Making movie clips for E-Prime?

[David Vinson]

Thanks David for this detailed description!

I work with videos a lot and there are all kinds of possible pitfalls, 
so I thought I'd weigh in on a few of the issues while the topic is 
fresh in my mind.  It's incredibly important to develop dependable and 
consistent systems of creating stimuli that work for your hardware and 
software.

I very much agree with David's option: that wherever possible it's best 
to create video files with stock E-Prime in mind, rather than trying to 
coerce E-Prime into working with whatever video files you have. With the 
latter you may experience weeks of trouble (and repeated back-and-forth 
correspondence with E-Prime Support who may or may not be able to 
replicate your video problems on their hardware!). The Codec Config tool 
is very useful in dealing with various video formats but it only goes so 
far - and files that pass the codec config test may still cause crashes 
when run for real.

Our test procedure when developing video experiments is something like 
this. At each step there's a chance for failure (but less if you follow 
the same steps as in a previous successful study)!
(Our experiments: we are mostly using clips of sign language stimuli 1-2 
seconds long, perhaps 150-200 per experiment, possibly repeated a few 
times in the course of a study).

1 - convert the very first video clip into the desired format/size etc.
(we prefer to display videos at their actual size rather than using 
E-Prime's stretch function), and run Codec Config to see if the clip 
passes this simple test.

2 - create the simplest possible program that displays that one video 
using a MovieDisplay.  If it works proceed to the next step.

3 - convert some more clips (15-20 should suffice, or just do them all 
in batch mode).  While you (or someone else) proceeds with the editing 
process, set up a "stress test" experiment that repeatedly displays 
these clips, one after another, with a reasonable interval between each 
(otherwise you may start getting build-up of loading lags and thoroughly 
unpleasant performance)
Make sure this experiment logs data (even if it's just non-responses). 
I might set it to 100 repetitions of a randomized List.  Then just set 
it off - if it crashes along the way (ie, doesn't get to the end and 
convert the running text log into an edat2 file) this is a sign you may 
have a harder-to-detect codec issue to sort out.  Better that it happens 
this way than with a real participant.

4 - if you get this far, set up your real experiment and test it 
completely under "realistic participant" conditions.  Keep in mind that 
the stress test in (3) may succeed but the real experiment may still 
crash for one reason or another.

In addition to making sure your experiment does not crash, if timing is 
an important consideration (and when is it not?), keep in mind that it 
takes significant time to load and display video files. There are a 
number of additional time audit features specific to MovieDisplays - 
it's very wise to log these and get your timing sorted out before you 
start running.  Timing can be a real mess when you deal with video 
stimuli and seems to depend a lot on the hardware.

For all of this, if you run into real problems, I strongly urge you to 
open a support request with E-Prime support (even though you may want a 
solution "today" and the support queue may be 10-12 days). Video 
problems can be so complicated and idiosyncratic, and there may be known 
bugs or still-undocumented features that could help you out a great 
deal. I've had very good responses from support staff over the years, 
but patience is required especially during the peak periods (eg start of 
academic terms).

Whew, that's enough about my life story with video problems. When it 
comes to video capture and conversion - our main setup for generating 
E-Prime video from scratch involves initial processing on Apple hardware 
(Final Cut Pro) from a range of DV cameras (our studio/video hardware is 
apple-based), although I still use VirtualDub on some occasions.

Our biggest problems have come from videos that have gone through 
quicktime formats (ugh!) and my advice is always to minimize the number 
of conversions if you want an E-Prime-friendly video type - and stick to 
a standard process using the same camera/settings/etc once you get 
something that works - different cameras have different default settings 
and it's best to stick with something that works wherever possible. And 
don't ditch your original video files (even if they're huge) until you 
know you have a conversion that works.

Many researchers will work hard to develop video stimulus materials that 
display as well as possible in a media player, but this does not always 
transfer to clean E-prime performance - if you know you are going to use 
E-Prime with your videos, it's a good idea to get them E-Prime-friendly 
first (really this is just as important a part of materials development 
as other experimental controls you include).  If your experiment doesn't 
work there's no point in having nice presentation quality video stimuli.

hope this is useful to somebody!
-dv

On 12/07/2011 22:24, David McFarlane wrote:
> OK, we got something to work, so here is my report in case anybody else
> can use this. I would be very interested if others could weigh in with
> what they have used.
>
> First, turns out my user really wanted to extract short clips (~1 min)
> from commercial movie DVDs. In short, I managed to use AoA DVD Ripper
> (http://www.aoamedia.com/dvd_ripper.htm ) to visually select a segment
> to clip from a commercial DVD, set DVD Ripper to encode it as MPEG-1/VCD
> (*.mpg), and the resulting file played in stock E-Prime with no further
> fuss. (That will work fine as long as the DVDs that the user wants do
> not foil us with some futher copy protection, and assuming that our use
> falls within Fair Use.)
>
> More details for those who care...
>
> In general, we have two strategies we might apply here -- configure
> E-Prime to work with movie files as we supply them, or make sure that
> the movie files we supply are configured to work with stock E-Prime. I
> prefer the latter strategy, because (1) I figure that the E-Prime
> developers have optimized E-Prime for a limited range of formats even if
> we can trick it up to accept other formats, and (2) I would rather not
> have to reconfigure E-Prime with custom codecs every time we move the
> experiment to a new machine.
>
> So first I used GSpot (http://www.headbands.com/gspot/ ) and MediaInfo
> (http://mediainfo.sourceforge.net/en )to see what format/codec E-Prime
> uses for its own examples. They use .mpg files, encoded as MPEG Video
> (Version 1). So I set that as my goal.
>
> I tried two products for ripping the DVD samples. First I tried
> HandBrake (http://handbrake.fr ), but (1) the interface does not seem to
> have a good visual way to select arbitrary clips, (2) it was *very* slow
> (we canceled before it finished), and (3) HandBrake itself insists that
> it is only a transcoder, *not* a DVD ripper.
>
> After Googling around a bit I moved on to AoA DVD Ripper. This provides
> a good-enough visual interface for selecting arbitrary clips, worked
> pretty fast, and produced an .avi file that played in Media Player. That
> file did not play in stock E-Prime, however, but once I reconfigured DVD
> Ripper to output MPEG-1 all was well. (I also set File Split Mode to
> Infinite just to avoid splitting files.) Note that by default DVD Ripper
> outputs .avi at 720x480 resolution, or .mpg at 352x240 resolution, so
> you might want to fiddle with that further. (It costs US$40 (cheap!) to
> register DVD Ripper, and I do not know how the paid version differs from
> the free one, but if you find the program useful then please do the
> right thing and pay for it.)
>
> Now as it turns out, when I installed AoA DVD Ripper it also installed
> an Xvid codec, so then the .avi file played in E-Prime as well. But the
> .mpg file continued to play in E-Prime even after I uninstalled the Xvid
> codec, so I feel safer sticking with .mpg (MPEG-1).
>
> For the record, if we did want to fiddle with codecs for E-Prime, in
> addition to letting AoA DVD Ripper install Xvid, PST recommends the
> ffdshow (http://www.free-codecs.com/download/ffdshow.htm ) and SUPER
> (http://www.erightsoft.com/SUPER.html ) codec libraries (see
> http://support.pstnet.com/forum/Topic635-12-1.aspx and
> http://www.pstnet.com/forum/Topic2986-5-1.aspx ).
>
> Finally, as mentioned earlier in this thread, for more extensive video
> editing VirtualDub (http://www.virtualdub.org ) might come into play.
> This outputs only in .avi format, so would require installing codecs for
> E-Prime, or using A0A DVD Ripper or HandBrake to transcode the output
> files as needed for use in E-Prime.
>
> I still have no idea what system to use in case we ever want to generate
> video for E-Prime from scratch, but I guess we will cross that bridge
> when we come to it.
>
> -- David McFarlane, Professional Faultfinder
>
>
> At 7/11/2011 04:19 PM Monday, David McFarlane wrote:
>> Telephoning and Googling around trying to answer my own question #1...
>>
>> First general-purpose answer I got was Windows Movie Maker, which
>> already comes with recent versions of MS Windows. Have any of you
>> found success with that?
>>
>> A more serious source recommends Final Cut Pro (or perhaps the less
>> expensive Final Cut Express). Anyone have any experience with that? It
>> will render in a host of codecs, so it should do a good job of
>> rendering clips for E-Prime. But it is a Mac-only product.
>>
>> So I looked for Windows alternatives to FCP, and came up with Adobe
>> Premiere, or Sony Vegas. Can anybody tell me something about these
>> products?
>>
>> Thanks again,
>> -- David McFarlane, Professional Faultfinder
>>
>>
>> At 7/11/2011 03:36 PM Monday, David McFarlane wrote:
>>> Well fellow E-Prime mavens, now I could use your experience & advice.
>>> I first checked in the New Features Guide, and did a cursory search
>>> through the PST Knowledge Base, the PST Forum, and the E-Prime Google
>>> Group, and did not find this addressed anywhere, so I hope I have
>>> done my homework at least as well as I expect others to :).
>>>
>>> Two related questions:
>>>
>>> 1) Suppose we want to make some short clips from a larger, existing
>>> movie file for use in E-Prime. What software would you recommend for
>>> this?
>>>
>>> 2) Suppose we want to record our own movie clips from scratch for use
>>> in E-Prime. What systems would you recommend for that?
>>>
>>> I might also reframe those questions as, "What did *you* use?" or,
>>> "What worked for *you*?"
>>>
>>> Note that the KB and online discussions have addressed at length how
>>> to get *existing* clips to work with E-Prime (installing codec
>>> libraries, etc.). I instead want to avoid those problems by preparing
>>> our clips in the first place in a way that plays well with E-Prime
>>> "out-of-the-box". Any advice?
>>>
>>> Thanks,
>>> -- David McFarlane, Professional Faultfinder
>


-- 
David Vinson, Ph.D.
Senior Postdoctoral Researcher
Cognitive, Perceptual and Brain Sciences Research Department
University College London
26 Bedford Way, London WC1H 0AP
Tel +44 (0)20 7679 5311  (UCL internal ext. 25311)

-- 
You received this message because you are subscribed to the Google Groups "E-Prime" group.
To post to this group, send email to e-prime at googlegroups.com.
To unsubscribe from this group, send email to e-prime+unsubscribe at googlegroups.com.
For more options, visit this group at http://groups.google.com/group/e-prime?hl=en.