footage, often times I'll send clients cuts, different versions, etc...
and exporting RED footage takes quite a long time even on a late 2012 imac with all the trimmings.

In FCP7, you would need to transcode, but after that you could export to Pro res really quickly. Then create a mp4 of that in Adobe Media encoder pretty quickly.

My final export is typically mp4 for these low res previews but I'll export to anything if it's faster.

So what should be my workflow for the fastest export once I have a RED project ready to export...render, then use previews?

Thanks

Hi Joe, one thing you can do with Red footage is to right click on the clips and go to source settings and lower the debayer resolution.

The other setting that affect render times is GPU. If you have a fast GPU, that can speed things up but if your GPU isn't that fast you may get better exporting times using software rendering. This can be set in AME on export.



www.strypesinpost.com

Can you batch lower the debayer settings? Doing them one by one would not be worth it.

I have the GTX 680MX 2gigs of Ram, not the fastest but okay.

So is it a no to the render then use previews with RED footage?

Yes, you can batch lower the debayer settings by selecting a bunch of clips and setting it to half.

Prores is a smart render codec so if you need to render during the edit, it will allow you faster export to a Prores master. In theory, if you have a fully rendered timeline, exporting with used previews checked should let Premiere use the preview files, but do some tests with that. There shouldn't be a speed difference in total render time between rendering during editing and exporting with use previews or not rendering until the final output.



www.strypesinpost.com

.

How do I batch lower the debayer settings for all clips?

I'm also noticing that most of my clips have source settings grayed out in the project window but
not in the time line....maybe another merged clip issue.

Select clips in the project panel or timeline, right click and choose source settings. Set to half.



www.strypesinpost.com

I have my files separated in bins, when I do select all, source settings is grayed out,
When I try to select all clips just in a bin, source settings is grayed out
when I select a merged R3D clip, source settings is grayed out,
when I select just an r3d file source settings is there.

I don't think CC like me...

I'm getting a glitch in my final export, there seems to be no rhyme or reason as to why,
out of 8 exports it appears sometimes and sometimes not.



It's this white and green line across the top of the frame for a couple seconds, and it
hapens right after a fade from black.

Source footage: Red files, effect colorista (it may have something to do with colorista but why would it happen intermittently?)
Export: Mpeg2-DVD (tried cbr, vbr, and vbr2-pass)

Could be tied to resource usage. I hardly use CII, tbh. It's still very laggy in Premiere.



www.strypesinpost.com

Can you try going through the Media Encoder by hitting 'queue' instead of 'export' in the export window to see if this makes a difference? I've found AME to be WAY faster than a flat-out export in the last CC bump on a 2010 Mac Pro.

If there's still a problem, maybe try exporting an intermediary file doing less compression at once, like a pro res - then go from this to MPEG2.

Hey Jude,

Thanks, like you I always hit queue because I too find it faster than a straight export from premiere.

I just have to assume it's something with Colorista, I managed to export it with it and not get the glitch but just frustrating because sometimes it's there and sometimes not.

making a master to pro res is not a bad idea.