Yesterday, I finally posted the long awaited (or, since no one cares, randomly new) camera reviews and an overview of my Asus EeePC 1000HE netbook. Again, since no one cares, I'll leave it up to you to view the Youtube videos.
Anyway, the two camera reviews were uploaded directly to Youtube (the Powershot SD780 IS review was shot with the HF20 and vice-versa) and varied in terms of processing time. Since the HF20 review was shot in 720p/30fps .mov, it uploaded an processed fairly quickly. The Powershot review was uploaded in 1080p mode in "24p," my framerate of choice, in .mts format. This one took longer, one because the file was bigger and in a less accessible format, but also because everything about the file was larger/more complicated. Nevertheless, they both posted without a hitch.
The Asus review, however, was not so lucky. Due to my own idiocy, I over-ran the time on the review by about 6 seconds. This wouldn't normally be a big deal, but considering Youtube is a bitch about that 10 minute rule, it wouldn't do. "No problem," I thought, "I'll just throw it in Vegas and do a quick edit/render."
No problem indeed.
The editing wasn't an issue. My system (2.66GHZ Core 2 Quad 8200 with 6 GB DDR3 1600MHZ RAM running Windows 7 Ultimate) can handle the AVCHD files natively with no hiccup. My Vegas Pro 9.0 is the 64-bit version, so I eke out a bit more optimization from the cores. Like I said, the editing wasn't an issue. And the PROCESS of rendering wasn't an issue either, as the system was more than capable of performing the task.
Problem is, I had to render it (18 minutes for a 10 minute clip) SEVEN TIMES to get an acceptable file.
That's my problem with any editing, and I'm new to this so it might just be my noobosity checking in. But why in the hell can't I load in like files (or even trim ONE file) and have the output match the input? I had to test tons of different output render settings (tried MediaConcept .mp4...shit. Tried uncompressed .avi...shit. Even tried .wmv...shit) with no luck. Frustrating/time wasting to say the least.
Here's how I finally got a decent output file, and this was only after scouring/begging on the internet for hours:
1) Re-started the whole project as new.
2) Set project settings to 1920x1080 at 23.967 progressive scan
3) Loaded .mts file in
4) Edited it down
5) Chose render as, with the following settings:
Sony AVC, .mp4, 1920x1080, CABAC encoding, 20000000 bit-rate, best render settings, default audio.
Took 23 minutes to render. When it was finally done, I checked the file really quick just to see if the quality was bearable, then I loaded it into Youtube. Twelve hours later, it finally processed and I realized that, while the tolerable, the video quality left much to be desired. The main issue was the ghosting (watch the video and see the part where I wave my hand really quickly over the netbook towards the beginning of the video itself...blur and ghosting like a bitch). The quality itself isn't bad, but the ghosting ruined it for me, and I couldn't fathom what the issue was.
Enter: Eugenia.
Here are three links I found to be insanely helpful, and if you are encountering this problem with any of the Canon AVCHD camcorders, I suggest you check them out.
If you're not interested in reading those posts, I'll summarize.
Basically, the problem isn't with Vegas per-se, but with the footage itself. What Canon says is 24p isn't ACTUALLY 24p at all, but an odd restructuring of 60i. It looks like 24p in raw footage, but Vegas doesn't recognize it as such.
Now there are a couple of fixes for this. First off, I could just shoot footage in 60i from now on, but that isn't what I really want. I want that 24p, that filmic quality. So that option isn't really an option. The next fix is to remove the pulldown from the footage, which results in an honest to God, genuine 24p image. Problem is, this can get expensive.
For people without FCP or a Mac for that matter, the best, and one of the only options for pulldown removal and intermediary footage is Cineform's Neoscene program. The program is available from Cineform's website for $129. Normally this wouldn't be a problem, but my normal *ahem* "sources" are fresh out of legitimate Neoscene programs. The trial is free, but lasts 14 days, which is a stop-gap solution more than anything.
There ARE freeware methods, but reading the tutorials makes it more trouble than it's worth. You would have to download somewhere around 10 different programs and codec packages, essentially render the video from the get-go, and RE-RENDER after editing, and even then the results aren't stellar.
Yet these two options are the only way to avoid ghosting when using Vegas.
Solution: Adobe Premiere CS4 and Adobe After Effects CS4.
While I wanted to avoid the Adobe suites (a little too daunting for me, at least from my limited use of Photoshop), it seems as though the CS4 suites not only handle AVCHD natively, but allow for 24p pulldown removal natively as well. This not only cuts down on time, but also the hassle of jumping through numerous hoops to get a working 24p render.
So, to summarize, it isn't really a Vegas issue, but an issue with the setup of the Canon AVCHD camcorders. To lie and tout true 24p is wrong (with a little workaround, it is true, but not out of box), and the fact of the matter is I shouldn't HAVE to remove pulldown. That said, a suite that is otherwise comprehensive, like Vegas, should include native pulldown removal, or at least a downloadable codec/plugin to add such functionality.
Haven't had a chance to do anything with the Adobe programs yet, but I'm off tomorrow so I'll be sure to tinker with it. Needless to say, it's been a huge hassle so far trying to get non-ghost images from my final renders, and I'm hoping the benefits of Premiere/After Effects native abilities outweigh my personal aversion to the software itself.
Who would have thought that my biggest problem wouldn't be the on-system editing of raw AVCHD and instead some bullshit pulldown issue?
So it goes, so it goes.
More to come.
No comments:
Post a Comment