1GB Video is exporting at 40GB
I am new to Adobe CS6 hence my question.
I have a 10 minute video recorded on a Nikon D5100 DSLR Camera.
The video was captured at 1080p 25FPS and is about 1GB in size.
I imported the video into Adobe Premiere Pro and made my edits.
Now I want to export the finished video so I can upload it to YouTube.
However once I choose File > Export > Media and choose Export Settings > Match Sequence Settings to preserve my 1080p video, Adobe Pr CS6 exports it out at a whopping 40GB!!!
My ISP would have kittens if I tried to upload videos that were 40GB a go.
Can anyone tell me how I export a 1080p video that isn't 40 times larger than the original 1080p video.
What am I doing wrong?
Thanks in advance.
Hi Tommy, this all has to do with bitrate and codecs. Premiere exports your video at a higher bitrate than your DSLR recorded it at, hence the bigger file size.
Here are some video tutorials to help you get started:
How Codecs Work:
DO NOT MATCH SEQUENCE SETTINGS... ...EVER... ...When going to YOUTUBE. It will match the video as it is rebuilt (bit rate etc, with little compression).
[Eric Strand] "Hi Tommy, this all has to do with bitrate and codecs. Premiere exports your video at a higher bitrate than your DSLR recorded it at, hence the bigger file size.
Here are some video tutorials to help you get started:
How Codecs Work:
Understanding Exports: "
Close but not quite right yet. The camera didn't record a low bit rate. It recorded compressed bits where the bit rate would vary greatly, but the quality matched a set standard of retention on blowing up the file again. The compression algorithms in the camera provide for high quality with more frames actually captured. The chroma, Gamma, luminance and color data are all compressed heavily, while the images themselves are less so. The images are actually compressed by storing the full versions of several specific frames, but the values are then processed through drop in Luma, chroma and all that and stored that way in the file. Only the changes in the images between the important frames are stored beyond those frames, and again the luma etc are all compressed and only larger changes are noted. When played, they are then read as the difference from the cameras base setting, then changes are then read as changes from that value, and the rest of the frames are rebuilt by guessing the values around the changed ones based on the originals.
Youtube compression to H264 will compress the images by change-calculation and motion estimation (much like the above), but it will not drop the color balance information down. This usually accounts for much of the difference, but also, remember to set your bit rate to around 10mbps (or 10,000 kb data rate), also the audio should be compressed to 48k16bit for smooth playback over net speeds. 10mb for a target, 12 max. This is standard bandwidth allocation for entry-level broadband to keep up with. People with slower speeds will buffer several seconds before playback, and may get some buffering, but it will be playable. Your bit rate is not a measure of quality, as most would have you believe. It's a measure of compression; lossy compression, of course, but not as direct a quality measure as they would have you believe. If you compress something that's already compressed, you're calculating your new compression based on a scrambled mix of the data being roughly blown up and recalculated, and very slowly... ...along with very poorly. But if you compress something that's already blown up, you aren't roughly blowing it up, you've blown it up full freight; and the algorithm is much more accurate in compressing that without losing quality. Bit rate doesn't measure quality. It measures the strength of the compression, and (roughly) how fast to move the data you end up with. If you set a variable bit rate, and set the max to 10 or 12, the target to 8-10 and the bottom to 4, you should be able to make a YouTube capable file that really rocks. BUT... ...I suggest you follow the workflow below to get a better idea of how to retain quality. I suggest you keep a copy of this somewhere in case you want to go over it again, too.
Bit rate, Compression, And Quality:
Most people prefer to retain max quality, while getting the smallest file possible, when compressing a video. If this is you... Consider the following: Compression will take a file and calculate how much data it has to drop (dump; delete; forget; goodbye little bitsies) to match your data rate settings; that is, it will make 1second of video take that much bandwidth by calculating where and how to drop or squeeze data to get within a certain threshold of the data rate (bit rate) you set. If you are already compressed, it uses a baseline algorithm (a very rudimentary one that's low quality) to blow up and calculate the frames. The existing compression also slows the whole operation because it takes more steps to complete. IF there are edits, it probably has to CREATE frames and blend frames as well, which is also more difficult with files that are already compressed.
Why not start by blowing it up to an AVC-intra first in media encoder. This will put your color data back in, and blow up the frames. Start with AVCintra100, and you can add a second job for simultaneous render, do an AVCIntra50 (this is a proxy format, should have about half the size of 100 and will edit nicely). If you set your previews as iframe only mpeg, you can then tick use previews when you export.
First, relink to your AVCIntra 50:
Go to your project panel, right click your clips and select make offline. Then right click them again and select relink. Pick the AVCintra 50 proxy file. Check your edits. Render out previews by Rendering Work area etc if you wish to check the quality of your edits. Finally, select the clips in the project panel, right click, select make offline, right click again to get a menu and select relink. Select the AVCintra-100 file. Export the sequence to an AVCintra-50 or 100 file (dealers choice here, the 50 isn't as high quality as the 100, but gets really close and does a little compression). When that's done, drop that file into AME and export it to a youtube compatible with the following settings:
Variable Bit Rate---min 4to5, target 10to12, max 10to12. These youtube settings rock with 1080 or 720 video. If you do 2k or above, bump each of those up by 4, and you should do fine for quality, as long as you are working from a full format file and not a compressed one. DO NOT MATCH SEQUENCE SETTINGS... ...EVER... ...When going to YOUTUBE. If you're apple-centric and have Compressor, I suggest using Compressor, as the H264 algorithm does a better job of fitting to your bit rate settings while keeping quality. I also suggest using Pro-res LT and proxy for smaller project videos (LT is a mid level between proxy and full-format-every-frame, in pro-res, and you can't tell any difference between LT and full at 1080 sizes in the 422 codec; 444 can do 4k high color).
[Ht Davis] "Close but not quite right yet. "
The original poster asked why his 1GB file turned into a 40GB file. As you said, it sounds like he exported with Match Sequence Settings, thus Premiere exported his file at a higher bitrate than was recorded in camera, which therefore is the reason his file is 40GB.
[Ht Davis] "Why not start by blowing it up to an AVC-intra first in media encoder. This will put your color data back in, and blow up the frames"
The Nikon D5200 records a 4:2:0 color space, it's impossible to put the color back in. AVC-intra will be easier to edit with, but that color information was thrown out at the source.
[Ht Davis] "Start with AVCintra100, and you can add a second job for simultaneous render, do an AVCIntra50 (this is a proxy format, should have about half the size of 100 and will edit nicely)"
I'm not sure why he would need an offline/online workflow unless he is editing on an old machine. Any relatively modern machine running Premiere CS6 should be able to edit Nikon DSLR footage or AVC Intra 100 without trouble. An offline/online workflow is going to take unnecessary time and storage space.
[Ht Davis] " I suggest using Compressor, as the H264 algorithm does a better job of fitting to your bit rate settings while keeping quality."
This may be true for Compressor 4 but Apple's Compressor version 3 is widely recognized as having an awful implementation of the H.264 codec.
Hi Eric Strand & Ht Davis
Thank you both for your invaluable input.
I was indeed ticking the box that said "Match Sequence Settings" and my format was "AVI" this resulted in a 40GB sized file.
CreativeCow isn't allowing me to put hyper-links inside my posts yet, because I am a new member of the forum. So all I can do is give you the title of the video I just watched on YouTube, it was called
"Best YouTube Export Settings With Adobe Premiere Pro CS6"
I will still try and post the video's URL just in case the restriction has been lifted:
http://www .youtube. com / watch? v=4FbTsk_3QD4 (remove spaces)
I followed the advice in the video. I didn't tick "Match Sequence Settings" and the Format is now H.264. My preset is 1080p 25. I am a European (Hey someone has to be!) and my camera came pre-set to PAL. I don't know if NTSC is better or not??? so I never bothered changing it. So anyway my file size has now dropped from a colossal 40GB to a more respectable 2.8GB. Still more than twice the size of the captured footage!
Gents if you find the time, please watch the YouTube video its just 6m36s in length. I would be extremely grateful for your opinions on it please. As I am new to Premiere Pro so I don't know if the guy in the video is offering good or bad advice or whether his settings can be improved upon.
In particular I would like to know what you think of what he does at 3m 26s into the video, especially the adjustment he makes to the bitrate.
Thanks very much.
Hi Tommy, NTSC and PAL are technical standards. Europe uses the PAL standard, North America uses the NTSC standard. One is not better than the other. I'm not loving the Best YouTube Export Settings With Adobe Premiere Pro CS6 you posted. The guy in that video has a standard def sequence that he is telling people to uprez to 720? Then he sets his level at 5.1, which auto adjusts his frame size to 1080, but he doesn't say anything about that. Seems odd.
Have you tried using the YouTube 1080p25 preset that Adobe Media Encoder comes with?
Also, I'm hoping you watched the links I posted in my first reply, as they explain the basics of exporting and compression, something you're going to have to learn.
Sorry for my slow reply, Easter happened :-)
I was aware NTSC was USA and PAL was Europe, and that NTSC was 29 FPS and PAL was 25. Other than that I am oblivious to their differences. I keep wondering which is better. On-line opinion is divided; I was hoping someone with experience like you could enlighten me please? Which format makes for better video footage?
Thank you for taking the time to watch that video. As I suspected he was not as authoritative as he first appeared to be.
I am recording in 1080p and I wish to preserve that. So feeding my video through the Adobe Creative Suite and out the other side should not result in anything being "destroyed" i.e. nothing should suffer deterioration or degradation. Clearly Exporting Media is the number one place to get it wrong!
After watching that video I realised he was working with 720p footage so I decided to tweak my export settings within Premiere Pro as my original footage is 1080p.
So I picked the following settings:
Source Range: Entire Sequence
Match Sequence Settings: “unchecked”
Preset: HD 1080p 25
Export Video & Export Audio: “checked”
Frame Rate: 25
Aspect: HD Anamorphic 1080 (1.333)
[this was the only choice that made the source and output tab views match]
TV Standard: PAL
Bitrate Encoding: VBR, Pass 2
Target Bitrate: 32
Max Bitrate: 40
Use Maximum Render Quality: "checked"
This results in a file that is 2814MB in size or 2.8GB, nearly three times the size of the original.
If I change the target bitrate to 16 then the size is 1.4GB
And if the bitrate is 8 then the size is 716MB
This all makes sense as the size halves each time.
What is the cut-off point? When will my video’s quality begin to suffer? When will it be detectable to the eye?
On a separate issue, I am trying to correct colouring of my video.
So I am using Adobe Speed Grade. However I cannot open an H.264 MP4 with Speed Grade as it’s not a format it recognises.
If I change my format to QuickTime then confusingly the highest Preset available is 1080i there is no 1080p.
And finally if I choose to send the file to Speed Grade rather than exporting it, Premiere Pro creates a file that is a whopping 238GB in size!!!
Thanks very much for your help, its greatly appreciated.
[Tommy McMahon] "Which format makes for better video footage?"
PAL and NTSC don't affect the quality of your footage, they are technical standards.
[Tommy McMahon] "What is the cut-off point? When will my video’s quality begin to suffer? When will it be detectable to the eye?"
Typically 1080 footage is encoded around 8-12 Mbps. YouTube's own standards advise using 10 Mbps. If there is a lot of motion in your video, bump the bitrate up.
Unfortunately, I don't know enough about Speedgrade yet to help you with that issue. I've been meaning to learn it myself!
Is your source footage 1920x1080 or 1440x1080? You suggested it was the latter, however that wrong Sequence Settings might have been used when you set up the project originally. You can right-click a clip in the Project Bin to see the clip specs. Can also right-click clip and select "New Sequence from Clip" to ensure that the Sequence matches the source video.
If the source is indeed "anamorphic" HD, meaning the pixels are not square, you do NOT want to export as anamorphic. More often than not, it will not play back correctly as 16:9 when you do that. You want the 1.0 Pixel Aspect Ratio, not 1.333 PAR.
Whether source is 1920x1080 or 1440x1080, export as such: Choose H.264 for Format, and for Preset, scroll down to the bottom of the list and choose "YouTube HD 1080p25". This will use 1920x1080 with 1.0 PAR. Expanding the 1440x1080 source to full 1920x1080 will not hurt anything.
Alternately, one could also choose the "YouTube HD 720p25" and downsize the 1440x1080 to 720p. Then you can check the "Max Render Quality" box to assist with downscaling.
And back to NTSC or PAL, just MATCH the frame rate of the source when exporting, so use 25 in your case.
Don't over-think the whole process. Edit with correct Sequence setting, export using proper YouTube preset. Simple.
Safe Harbor Computers
Thanks for your advice, I was definitely over thinking the problem.
After watching a few tutorials I feel much more confident about my workflow.
OK thank you that clears up NTSC vs. PAL for me thanks.
It's great to know I can go as low as 10Mbps for rendering out to YouTube, my file sizes will be much more acceptable.
I watched both of those videos David Kong videos that you linked to in your earlier reply. They were very helpful. Thank you.
SpeedGrade is fantastic, I highly recommend it, after watching another tutorial I realise the best way to use it is at the start of one's workflow. i.e. colour grade everything before you begin cutting, editing, adding audio and effects as it makes for a much simpler process. This cuts out the headache of trying to export a video from Premiere to SpeedGrade.
Thank you very much for your help.
Jeff is right about the preset. Ue the one that came with AME, but adjust your bit rate settings as necessary.
Compressor 3 had a bad implementation until 3.5 and the updated components in the codec, just before compressor 4 came out.
By the way sorry about the rant. At 3 am, I get a little punchy. Especially since I have to be up at 5 or 6 am, alternating, including weekends.
As for the comment about older machines... ...I do my work on varying arrays of machines. My workflow is designed to remove all problems that MIGHT occur with ANY FOOTAGE. I get clips from everywhere and anywhere. I remove any Frame rate and sync issues early, and get everything into a single format so I can color grade and apply it at the best quality, with a short render (only a few seconds worth in each clip) that allows me to gauge the color balance. You call it slow, I call it complete. I get pro quality, and exceptional quality on a 2008 MBP with a 2.16ghz core 2 duo and 4gb ram. I do my output to a full format, then carry that over to a lab to compress it, cost is usually about 4 hours of render time in compressor with one machine and down to an hour per hour of video with a set of machines all running compressor with the right settings. On top of that, I archive the entire project to disks (excluding the full format blow ups) so I can revisit it later. It's easy because I place it al in the same folder structure on disk images. I set the size the project should take, and grow it when I need to. I still get quality that makes video guys wet themselves laughing at the joke when I tell them I got it from an '08 mbp with 4gb ram. It's no joke. It works. Adobe CC is great, but I prefer my cs6. That said, I've tested alterations to just about every setting I can think of for speed hikes. My findings on this issue are thus:
Cache files can be tricky--on mac they need the mac permissions structure and disks that conform to it; on windows they are less permission dependent in cs6, but in CC, there have been stories of mishaps outside of the NT file system.
Previews can go anywhere, but keep them together (audio and video in same location), and it will index them faster. Keep them with your project file location, and it will default to checking the top level folder when looking for resources (a relative path) that is less likely to break when moving between operating systems.
Other files can be placed anywhere, but I like to keep my proxies and original video in the same volume as my project because it will find them faster when searching for them to relink when moving between machines.
I can put the output from premiere anywhere, so long as the other machines I use to compress it can read that location. I place the compressed file I get from compressor into an output folder on the disk image, and link that to encore as a transcode of the premiere sequence, and I also place the audio next to it. When linking, Premiere and Encore will seek the top level folder, then seek the other files from there by matching the relative path.
At the end, I close the image, I RAR the image to split it into pieces of set size, then burn to discs and wipe my drives with a quick wipe and start fresh. I can output for youtube, dvd, blu-ray... Not just for slow machines. For a complete workflow...
I get rid of any possible problems first, I don't waste time checking for them. On top of that, I provide archival. I can rar, and burn at any time to make a backup. I already have an incremental imaging program handling that by backing up to 2 externals, slowly, but it happens every other night. Knowing that, I get my files ready the first day, start work, and do an initial backup, then update only the changed files every other night after. No whoops moments.
My working drives are all firewire and esata. My backup drives are usb and firewire. I use almost every port on my MBP for drives. Of course, I image all the cards\cameras first, and then transcode, but that's all usb. Takes about 3 hours for 4 cameras with 2-3 hours of footage each (17-25gb per card). Then it takes a lot less time to get the transcodes from faster drives, and I get to keep a copy of the original card, freeing up my cameras for more usage and footage. A complete workflow for craziness and constant work. My minimum deadline is a week to 10 days, to allow for basic editing and output. With added 3d effects and such, 2 weeks. Most are in line with that.