Again I find myself in HDV Hell. Further proof that this format is pure and utter crap. OK, fine for shooting…whatever. But never ever ever NEVER work in an HDV sequence. Ever. Those who do, if they are well informed and choose to ignore the advice given here and on the forums, deserve to be flogged. Beaten to death with dead rodents.
What now you ask? Let me tell you…
I was approached to help figure out a problem with a project. A 90-min documentary shot on HDV, edited with FCP, and using an HDV sequence setting. The show is fully rendered, yet every effort to export a Quicktime Movie, or go via Compressor (this is FCP 6, so no SHARE option), results in either a GENERAL ERROR (love that one) or an OUT OF MEMORY error. I have 8GB of RAM, I’m out of memory? I can’t export the full show…I can’t export even five minutes of the thing without getting that error. Absolutely maddening.
This project is LOADED with graphics. Text effects. Tons and tons of moving text and keyframed pictures with text overlays…stills, video, text..sometimes 18 video tracks deep! It’s intense. The original editor is running a G5, and they thought that the machine was the issue. But no, I get the project and media here, on my Octocore MacPro 3.0Ghz machine, and I have the same problem. So it isn’t an Intel vs PowerPC processor thing. I just can’t export a darn thing. And I have been tasked with making a DVD and an H.264 file for the web.
Well, the DVD was easy. Just output the video from my MXO2, downconverted the signal to SD and connected to a DVD Recorder. Press record on the Recorder, play on the timeline…realtime output. Still one of my favorite ways of doing this. MUCH quicker than authoring. No menus, but I don’t need those for this stuff.
But now the real trick…the H.264 QT movie. I can’t export anything from this HDV timeline, so I am stuck in the mud. I did notice that the Render Controls were set to render as ProRes, but that didn’t seem to be making any difference. The render files are ProRes, but still no go. I guess that something in there doesn’t need to be rendered, and still references the HDV master…so… I don’t know, just grasping at a cause really. When I try to export a self contained movie, FCP says it will take 5 hours! And then it bombs out…but note that time. When I tried the 15 min chunk, it said just over an hour. And still bombed out. OK, I made a few attempts to figure out a work around for this.
Ok, the first thing I thought of was to make a new sequence…this one a ProRes sequence. I copied and pasted the HDV sequence content into this one and rendered. After about an hour the first 15 min was done. Not bad. And then I chose a 15 minute chunk, and exported that as a self contained file. It took 8 minutes. FAST. ProRes is definately better than HDV…proof #1 there. But then I watched the export and…OOOPS! All the graphic elements were messed up! They didn’t line up like they used to and they were squished. OH, right! HDV is thin raster 1440×1080…anamorphic. And ProRes is full raster 1920×1080. Those clips will have some DISTORTION in their motion attributes. So when I try to just remove the distortion, they still didn’t look right. They were way off of alignment. So no quick fix here. These would all have to be redone manually, as they were all done internally in FCP, and not in Motion or After Effects and rendered out. And there are a lot of these effects. I can’t tell you how many text effects there are. Almost more than there is footage. A lot. So that wasn’t going to work. Not without a LOT of work.
OK…I was able to output a DVD via my MXO2…why not try something similar. I happen to have TWO computers, a MacPro and a MacBook Pro. And I have two capture devices…the MXO2 and the MXO2 Mini. Why not play out from one machine to the other, capture the signal as ProRes, and then render out that file to H.264? That sounds like it’d work. So I connected the devices via HDMI, and I got the capture tool to see the image. I played from the Laptop, as the tower would need to do the capturing…as the MXO2 does capture ProRes, but it relies on the COMPUTER to do the encoding. So the laptop isn’t quite up to snuff for 1080…it does 720p at 23.98, but this was 1080i 29.97. So I pressed CAPTURE NOW on the MacPro…pressed PLAY on the Laptop…and recorded for 12 min when…dropped frames. Damn. OK…try again. I got 8 min before the same thing. Hmmmm…. Let me turn off the dropped frame warning, it won’t look that bad, right? Wrong…there were two cases where I noticed it…a freeze frame for 1-2 seconds. Unacceptable.
OK, but how about the other way around? I know that I can’t to 1080 on the laptop, but it does 720…and I do have a 720p 29.97 setting. And the MXO2 does cross convert upon output, so it should work. So I swapped the cables from IN to OUT on both sides, and tried the other way. I got 23 min in before the laptop stopped capturing due to dropped frames. Damn. I have captured 720p 23.98 in one hour chunks before, so I know that works. I guess that 29.97 is just too much. OH, and the drive I was capturing to was a 2TB G-Raid3 via FW800. Playing from a G-Drive via eSATA. OK, so machine to machine won’t work, unless I have two MacPro towers. Darn it.
When I was sitting here pondering what to do next, I recall how on one job we’d play out from the Avid to a PC with a capture device and PC software that would encode straight to MPEG-1 (this was 5 years ago) while we played. VERY slick. And I was wishing for something like that, when I remembered, “Hey, didn’t Blackmagic Design make something for this?” I went to their website and looked and sure enough, the BMD Video Recorder did just what I hoped. It captured a signal straight to H.264…just what I needed. I called a local dealer, they had a bunch in stock, so I drove into Hollywood to get it. I asked “does this take HD video?” and the salesperson said, “yup.” So I bought it. When I tried to use it with the HD signal…it didn’t work. Then I looked at the manual and it states, near the end, that “this doesn’t work with HD video signals.” Son of a…! Drat…I didn’t research this well enough. I expressed my frustrations on Twitter when a follower said, “yeah, for that you need the PRO model.” I look on the site, and there is the nice pro model. Yeah, that will do what I want. But it is more than the other one…as it should be. And it was not only a tad outside my budget at this time, the dealer didn’t have any in stock.
I did a test capture and compared that to the H.264 encode of the ProRes output (the bad one)…and it was noticeably softer. This H.264 is intended to look GOOD, a way to attract investors for distribution. So this model, while darn handy for quick outputs for producers or clients to review the cut, it isn’t all that great for FINAL compression. Compressor is better for that. And I’ll soon see if the PRO model does a better job.
OK, I know that if I can get this to ProRes, things will go smooth and I can use Compressor. But that darn full raster issue. But wait, when you capture HDV as ProRes via firewire in that little trick that Apple made available, the ProRes setting that it is captured as is 1440×1080…anamorphic. So why can’t I make the timeline settings ProRes 422 at 1440×1080? Let me see… I go into the Sequence Settings, see that it is HDV, that there is no field dominance…so not interlaced. And I simply change the compressor from HDV 1080i60 to Apple ProRes 422.
BAM…I needed to render everything. And when I tried to do it all in one clump, I got the GENERAL ERROR and OUT OF MEMORY error again. So I rendered in small chunks…5 min…2 min….and it worked out. When I looked at the graphics, they were perfectly fine. So now I have a ProRes sequence, and the graphics are all fine. And after I rendered out 15 min…I exported the QT and it was fast. PERFECT! But the big hitch in this is that I have to render those small chunks, and babysit them. I am still doing it, as I write this. It has been an 8 hour ordeal, but it is almost done. VERY close. I did find that the estimate that FCP gave was never right. It’d say 15 min, but really be 24 min. Two minutes is six minutes.
Now it is done, and I have exported the self contained file. It took 30 min, but it is out and done. SUCCESS! Now compressing the H.264.
Don’t edit HDV on an HDV sequence. Not if you like things to go smoothly. Yes, you will have the green render bar above the footage, and yes, eventually you will have to render everything. But if you do it while you are working, bit by bit, you avoid the one massive render in the end. And you avoid the HDV madness. So if you want to save space by not capturing HDV as ProRes, and I can understand wanting to do that, then at least use a ProRes sequence while editing. Even if it is full raster, which is a good thing, the HDV will fill up the space.
And for all you Avid people that might feel like chiming in with, “Well, if you editing with Media Composer you wouldn’t have had this issue,” that is true. Because you’d be editing with DNxHD project and sequence settings. Just like if someone in FCP land edited HDV on a ProRes sequence. In both cases, you’d need to do a full render when you were done and ready to output.