Optimum settings for Streamclip to get best results with FCP

[Andrew Waite]

I just wrapped my first feature length project last night and now ramping up for post. The entire film was shot with a Canon XH-A1 and a Brevis 35mm Adapter in 24F HDV. I used my 2.4 Ghz Macbook Pro to capture all the footage live with Adobe On-Location (via bootcamp) into M2T files. I plan to cut in FCP, Edit in HD and Output to BOTH HD and SD in hopes to distribute the film.

My question is this. What is the best method of conversion of the M2T file format to Quicktime with streamclip, or any other program for that matter? I'm looking for the highest quality I can achieve. There are many codecs available (i.e. Apple Intermediate, Prores, Jpeg something or another, etc.) and many filter options and quality settings (i.e. upres to 1920x1080, deinterlace, etc.) and I'm a little lost on which is best and for what reason. I am new to the Mac platform (coming from Premiere Pro) and am unfamiliar with the various Quicktime codecs. I have been unable to find any threads that address this particular topic and any help would be appreciated. Maybe streamclip isn't the best solution and if you could suggest a better option that too would be appreciated. Thanks!

[Andrew Waite]

nobody? nobody can help me on this one?

[Andy Mees]

i will have a think about it ... but for the moment I am laughing too much :)
in the meantime, have you had any joy with Log and Transfer directly in FCP?

[Meryem Ersoz]

what are your plans for delivery? best to know this up front before you select a codec....

[Andrew Waite]

Originally Posted by Andy Mees

i will have a think about it ... but for the moment I am laughing too much :)
in the meantime, have you had any joy with Log and Transfer directly in FCP?

what's so funny? i have not log and transfered, no. I used on-location for several reasons... besides the scopes and monitoring options it was the only way i could be assured that i wasn't going to find that the only take i got had dropped frames and was unusable after it was too late. i did use tape as a backup measure. also i found streamclip appealing because it has an option to flip the image which can save me time being i shot this project using a brevis 35mm system. so, rather than dumping over 30 hours of tape it would seems quicker to just use streamclip on the 3 hours of usable stuff i already have on my hard drive. right?

as far as final delivery... short answer is DVD. i don't have to give this project to a client or anything, but dvd sounds like a logical choice. i would however like to edit in hd so i always have the option of outputing to hd-dvd or bluray should i get the opportunity... i shot the damn thing hdv after all.

[David Knaggs]

Hi Andrew.

I've thought of a suggestion which might (ultimately) lead to the best possible output to both SD DVD and HD DVD (and Blu-Ray, once DVD SP can work with it). But it will add an extra day or so to your workflow.

I'm going to issue a sort of "disclaimer" first. I only work with HDV1. So I'm really only guessing when it comes to HDV2 (which is what your Canon XH-A1 shoots). But until a Canon user is kind enough to post a solution from their own experiences about using MPEG Streamclip for optimum results with cinelike-mode 24f 1080i HDV2 footage, this might at least offer a workable solution instead of waiting.

I would skip MPEG Streamclip totally (unless you get some GENUINELY expert advice) because you really don't know what you're doing if you try to de-interlace "24f" 1080i footage to get a more "progressive look". You might unwittingly degrade the footage.

I recommend the simplest method (stay with me on this, as I will explain why).

1/ Log and Capture the three hours of footage (not the thirty hours) directly with FCP using the Easy Setup for your particular type of footage. (I'm assuming that an Easy Setup exists for this in FCP6.)
2/ Edit in a native timeline.
3/ Export directly from the timeline via Compressor to create your SD DVD assets (.m2v and .ac3 files).
4/ Export directly from the timeline via Compressor to create your HD DVD assets.
5/ Import those assets into DVD Studio Pro (DVD SP) to author (separately) your SD DVD and HD DVD. But note that Macs can currently only burn a regular red-laser DVD, so your HD DVD (unless it's only 20 minutes or so long) will probably have to wait until proper HD DVD burners are available and compatible with DVD SP. But at least you'll have your HD DVD assets ready and waiting.

The reason I recommend native capture and editing is because every time you transcode (change the footage to another codec) you WILL introduce a degradation (it might be only very slight with the better codecs). When you go to transcode your .m2t files in MPEG Streamclip into AIC or ProRes or whatever, you will introduce some degradation.

That means you'll have introduced 3 separate compressions from the capture of the image to the final DVD product, i.e.:

a) Recording the image (MPEG-2 compression from your camera).
b) Transcoding in MPEG Streamclip (a new compression in AIC, ProRes or Photo-JPEG 75%).
c) The final DVD compression (whether SD or HD).

If you capture and edit natively, FCP is set up internally to not add any extra changes to your footage until you actually export. If you export directly through Compressor it will only then do an MPEG-2 compression for your SD DVD assets. So your images only get two compressions (the original camera compression while recording and the final DVD compression).

Whereas, if you export as a Quicktime movie (rather than through Compressor), or even a Quicktime reference movie, it will add an HDV2 (MPEG-2) compression to all of your transition points and effects/filters (such as colour correction). Then, when you import the Quicktime movie into Compressor to make the DVD assets you get a third compression. That's why I recommend exporting directly through Compressor, if you want the very best visual results. But it has the drawback of tying up FCP until the assets finish encoding, so I know that a lot of working editors (when working with high-quality codecs which give minimal degradation) will export as a Quicktime reference movie and then import into Compressor so they can keep working with FCP in the meantime. It's a trade-off, but a very acceptable one in most cases (especially when working with codecs that are considered "visually lossless"}.

Anyway, those are your export options.

Finally, the reason I suggested logging and capturing from the tapes is that I am assuming that your on-set capture software "Adobe On-Location" captured the .m2t files with timecode. So I figured (rather than trawling through 30 hours of footage) that you can open FCP and set it in Log and Capture mode, then open each .m2t file on your hard drive (only 3 hours worth) in MPEG Streamclip and apply "Fix Timecode Breaks", then note the In and Out timecodes, toggle back to FCP (Apple-Tab) and log those timecodes, load the next .m2t file into MPEG Streamclip, etc.

This should take less than three hours, then you can get FCP to capture all of the logged clips automatically. All you have to do is insert the next tape into the camera or deck when it tells you to. So, hopefully it will only add a day to your workflow.

But, bear in mind, this is all a suggestion from someone who doesn't work with HDV2. But native capture and editing in FCP is a reliable and "safe" option that can end your period of waiting for a Canon user to volunteer his/her expertise.

[Meryem Ersoz]

david, can you (or somebody) explain the difference between HDV1 and HDV2? i've heard these terms floated around and am not really certain that i understand them, nor the implications for using one or the other....

andrew, i'll try to answer your question, but i'm probably not the most qualified person here...i'm still struggling with HDV workflow issues of my own.

first of all, i transcode m2t files with MPEG Streamclip to get them into FCP, and i think it works well. but as david suggested, every transcoding that you do adds compression to your file. i don't see an appreciable difference between a transcoded m2t file and a natively captured HDV file, but i suppose it is there. someone with more technical expertise can address the "whys" of this. i can only calls 'em as i sees 'em.

advantages of the various available quicktime options:

HDV

advantages: looks nice, and you're working in the native stream

disadvantages: long, frustrating render times of an intra-frame codec, unless you have a pretty powerful computer. i have a 3.0 ghz dual core tower loaded with RAM and and a 2.17 ghz macbook pro with 1 gig RAM, and the render times are night and day. i don't know how the performance is of the 2.4 ghz notebook, but you may find rendering frustrating. i definitely get frustrated with my laptop, but with a little kid around, sometimes editing on the fly is the only way to go, while watching High School Musical.

also, HDV can do some glitchy things at the delivery stage. if i'm editing in HDV and going out to the web or to DVD, there can be some nasty artifacting. that's why i asked about delivery. it can really vary, depending on your delivery codec. for instance, if you try to go straight out to Sorenson 3, you may get some weird audio and image issues unless you transcode to AIC (or some other inter-frame codec, like ProRes) first. and if you're shooting interlaced (which you're not), de-interlacing seems to be an absolute necessity to get to DVD with Compressor 3. my big disappointment with FCS is that Compressor 3 seems to be a step backwards from Compressor 2 in terms of transcoding HDV to SD DVD. i never used to get the jaggies until i upgraded from 2 to 3. some upgrade! i have not had time to de-/re-install, but i plan to.....



AIC:

advantages: it converts from long-GOP to an interframe codec, therefore render times are substantially faster.

disadvantages: i don't hear much reference to this one, but my opinion is that AIC has a more washed-out looking color space that HDV--that's the biggie. the file sizes are huge, and storage may be an issue on a longer project such as the one you're describing. but with the price of cheap storage, probably not an authentic issue.

ProRes:

disadvantages: well, you can't capture with it, not without a capture card, so that's a big disadvantage. there is no native acquisition format. so no prores without transcoding. of course, this is a capture-from-tape issue and does not apply to you, since you have already acquired your footage straight into your system.

advantages: i'm still trying to figure this out. i bought FCS 2, mistakenly thinking that this would solve some of the editing issues with HDV, but so far, i am not seeing any advantage over AIC. in my experiments with it, transcoding into it it seems to keep the color space intact (unlike AIC) but seems to produce a softer image than native HDV, to my eyes. if i had a capture card to capture it natively, i might see better images than i have with transcoding from HDV.

i have not tried this, but if i was in an experimental mood and in your situation with already-captured m2t files, i would try the 3rd-party bitjazz sheer lossless codec. the people who use it really seem to like it.

that's what i know to date. i haven't experimented with photo jpeg compressions or DVC proHD compressions--although some folks claim that the HDV color space is enhanced using this codec, i have not really tried it.

for your own peace of mind you might want to just try a few quick experiments. burn an SD DVD with a quick clip which challenges the codec (in terms of motion handling, color, noise) and see what you prefer. if i was invested in a feature film, that's what i would do. a day spent finding a satisfying workflow and, more importantly, a satisfying result in terms of how it looks upon the delivery of your choice, is well worth the time spent, especially if you are not up against a particular deadline.

i know that i have considerably more experimenting to do before i am wholly satisfied with my workflow and final products. what i have concluded is that you can read all you want about how other folks are doing it, and the tools that they are using, but there is no substitute for your own eyes. and you will see things that you may or may not see described in internet postings.

hope this gets you started. ah, the halcyon days of DV. things were so much easier way back then, in the good old days, six months ago! not quite as pretty though, HDV is a pain, but it sure makes some lovely images. bring on blu-ray!

[David Knaggs]

Hi Meryem.

Our own Chris Hurd defines HDV1 and HDV2 beautifully in the last paragraph of this page:

http://www.hdvinfo.net/

[Andrew Waite]

wow, i had no idea. i always thought that m2t WAS the raw format. on-location writes captures all hdv to that format and i just figured that was as pure as it gets. so what codec or format is recorded on the tape itself? i was under the impression that on-location wasn't compressing anything and it was just eliminating the need for tape which has given me problems with drop outs in the past.

[David Knaggs]

Hi Andrew.

The same compression is on both the tape and the "On-Location" .m2t files. It should be HDV 1080i (MPEG-2 Transport Stream).

When your Canon camera captures the image, it applies an MPEG-2 TS compression (HDV 1080i) and sends it out to the tape and also through the FireWire port. (I presume your On-Location was hooked up to the FireWire. Some cameras have HD-SDI ports which bypass the compression, and if you happened to be capturing that, then your footage would be uncompressed, of course. But, as you said your files are .m2t, then it's been compressed with HDV 1080i.)

Final Cut Pro will not recognize an .m2t file. It recognizes and works with Quicktime files. So you have to put a "Quicktime wrapper" around the .m2t file.

FCP native capture (from the .m2t files on the tape) simply places a Quicktime wrapper around each captured .m2t file with NO transcoding or alteration.

MPEG Streamclip will only transcode and THEN place a Quicktime wrapper around the file.

I once emailed the originator of MPEG Streamclip and requested the function be added where a Quicktime wrapper could simply be placed around the .m2t with no transcoding. But I received no reply and, to date, this function has not been added.

Now, of course, if you used MPEG Streamclip and set the "codec" to "Uncompressed", it would then NOT be adding any more compression while adding the Quicktime wrapper. But it would depend if you have enough drive space, etc.

That's why I recommended native capture and editing. It gives you quite small file sizes while not adding additional compression steps (until you export). And, as it's a feature film and probably your "pride and joy", I figured I should mention these options.

[Andrew Waite]

wow... thanks! i'm starting to understand now. i've never heard of a "wrapper"... although, what you said makes sense to me. i think i will probably start dumping my tapes then using FCP. i just there was a way to have the footage automatically flip as it was capturing or something.... i really thought i would be saving time by going direct to disc, i guess i was wrong. the only real benefit i guess was scopes and double redundancy in the event a tape dropped out at a critical moment.