MPEG Streamclip - How to think about settings to convert files?

The thing I hate about MPEG Streamclip is *TOO MANY OPTIONS*. (I prefer Quicktime 7 Pro.)

You *never* want to export to AVI or WMV unless you will be editing on a Windows machine.

You *never* want to use JPEG anything - it is too highly compressed. For FCPX, always select a ProRes codec (for MPEG-2, even ProRes 422 Proxy should be good enough - that would be the highest compression you should tolerate. I would generally recommend ProRes 422 LT as the lowest quality. The highest quality would be ProRes 4444 which retains the chroma at 4:4:4 with the extra 4 bits for an alpha channel [transparency, which you will not need for any other video transcode.])

If you will be working with video in FCPX, you ALWAYS want to use 100% of any optional parameter.

For FCPX, sound should always be WAV, AIFF, or CAF, linear PCM - and 48kHz -- NEVER AAC/MP3 or any other compressed format.

Stay with the original Frame Size, whatever it is. (NEVER zoom up [ > 100%])

Know the frame rate of the original and use that. I *assume* a blank frame rate in M-S means it will use the original (however - in case of a variable frame rate, which is possible, you will need to specify the nearest best framerate that FCPX will accept: 23.98/24, 25, 29.97/30, 59.94/60 — with the exception of 25, use the fractional framerate where applicable -- FCPX will not accept video that is variable frame rate.)

You will almost certainly WANT to deinterlace MPEG-2 IF it was originally interlaced... DO NOT deinterace progressive footage under any circumstances (you will lose half your resolution.)

IF YOU MUST deinterlace, then you must check Deinterlace Video and you select Lower Field First for MPEG-2/DV/DVCPro and all related "DV" video sources (MPEG-2 is "DV"D). ALL OTHER video sources will be Upper Field first. If you get this backwards, you will end up with video that will appear to go 2 frames forward, 1 frame backwards going frame by frame, or very "jittery" video upon playing.

Quality "loss" rarely happens when you go UP in bandwidth (expansion) which you will automatically do transcoding to any version of ProRes from MPEG-2 [any changes from the original will be virtually unnoticeable]. You only lose quality going down in bandwidth, or compressing greater than the compression of the original (in general -- the H.264 codec can compress quite a bit better than MPEG-2 so the "virtual" loss might not be as noticeable compressing to h.264 — but you shouldn't be doing that going into FCPX anyway.)

For sound, you will want uncompressed -- 48kHz. Uncompressed audio is Linear PCM (generated by M-S — I checked.)

For interlacing, in the player window, advance frame by frame -- if you see any "combing" effect (lines of the image offset from each other) the video is interlaced. If every frame looks like a "perfect" image, it's progressive. MPEG-2 is used quite a bit in broadcast television (old style) and for DVDs. Progressive DVDs became somewhat standard over time, but television stayed with interlaced media. If your source is from broadcast television, it is 80-90% likely to be interlaced. [Also a very good chance any camcorder of that era would output interlaced video as a "direct to broadcast" "feature".]

You can opt to deinterlace if you want. I would recommend deinterlacing DV/MPEG-2 sources, but not others. The quality of the video is just not there to matter much. It's basically "TV quality" (standard broadcast television.)

Like I originally said - I do not like MPEG Streamclip because it has too many options... too many opportunities to "screw it up"!

Frame Blending is a method used to "smooth" the visual perception of the playback of interlaced video. There are several methods for blending and I do not know which one M-S uses. Not using it usually means that every other field is duplicated or doubled, basically reducing the vertical image resolution by 1/2 (not the size, just how many "lines" are used.) With it turned on, the video will still look interlaced, but every other line is "averaged" from the lines before and after it — and that's basically "baked" into a progressive frame (so, no way to really get rid of it — doubled, or duplicated generally looks better to me). Downscaling is pointless -- you should not be changing the size of the video *at this stage!* I personally would leave them both alone.

No matter what software you use, you want to get it from whatever format you have into ProRes/WAV (or AIFF/CAF) format with as little change as possible with the minor exception of deinterlacing any DV footage — and even that is not that important -- FCPX will be able to handle it for you as long as it knows the correct Field Dominance is Lower First. The only reason I recommend deinterlacing DV/MPEG2 at *this* stage is because FCPX will expect Upper First for interlaced media if it isn't in its original DV wrapper. If the video looks bad after transcoding with deinterlacing, try again with Frame Blending turned on. If you still don't like it, turn off deinterlacing and frame blending altogether. You can change the field dominance in FCPX by selecting the clip in the Event Browser and going to the Info > Settings pane in the inspector to find the Field Dominance option (I don't recommend turning on Deinterlace [which will be in that same pane] within FCPX at all. Adding the clip to a progressive project will deinterlace automatically.)

Good luck!

Hey fox

Just curious - you said you prefered QT Pro for encoding before edit? I was wondering, what would you recommend for exporting, after edit, say, MP4 for Youtube? Would you use QT Pro there too?

Absolutely recommended. I've tried using Compressor (3.5 back in FCP7 days) and never was able to get satisfactory results (too many parameters for me — I always messed something up) and Compressor always took much longer to transcode. [I have not purchased Compressor 4.x — I'm perfectly satisfied with the quality I get from QT7Pro.]

QT7pro is straightforward. It gives YouTube exactly what it wants. If I stay within recommended limits, YouTube *will not* transcode the video again! Which means, my uploads are processed and ready almost as soon as they are finished uploading.

I export most of my projects out of FCPX to hard drive as ProRes 422(LT). Usually more than enough bandwidth. If I need more, then I'll jump to 422(HQ) and if I need maximum quality: 4444 (very rare.)

When you transcode for YouTube for 1080, these are the overall targets:

Make sure Prepare for Internet Streaming is set for Fast Start.

The compression settings are as follows:

Set the Frame Rate to 29.97 (or the exact framerate of your project). Key Frames: Automatic (let the compressor figure it out). Turn OFF Frame Reordering (Quicktime can handle reordering, but YouTube is not prepared to deal with it.) Data Rate: Restrict to 8000 kbits/sec.

This is based on the current YouTube requirements: https://support.google.com/youtube/answer/1722171?hl=en (if you look down the list to Bitrates, use the maximum Video Bitrate for the format you are trancoding. Do not use the High Quality settings listed unless you qualify as an "enterprise"!)

When you select Restrict To, the Compressor Quallity slider will max out at High (you cannot select Best.) The Encoding option should always be Best quality (Multi-pass).

A few years ago, YouTube used to have a problem playing back 1080HD without "squashing" a few lines (it was obvious at the time... and annoying, especially for screencapture tutorials.) I discovered if I set the Preserve Aspect Ratio using: Letterbox, that the squashed lines disappeared. It's become a habit.

When setting Dimensions, always use the sizes above the NTSC/HD Broadcast settings (like HD 1920 x 1080 16:9)... There is a difference and the HD broadcast settings will not work well on YouTube.

These are the sound settings:

I use 160 or 192 kbps VBR AAC (the Sound Quality slider is not specific, but if you refer to the first image above, you'll see the kbps selected there). As far as I've been able to tell, this is perfectly acceptable for YouTube (even though they recommend Constant Bitrates) and the VBR setting will not cause YouTube to re-encode.

If you exceed the Advanced Encoding Settings laid out by YouTube, I guarantee your video will be re-encoded by YouTube to fit. That means a second compression on your work. That means less quality than you expected. If you do all the heavy lifting yourself before uploading, you will already see what you can expect to get from YouTube. 8000kbps for 1080 in H.264 is going to result in more than satisfactory video quality 99% of the time (it used to only be 5000kbps!!) and your upload will be WYSIWYG.

Uploading directly from FCPX to YouTube has never made any sense to me. I do not trust YouTube to do a good job of transcoding the video with quality. Sharing to H.264 from out of FCPX is a great way to get a high quality movie for archival purposes, but it is completely unusable as an upload source for YouTube (and you have no options to adjust for bitrate/bandwidth). You never want YouTube to transcode your video for you... period.

QT7Pro is easy to use. There are not nearly as many pitfalls as there are in Compressor. Also, this workflow is an order of magnitude less time consuming than trying to upload straight to YouTube from FCPX, or sending the project to compressor. FCPX takes almost no time to export to ProRes LT (especially if the project is pre-rendered), and QT7Pro transcodes to H.264 in about half the time Compressor (3.5) would. And when you're video is live almost as soon as it finishes uploading, it saves time all around. [By almost I mean that after upload/processing, YouTube takes time to find a set of Thumbnails from which to choose... that seems to take a minute or two.]

As a PS - export should be to .mov, not .mp4. YouTube handles .mov files just fine (contrary to the document I linked you to above — former versions of that document, before Google took it over, included .mov container files as an acceptable upload format [and included a separate section for Mac users!]). Always use H.264 codec. MP4 is an older, less efficient codec and not anywhere near the quality of H.264.

Awesome! Thank you so much! 🙂

Did you say you have tutorials on youtube? What's the channel? I'm ready to subscribe!!

https://www.youtube.com/user/fxmah

And thank you for subscribing!