Hi all,
I've heard good things about this forum so I figured it was worth a shot to ask you for help with a multicam concert video I'm trying to cut. I knew nothing about video the day before yesterday - I'm an audio guy - and I'm trying to handle an 11 camera angle concert video we had the fortune to record recently.
The shots were delivered in 29 files. Most were .mp4 files from GoPro cameras, but there were some .mts files as well. This is what I noted about them (you will probably see I don't know what I'm doing):
Three different GoPro-types with .mp4 file extensions:
- MPEG-4 (Base Media / Version 2) with AVC High@L4 video format profile. 1920x1080px, 16:9 and 25.000 FPS.
- MPEG-4 (JVT) with AVC High@L5 video format profile. 2704x1504px, 16:9 and 25.000 FPS videos.
- MPEG-(Base Media / Version 1) with AVC High@L5.1 video format profile. 3840x2160px, 16:9 and 25.000 FPS.
Three other camera types with .mts file extensions:
- BDAV (Blu-Ray Video) with AVC High@L4 video format profile. 1920x1080, 16:9 and 25.000 FPS.
The total size of these files were around 100GB.
My system specs (I'm sorry to say):
i7-3770K 3.5 GHz
16 GB RAM
Win 10 Pro
Asus R9 270 3GB VRAM
I've managed to splice them together into 11 files, and cut away what I could at both ends, resulting files of about 25 minute length. I have managed to import them into DaVinci Resolve and sync 5 to create a multicamera shot from which I can cut the concert video. I had a working proof of concept to try the whole thing. That, however, is where it broke ... my GPU clearly isn't up to task once I start loading more cameras and/or the higher resolution files.
A new GPU is out of the question, so what are my options? Can I re-encode these files in a way which gives me good 1080p footage all round, and probably allows me to edit as an 11 angle multicam shot? If so, into what? I am so out of my depth here with all these formats, codecs and containers... I wish to preserve as much video fidelity as I can (audio doesn't matter), but will never render anything beyond a 1920x1080 file from this. I also don't want to spend a month re-encoding in various formats to find the one that works.
We are not a professional band. I just want to do this thing decently for us now that we got this footage. Hope someone can help me out here!
Cheers,
Kvebbs
Try StreamFab Downloader and download from Netflix, Amazon, Youtube! Or Try DVDFab and copy Blu-rays! or rip iTunes movies!
+ Reply to Thread
Results 1 to 5 of 5
Thread
-
-
11 HD+ angles simultaneously is a LOT for a PC to deal with, no matter what you do.
My recommendation: Create low-rez (SD or maybe 720p25) proxies, using a virtually lossless (but lossy) intermediate - Prores, Cineform/GoPro, DNxHD/HR, Edius/GV HQX codecs - so that you have lower CPU/GPU issues (but will need more storage space).
Edit the proxies and then conform the originals to your edit.
Scott Warren -
Thank you, Scott. That looks like the way to go
One further question, just to be safe. Last night, the GPU warning popped up when I tried to render. That won't happen when I conform back to originals?
-Kvebbs -
I'm so sorry. I sent a reply earlier today, but it happened just as I was leaving WIFI area, and I fear the message (long as it was) was lost...
I cannot guarantee that you won't have issues, but the whole point of doing proxies is to alleviate the burden on the cpu/gpu compositing/rendering engine. You do the editing & compositing in one stage, where the quality isn't a main requirement.
Once the elements have been decided upon and the final image layout is created, your proxy conforming will just be using the fewer elements that comprise the finished product, so even rendering will likely be operating on just a small fraction of those original frames.
IF you still have issues with a rendering of the whole session timeline, one option to try is to render to a less onerous compression format, such as a lossless intermediate.
Another, even more helpful cheat is to render only segments at a time and then concatenate them into the whole.
IOW, render part A then part B then part C, and then make D = A + B + C.
I had to do a similar project 4 years ago when I shot & produced a 4-angle multicam videos for a dance festival. It lasted >18 hours throughout 2 days ( ~13 on day 1 and 5+ on day 2).
I had to render (using ProRes) an hour or 2 at a time.
Make sure you do not overlap (or skip) frames!
Render Video A + B + C, etc. Then render the whole Audio track (it is not as burdensome, so should be easy to do the whole thing, and it doesn't react well to overlaps/skips - creates pops & ticks).
Concatenating doesn't even have to be done using a full-blown NLE - ffmpeg or AVIDemux can do that kind of things properly.
In fact, major complex studios push this style of rendering to the extreme when they render to still image sequences (in whole or in part).
Hopefully, you won't even have to deal with that, though.
Scott -
Thank you so much for taking the time to write all that out for me! I'm going to try this out tomorrow! Fingers crossed.
Cheers,
-Kvebbs
Similar Threads
-
GPU or CPU encoding?
By m00511 in forum Newbie / General discussionsReplies: 44Last Post: 17th Nov 2019, 19:03 -
HEVC or H264 GPU vs CPU Another time !!!
By gringito38 in forum Video ConversionReplies: 2Last Post: 23rd Feb 2018, 05:28 -
There is too much about GPU vs CPU encoding, but how about GPU filters?
By Bernix in forum Newbie / General discussionsReplies: 24Last Post: 16th Feb 2018, 16:17 -
Multicam stream setup and multicam end user switch option
By SBJ in forum Video Streaming DownloadingReplies: 2Last Post: 16th Dec 2016, 08:35 -
encoding video why are cpu best vs gpu ?
By tola5 in forum Newbie / General discussionsReplies: 15Last Post: 11th Jun 2015, 15:39