I am editing a project in CS4 that was (unavoidably) shot on 4 different cameras. We used a canon xha1, hv20, 7d, and a kodak zi8. Naturally, the footage is all different and it's giving me hell. The xha1 and hv20 I know are in 1080 60i, so that is what my sequence is in. However, my 7d shoots with a huge resolution and the video jerks a bit. I re exported it to avid format and the video didn't show up in my source or program windows. The Kodak I believe records in the same crappy h.264 codec and is also giving me junky playback. If I open a sequence to fit the 7d, it will be to big for the other 2 canon cameras (and possibly the kodak).
I'm not sure even what to ask, but I just felt like sharing my impossibly over complicated work flow.
Edit: http://vimeo.com/forums/topic:23548 is where I got the idea to convert the 7d/kodak footage
+ Reply to Thread
Results 1 to 16 of 16
-
-
What format is your end goal? e.g. blu-ray, dvd, web ? etc...
What format is the 7d and zi8 footage (e.g. 1080p24, 720p60 etc....) If you don't know , use mediainfo
In order for PP to "see" dnxhd footage, it has to be installed on your system, with quicktime . To be honest, it's not that much "snappier" to edit on windows than the native footage. This is because anything stored in .mov container is poorly optimized on PC. Cineform is much faster, but unfortunately, not free -
My final output will be 1080p 29.9fps meg2 mainconcept and I already have dnxhd installed
zi8 is 1920x1080 29.9fps mp4a: MPEG-4 AAC LC codec
7d is the same, but with sowt: 16-bit signed little-endian codec
I might also add that the 7d and hv20 are mine, but this is my first real project (not just on test shots) with the 7d
Edit: I just waited on a 20 minute render w/ clips from all cameras. Everything has decent playback with the exception of the 7d footage. It looks strange (almost compressed) -
what do you mean compressed? do you mean bitrate/size wise, or do you mean levels or gamma conversion
how did you do the conversion to dnxhd? One thing to be aware of is a levels and gamma shift that premiere does with dnxhd/mov as it converts to RGB internally
There's probably several ways to do this, and it will depend if you favor speed vs. quality.
Since your format goal is 1920x1080p30, I personally would deinterlace and resize the 1440x1080i60 hdv footage to 1920x1080p30 using 3rd party deinterlacer in avisynth (premiere is horrible for resizing and deinterlacing quality wise). The intermediate you could use is dnxhd also (so everything now would be 1080p30 dnxhd). I would use a custom sequence setting for all the footage (1920x1080p30, which matches your export settings). The negative to this is your hdv footage will now get yellow/red render bar , and will be converted to RGB internally (usually not an issue, as hdv is easy to decode, and you usually convert to rgb anyway when adding filters, color correction etc...) , and of course time taken to encode to the intermediate -
there is a small "ghosting" that occurs in the rendered native 7d footage. The converted footage did not work right, I am sure I messed something up there.
-
ghosting implies you did a framerate conversion, or did some other processing like blend deinterlacing
again, how did you do the conversion (e.g. though AME, ffmpeg, mpegstreamclip etc...)
look with mediainfo and check the specs are the same (1920x1080p30 before and after)
(by the way, when i say 1080p30 i really mean 29.97 fps) -
I am using the native file, friend. I converted nothing. This is the file I pulled off my compact flash card.
I think it is going to come down to having to render a lot. Thanks for your input about the using the custom sequence! -
are you seeing the ghosting in quicktime player by itself? if so, what were your camera settings, shutter speed - because this would suggest it's in the original file, not by premiere interpreting the file
Depending on how you view it or interpret the file in PP, or adjust with speed changes, you can get ghosting.
if you put a 1920x1080p29.97 native 7D clip in a sequence with the same settings, you shouldn't get any ghosting -
"jerky" is a world different than "ghosting". Make sure you are describing this correctly. Ghosting is a blend of fowards and backward frames, so you get a double image of sorts.
jerky is usually from slow/old computer or poorly optimized software for playback , but can be from duplicated frames due to framerate conversions / speed changes done in your editor
h.264 isn't bad per se, it's just that the software isn't well multithreaded for the .mov container on pc platform. Quicktime is one of the worst players available. The h.264 implentation of the 7d is poor also, it uses a very low quality profile, but at least it uses a high bitrate to compensate for these deficiencies. But in turn, a higher bitrate is more CPU intensive for decoding - which makes it harder to edit... a vicious circle
for example re-wrapping it into a .mp4 container and using better software decoder like ffmpeg in mpchc makes it play back just fine (even on older computers), even though the actual video stream is identicalLast edited by poisondeathray; 16th May 2010 at 20:53.
-
I understand the difference my friend. I have been futzing with filters and effects for about 8 years now. It is indeed creating a small "ghost" like feature as well as a jerky feature while unrendered. However, it seems to be minimized after a render. My problem was that the bar above it was yellow, and the clip will not render until I have to render another clip with a red bar (go figure?!). I am not sure to why this is. Fortunately this is a music video type project and I can live with the produced effect. (in fact it might make it more appealing)
"the software isn't well multithreaded for the .mov container on pc platform"
Couldn't agree with you more. I am a PC user and I run into all sorts of platform issues. I happen to like h.264 myself (i like the output quality and compression size), but I tend to be add it to my laundry list of issues and complaints on dealing with different codecs. Once I get the money I am investing in a Mac laptop simply for file conversion use. (i.e. when I receive a file that was captured on a mac that I need to edit on my PC)
This project has a deadline coming up and I seem to be working around my issues atm. But, in the future, I am going to try taking the 7d's footage and play with the output from Mpeg Streamclip and whatnot. I also happened upon a copy of Neocene. I need more extensive testing with the resulting conversion to avid/avi/mov/misc other output formats.
This is my FIRST real project with my new 7d, and ironing out my work-flow kinks is going to be a priority for me (lol after I finish this project).
as a side note... while I saved time avoiding a real-time capture, I made up for it by encountering other issues. -
If you're seeing this in the native footage (you can use another player like kmplayer or smplayer for example) , then likely it was shot with problematic settings (like bad shutter speed). If you see it in premiere, it can be due to a number of things depending on your settings.
If this is the type of thing you are going to be doing a lot , I would spend the money on CS5 and a proper graphics card. The mercury playback engine makes everything smooth , even several layers of 7d footage (or other kinds of h.264, even AVCHD, or 4K footage) with effects in realtime without rendering. This is truly amazing. Some people have "coaxed" the MPE to work with "lower end" , less expensive cards (instead of quadros, or top end consumer cards) through registry hacks. It really is the fault of software optimization and being able to leverage CUDA acceleration. A $1000 dual core system with compatible card is easily 10x faster editing/scrubbing on CS5 than an $5000 dual quadcore workstation editing on CS4 or unoptimized software. No joke.
Still, using 10-bit intermediate (such as cineform or dnxhd) can provide other benefits than "speed of editing". You get much less banding when you do levels / gamma adjustments with higher 10-bit precision, even if your export format is 8-bit. "v210" is uncompressed 10-bit 4:2:2 , and another option as well. You need fast I/O (e.g. SSD's , RAID0 HDD's ) in order to work with this because of high data rates. -
I am in the process of saving up for CS5
As far as system goes, I am running SLI with a GB of video ram and 6GB of DDR3 (soon to upgraded w/ the addition of CS5), and an overclocked 2.6 i7 processor. It just "Grinds my gears" whenever these things happen. I truly appreciate the info and advice. -
This is really interesting and something i want to look into. Not because i want to force it work with a lower card, just that I understand it would work with a GTX 285 but not a 295 like i have in my second system. If i can 'coax' it into working with it and using all that power thats in there then i definately will put cs5 on it.
Similar Threads
-
Correct Project and Sequence Settings for footage from Panasonic TM300?
By Bloodredthorn in forum EditingReplies: 0Last Post: 20th Jan 2012, 12:34 -
need some help with sequence maker
By beta72 in forum EditingReplies: 1Last Post: 17th Oct 2011, 20:44 -
FCP 7: HD Playback with NTSC Sequence Settings?
By TeeHee in forum Newbie / General discussionsReplies: 3Last Post: 2nd Aug 2010, 22:17 -
Recomended sequence.
By Tafflad in forum RestorationReplies: 0Last Post: 19th May 2010, 15:31 -
VIDEO TS sequence
By red lion in forum EditingReplies: 1Last Post: 25th Jun 2007, 19:15