Right now I'm using CS3 to do straight forward 2-camera multicam edits using SD DV files. After the multicam edits, I'll insert dissolve transitions, nest the sequence in another sequence to trim, add titles, chapter markers, etc. The typical timeline is 90 minutes.

I'm running this on 64-bit Win7, i7-2600 SandyBridge, with 16GB RAM and several SATA 2 & 3 drives for OS, Programs, Scratch, AVI files, etc. It runs really smooth. Scrubs without any lag, and AME exports a 90 minute video DVD mpeg for Encore in about 20 minutes.

Now I'm considering moving to a HD camera, and it outputs in the AVCHD format. I know CS5.5 will handle it natively, but what can I expect for *multicam* performance using the above scenario?

I've read that one should use a fairly high end system to natively edit AVCHD (which I have), but I haven't read much about multicam editing of AVCHD and how it performs when there are lots of transitions, titles, chapters, and nested sequences.

I don't want to make the investment to upgrade to CS5.5 just to find my system sluggish editing the multiple AVCHD files. I would then be looking at having to spend more to upgrade my system even further, or using an intermediate codec like Cineform. My thought is, if I'm probably going to have to use Cineform anyway, then I might as well just stay with CS3.

Thanks in advance.

-Dave-

A PC such as yours will sail through a 4-way AVCHD multicam edit with no problems, as long as you have a CUDA-enabled NVIDIA graphics card, preferably one that is approved by Adobe for CS5.

I regularly do multicam edits with 3 AVCHD and 1 HDV video tracks, plus transitions and colour correction and up to 8 audio tracks on an i7-920 with 12 GB of RAM and it renders the edit to an HD monitor in real time.

You will not need Cineform for multicam work.

I agree with Tony about your system being fine for a 2 camera AVCHD-multicam edit under CS 5.5. With your system, you will likely find a two-camera AVCHD multi-cam edit easier and faster under PPro CS 5.5 than running a Cineform conversion and editing under CS3. Do be aware, however, that switching to AVCHD cameras will significantly increase the time for coding to DVD format mpeg. This is simply because there is so much more data that has to be handled with AVCHD than with SD.

Let me be more specific.

As Tony notes, we are assuming your system includes a CUDA-based video card that supports hardware MPE. Additionally, I am assuming that: a) that you edit using PPro's AVCHD pre-sets; (b) that you are not trying to edit the newer 1080/60p formats (with which I have no current experience and -- at least in my copy 5.5 --- PPro does not yet seem to support 60p with LPCM audio); and (c) assuming that you left off the "k" in describing your I7/2600 (I suspect that a plain 2600 would probably be adequate for what you want to do but have to say that I just do not know.)

This is not to say that Cineform might not be useful to you, but I think you would find that a two-camera mulit-cam AVCHD edit on your system will run better under CS 5.5 than with Cineform conversions under CS3. Simply put, the processing power in your system far exceeds what CS 3 uses.

Having said that, I also should say that I personally have found Cineform Neo useful under CS 5 and 5.5 when running more than two multi-cam tracks and when I need to do color-matching adjustments between cameras. Like Tony, I often have projects with numerous tracks. On some of my theatrical shoots, I have run three or three or four streams of avchd and three or four streams of HDV. I've had a few large weddings where I have had to do something similar, as well.

I am editing on an I7/950 and 12 gb of RAM with an older GTX260 card plus a Matrox MXO-2 mini feeding an HDMI display to calibrated LCD monitor. I also have multiple RAIDS and additional hard drives in the system. In theory, an I7/2600k with a CUDA card and optimized drives should measurably surpass my system's performance. There have been numbers of posts on this subject by Randall Leong, among others. Do a search on his name here if you want more background. Also, have a look at Harm Milaard's PPro benchmarking website for PPBM 5, if you want further background. Harm posts here frequently and has given much good advice about optimizing systems, and a search of his name will turn up much useful info.

What I've found with my system --- comparable to Tony's --- is that editing with heavy multi-cam loads may result in the system starting to bogging down when sequences get to about 15 or 20 minutes in length. (Again, when i say "heavy" I'm talking about a lot more than two camera and audio tracks). Running my AVCHD footage through Cineform Neo pushes that out to about 30 minutes. Thus, for larger multi-cam projects --- when I have sufficient editing time --- I like to run everything through Cineform NEO and use Cineform's First Light application for color matching the cameras. I like First Light because it works in meta data. This means it can be run in parallel with CS 5/5.5 which allows for near instantaneous comparisons and adjustments to the footage being used in CS 5/5.5 and no performance hits on playback.

Mind, I do not always need to do much color matching and sometimes I do not have enough time for Cineform conversions. Sometimes my pre-show manual white balancing work pays off. Sometimes the magic is not sufficient. A month ago, I shot an Advent concert in a dimly lit church. Three of the cameras matched perfectly but the fourth --- the one I used for close-ups -- was in just enough of a different location that its video had an overly reddish cast to it and I had to adjust it in post. I cases like that, I have found that doing much color work on AVCHD files with PPro's own filters and effects can cause noticeable slowdowns in playback. Admittedly, in my case, at least some of the slow-downs seem to be caused by the MXO2 which can momentarily lag when switching between the source, time-line and multi-cam monitors.

Even so, I have had pretty good luck even with larger non-Cineform projects as long as I can work in shorter sequences and then assemble them to a final timeline . For one recent example, I shot a school Christmas concert that needed to be edited and delivered in four days so the DVDs could be distributed in time for Christmas. The deadlines did not leave me sufficient time to do Cienform conversions of all the footage. I needed some color matching. The concert was in a large space with a stage for choirs at one end of the floor and a stage for bands at the other. Each band and choir segment ran about 15 or 20 minutes. I set up two fixed cams for each stage and had two other cameras positioned in a balcony where I could control them (that is, a total of six cameras, with the ability to have four aimed at any given event.) With so many cams, it was ineveitable that some would match well and some would not. For editing, I made separate sequences for each event (i.e., one for the first band, another for the first choir, etc.) I had to adjust some tracks for color matching using PPro CS5.5 tools but the format of the concert lent itself to editing in shorter sequences. I was able to proceed smoothly with playbacks and multi-cam editing and get the project edited together. I used frequent dissolves and other transisitions for some sequendces. I added titles (every kid has to have his or her name in lights!). Each song had to be a chapter. It all worked smoothly and passed on to DVD in a few hours of editing. (Well, there were the momentary pauses when switching between PPro's source-timeline-mulitcam monitors, but that, as I noted above, seemed to be an artifact of the latest drivers for the MXO2 Mini rather than a PPro issue.)

On the other end of the spectrum, I just did a quick multi-cam project for which I shot an interview using two AVCHD cams with the audio mixed into one of the cameras. This 40 minute edit went through my system without a hitch using only the original two track that was nested and edited in a single multi-cam sequence. No slowdowns, hitches, or whatever.

Therefore, I think you should have no trouble with two-camera track mulit-cam edits on your system when using CS 5.5.

The speed of your AME exports may be a different matter. This is not a problem with CS 5.5 or your system but, rather, the processing time required in down-rezzing from AVCHD to standard DVD Mpeg2. On my system, I usually feed the timeline to Encore, build my DVD, and code from there (which also invokes AME). Coding an hour of AVCHD using two-pass high bit-bitrate VBR settings seems to take roughly an hour on my system for an hour of video. When I start with standard def video, the coding speed seems to be much faster. Lately, most of my SD coding has been for customers who want old VHS, 8 mm. and Hi8 family videos put on DVD. I have not timed anything exactly, but my system seems to code 1½ hours of such video to DVD in about 20 to 25 minutes. I say "seem," because I usually do this as part of building a disk image for making copies so I can only estimate. At any rate, it seems a lot faster than what I got with CS3/AME3. It should be faster since CS5.5 apps take advantage of 64 bit processing and multi-threading. I should also note here that AME 5/5.5 seems, to my eye, to do a better job of coding that I got back when i used CS3. Oterhs have commented on this here and you can search on that subject if you want.

Hope the foregoing provides you with the kind of detail you wanted.

Wow, Jay (and everyone else)... Thanks for the detailed reply. It's just the info I was looking for, and I hope that those who come after me searching for similar information will benefit from your insights.

The decision I now have is to just upgrade Premiere or upgrade the whole suite? I used to use Photoshop a lot for the photography side, but now I do most of my work in Lightroom, and seldom go into PS. But I do edit Encore menus in PS, but I assume that Photoshop CS3 can edit Encore CS5.5 menus just fine.

Thanks again for all the help. I really appreciate it.

"But I do edit Encore menus in PS, but I assume that Photoshop CS3 can edit Encore CS5.5 menus just fine."

I assume you mean that you've built a menu in Encore (maybe using one of the templates) and then sent it out to Photoshop for further work. In a pinch, I have edited an Encore CS5 menu on a laptop using Photoshop Express version 5. That version of PsE has to be at least six years old but it worked fine. It is possible (though I strongly doubt it) that CS 5.5 menus might have some additional coding that would not be recognized by an older version of Photoshop or PsE. I have not tried this with Encore CS 5.5, so I cannot say for 100% sure that it will work without glitches. As far as I know, all the Encore and Premiere menu templates are just standard PSD files readable by any recent version of Photoshop or PsE. The only issues I have ever run into with this sort of thing is that some versions of PS and PsE recognize different sets of fonts than others.

I still use PsE version 5 because my scanner (Canoscan 8800f) does not have 64 bit drivers that work with the newer, 64 bit versions of PS. So, I generally run a scan under PsE, save the scan, and then open the saved PSD in PS 5. If you need to do much image editing, PS 5 has some very useful tools such as the "heal" function for blemished images. Personally, I have never needed those tools for constructing DVD menus, though. YMMV depending on what kinds of images you are using.

If you are constructing the menus in older versions of Photoshop, there should be no problems importing them into either PPro or Encore CS 5.5. There might be some minor glitches in using CS 5.5 dynamic linking functions with older PS versions, although I have yet to find one, myself. If you did have a glitch, it would be easy enough to simply save the image in PS and then import it as an asset into Encore rather than dynamic linking.