Bad Encodes Out of Compressor
 

Guys this is my second try asking for help. My sequence plays perfectly through my monitor, yet when I export directly from the timeline I lose parts of the video during the MPEG encode. I won't acccept exporting as a reference movie or self-contained movie, as that is not a real solution--just a work-around. Does anyone have any tips for getting a good encode. My program is 124 min.

Sending a sequence out directly from the timeline is *not* the preferred method for sending something to Compressor, it's just an option, and works best with short sequences. If you're refusing to use the better, more stable alternatives then you're probably stuck with troubleshooting this ad-nauseam.

Every FCP certified trainer and all the Apple Pro series books suggest using either reference or self-contained movies as the best, most reliable method for working with any version of Compressor.

If you try those methods and you're still having problems you could have a corrupt sequence, which is more common with really long, single sequences.

From Timeline
 

It might not be the common or preferred method, but it is undoubtedly the best if quality, not time is of the most importance. The advantage of exporting directly from the timeline is that compressor automatically inserts i-frames at each edit point and both ends of all transitions. The manual says that this method has the best quality above all other methods, as it does not use rendered/re-encoded files to write the mpeg2 stream, unlike exporting a quicktime movie--it writes from the original source files, much like a compositing program (did I mention it takes a full day to write it?). I'm just wondering if I have an option or trick that I have yet to try which might help. My last resort is to use a quicktime reference movie, but I will if I must. Any thoughts?

Bryan can you share more information about your sequence settings, and which parts of the video come up missing? I am using export to Compressor the same way on a 50 minute FCP timeline, no problems so far.

Settings
 

Settings:

Compressor v 2.3
MPEG2 Variable 2-Pass Best
Mostion Estimation: Best
Max Bitrate: 9.3 mb/s
Avg. Bitrate: 4.5 mb/s
Anamorphic 16:9
TRT: 124 minutes
Audio: AC3 224 kb/s

I have several large TIFF files in my FCP sequence. They render and play back perfectly fine on my external monitor. Upon the compressor encode, ceratin sections show up as black space where the TIFF files were. I've encoded it 4 times, and each time the black sections are in different places in the video...but all on large TIFF files. If I break the sequence down into smaller segments, those encode fine, so it's not corrupt TIFF files. I know...Why don't I just export the sequence as a bunch of smaller segements? Well, that would defeat the purpose of the 2 pass variable bitrate encode, since the encoder has a smaller margin to work with when deciding how and where to delegate the bitrates for maximum quality. Any suggestions?

Are you working from a DV timeline, or HDV, or something else? Just curious. Also, did you say each encode results in black spots in different places? So are all of the TIFFs on your timeline not rendering, or only some of them?

Are you using a custom encoder setting or preset? If you use a Compressor preset just for the sake of testing it would control that variable.

Have you tried to replace your TIFFs with PSD versions? Don't know how large your graphics are but I've had little trouble rendering high-res PSDs in Final Cut.

PSDs
 

First, I'm working from a DV timeline. All of my TIFFs render fine and play back perfectly from the timeline. Compressor is just screwing up the encode. If I open up those TIFFs in Preview, resave them as PSDs, then reconnect them in FCP, will they retain their alpha channels. This will give me a huge file savings...but still...what could be causing the problem do you think? I have plenty of RAM...awesome Mac Pro computer...and plenty of experience on the platform, but I've never seen anything like this. Thanks for your help thus far.

Don't know about how Preview works with alpha channels as I use Photoshop for this kind of stuff. Might be worth a try to do it the way you described.

I can't tell you what is causing the problem. Could have nothing to do with hardware. All I have is hunches, which are a dime a dozen. Having said that, I think if you apply some quasi-scientific method troubleshooting to this you will nail it.

First of all, can you repeat this mistake/error? Or does the "missing video" phenomenon happen unpredictably? If you take out the TIFFs and send to Compressor, what happens?

.
 

PSDs will be my next resort. Either that or breaking it up into smaller peices. But the answer is no, the results are NOT repeatable, which is why it's driving me crazy. Should I mention that these TIFFs are 125 MB. Is that extreme or not. I have no clue.

PS: Those are good youtube encodes you have there. What source media were they shot on?

Re. YouTube, the best looking stuff on my page is HVX-originated 720P- but I think the lack of movement, at least in the Office Space clips, is the reason they encoded half-decent. My latest thing shot 1080i looks like crud on YouTube:

http://www.youtube.com/watch?v=Q4icdCfx97k

...so I think the movement is a real factor there.

re. TIFFS:
125MB? HUGE. Try Photoshop to convert those to more reasonably-sized PSDs and try again. As far as breaking it up, that sucks when you have to do that. You never said, are you using your own encoding scheme or Compressor presets?

My Own
 

My own presets...but it's within DVD specs...

Compressor v 2.3
MPEG2 Variable 2-Pass Best
Mostion Estimation: Best
Max Bitrate: 9.3 mb/s
Avg. Bitrate: 4.5 mb/s
Anamorphic 16:9
TRT: 124 minutes
Audio: AC3 224 kb/s

As Ben suggests, there's no reason for still images to be that big and probably why Compressor is having fits. If you're starting with full-res scan or digital images a 2Mb JPEG with a max pixel size just over the frame size of your video is more than enough. Still files larger than 5Mb in the timeline are not only unnecessary but are trouble spots for encoding.

I'd highly suggest picking up the book, "Compressor 3 Quick Reference Guide" by Brian Gary. It will outline the best methods for encoding. Ditto for the Apple Pro training series, DVDSP4 first or second edition.

I'll take it for granted you know what you're doing there...the reason I asked was to see if a standard Compressor preset would give you the same problems, or no. You could eliminate that variable...though something tells me the TIFFs are your culprit...

thanks guys
 

thanks guys. I have a strange problem though. I don't have Photoshop and these TIFF files were created in Photoshop Elements on a PC. I have to work on my graphics on the PC then bring them to the Mac. I know, I shoulda taken them plunge and got photoshop for the mac. My questions are:

Is there a way to compress graphic files from within FCP.

Can I save an alpha channel as a jpeg that will be treated the same as my current project TIFFs if I reconnect the media that way? Will they be the same size so that my photos motion parameters won't get screwed up?

I don't think you can have an alpha channel with JPEG; do your TIFFs have transparency?

If so, try selecting one in the viewer and make a FCP freeze frame from that; if I recall, it should preserve the transparency but possibly give you a more manageable instance of it to work with on the timeline.

If you have a Mac then you have iPhoto, which should convert the TIFF's to JPEG for you. If not there are dozens of free digi-image editors that you can download and get your conversions done. Don't need no stinkin' PC to do your work. (laughs)

.
 

That's a good idea but my graphic work involves lots of map fly-arounds and zooms. If I made my graphics fit into the viewer and save them they'd lose their rez when I do major zooms on them. I can't believe they even sell a turnkey editing system with final cut studio and there's not any imaging application that's included with it. I was just gonna download the free trial of photoshop, but CS3 is only for Leopard OX, and I have Tiger. Too bad there's no free trial for CS2........but I know some people.............

.
 

I need to preserve my transparent backgrounds though. How do I go from getting a 125MB TIFF file with a transparent background into something more appropriately sized for FCP........without Photoshop.

I don't think you can. If you're doing most of your work on the Mac you can call Adobe and have them sell you a transfer-license to migrate your Windows products over the the Mac. You'll have to send them the original PC discs along with a letter of destruction and for a small fee they will send you Mac discs with new install codes. That's what I would do.

Did you try what I said, selecting a TIFF into the viewer, make a freeze frame, and using that instance on the timeline?

.
 

I would Ben but I do a lot of zooming into these photos and I need the quality to hold up throughout the zoom. What format besides TIFF can I use in Photoshop Elements to save an well-compressed image with an alpha channel? Elements saves most of the same files as Photoshop. What settings does it need to be in for FCP to treat it properly?

When you save as Photoshop aren't you left with a PSD? Because PSD is what you want...

Try exporting just the TIFF parts as ProRes quicktime movies and re-importing them into the FCP timeline, then do your export to compressor.

As an aside put me down in the camp of saving these files as self-contained, saves a lot of headache from re-rendering and there is virtually no quality loss using ProRes. You have so many more options once your file is completley rendered. Especially if you want to test different mpeg2 quality settings or need to create a quick web copy, etc..

.
 

Okay I bit the bullet and converted all of the TIFF files to PSD's, thus cutting this file sizes in half. They're still pretty large though. With ones that didn't require transparancies, I converted to JPEGs. I'll update this posts with the results. One last question though. I have several nested sequences, and many clips have heavy filters applied etc. Should I render the underlying sequence AND the parent sequence before I export. If I understand it correctly, they're absolutely no need to render prior to exporting from the timeline (I've done it before), but is there any advantage of doing it?

Qood question about the rendering, been wondering that myself...

...let us know how the PSDs work out for Compressor.

.
 

Okay that didn't work. Compressor is choking on my very last sequence with a "Quicktime Error 0". I'm converting all of my graphic animations into their own movies, exported via "Animation" Millions of Colors+ (the + preserves the alpha channels). We'll see how this works out. It will be a drastic reduation in file size for sure.

Solution!!!!!!!!!
 

Okay fellas,

I was able to develope a work-around, and am posting this in case someone else might ever have the same problem. I recompressed all of my extremely large TIFFs that were not alpha dependent into JPEGs. This did not work, as Compressor choked on the large image files. I could render them and play them fine in the timeline of FCP, but compressor just couldn't handle it. Sometimes Compressor would leave out layers, and sometimes it would drop all layers to black. This was random, and after 6 tries, I resulted to othe rmethods. Note: This was only when I exported directly from the timeline. I could have exported a quicktime self-contained or reference movie and it would have worked fine, but direct from the timeline is THE BEST quality (even though so many seem to disagree).

To keep compressor from dropping those large image files, I copied all of the image animations to thier own separate timelines, and exported them with this setting:

Animation, Best, Millions of Colors+, Keyframe-All Frames.

Then I re-imported those Quicktime movies into FCP, and replaced the old image layers with their own single quicktime file.

I think what was happening, was that in an ordinary movie sequence, FCP looks at eack movie frame as a small file compressed in the DV codec, etc. With extremely large image sequence files (20MB+), FCP must treat that entire image file on a frame-by-frame basis. So, essentially if you added motion to a 30MB graphic for 10 seconds, it would be equivelant to a 30MB X 30 Sec X 30 Frames = 27,000MB, 10 sec. movie file!!!! My guess is that most applications would struggle with a bunch of 10 second, 27,000MB+ movie files. And that's with only one image layer. I was compositing up to five images of that size onto a single layer. That's a whole lot of processing, and when you send a 2 hr sequence like that into Compressor, your just asking for an accident somewhere along the encode process. That's why I think the drop-outs were random during each encode.