Thanks Bob,
 
I'm still waiting to hear back from them (Google) on related stuff (mostly concerning bulk uploading).  I know they prefer that I upload the full-res version, even though they shrink it down when streaming.  I'd actually rather them have the full-res versions, so that as future internet bandwidth paradigms allow higher-res streaming, they can automatically step it up without input from me.  Of course, uploading a 178MB file compared with a 3.5MB file -- multiplied several thousand times -- is another factor that cannot be completely ignored.
 
The ideal would be a JPEG2K-ish standard as you describe with user-specified resolution.  Rather than get Google to build another stand-alone app, I'd rather see them come out with their own browser with built-in support for Earth (and customized/optimized video streaming, Google Base access, Co-op features, etc.) all built in.
 
So far, no NDAs....
 
Aloha,
Rich


From: Bob Morris [mailto:morris.bob@gmail.com]
Sent: Tuesday, August 28, 2007 5:34 PM
To: Richard Pyle
Cc: Eamonn O Tuama; Timothy M. Jones; tdwg@lists.tdwg.org
Subject: Re: [tdwg] Species pages and video

JPEG2000 video standards---which are generally mp4--- probably address this problem, because JPEG2K can decompress at arbitrary resolution, that is you can tell the remote server what resolution you want the stuff sent at. I think, but am not certain, that this would be the case for the video standards too, since it should be doable frame at a time. Indeed, I vaguely recall a demo at a JPEG2K meeting in which video was streamed at resolutions which varied with time. There are lots of questions, but apparently few answers, of the form "does Flash support JPEG2000.  Of course rendering in the current browsers remains a problem if it doesn't, though people certainly tolerate standalone Google Earth, for example---so why wouldn't they tolerate standalone video viewers. Ask your Google video pals what's up. Preferably not under an NDA. :-)

Bob


On 8/28/07, Richard Pyle <deepreef@bishopmuseum.org> wrote:

Dear all,

I've been meaning to jump in on this conversation several times, but I keep
getting side-tracked.

For over two years now, we have been developing a protocol and associated
software tools (I use the pronoun "we" loosely as far as the software
development goes -- that has been entirely the work of Rob Whitton) to allow
us to harness the power of video for our scientific purposes.  We conduct
surveys of coral-reef fishes in the Pacific, and the use of hi-definition
underwater video cameras allow us to make dozens of "video vouchers" (as we
call them) of fish species in the context of their natural habitat on every
single dive (again, I use the pronoun "we" loosely, as John Earle is the
primary videographer on our surveys).  Though perhaps not as ideal as
specimens, the video is much better than in-situ still photos (especially at
hi-def resolution), because it gives us multiple angles on the subject
(increasing the probability of capturing that elusive but diagnostic small
black spot near the anus), as well as behavior (which can sometimes aid in
confirming identifications).  And it's a LOT better than just an un-imaged
observation record. It also allows us to document many more species on a
given dive than we could by collecting alone.

The software that Rob Whitton has developed is optimized for field-based
capturing of metadata.  We ( i.e., John) will generally catalog the video
clips on the same day the video was taken.  Metadata is robust, with full
locality/habitat data (including depth and other parameters), as well as
rich content cataloging (multiple identifications of the same imaged
organism, etc.)  At the moment, we (i.e., John & Rob) have something on the
order of 7,000 video clips cataloged -- representing nearly a terabyte of
video files (a mix of both standard-resolution DV and HDV).  Very soon we
will have an initial website online to allow searching/etc., and we have a
couple of major regional checklists in the works that will cite these "video
vouchers" in addition to more traditional means of documenting species at
localities.

So...the reason I am posting this now (rather than wait until the site is
online) is to ask others who are exploring the use of video content for
similar purposes how they plan to implement it.

Our current plan is to maintain an archive of full-resolution digital video
files on our local SAN, but the files are much too large to stream in
real-time over almost any typical internet connection, and moreover would
completely choke our bandwidth if the site ever became popular.  For this
reason, we want to use a video hosting service to stream the content, which
we will link to from our own web databases (which themselves will serve only
keyframes from the clips).

We've been working with Google to sort out a way to do batch uploads onto
Google Video.  I generally prefer the Google Video environment over YouTube,
but I'm not familiar with other video hosting services that are out there.

Here is a sample clip:

http://video.google.com/videoplay?docid=153611051098248174

Google Video allows me to dump all of the metadata into the Description
field.  Unfortunately, this is not very structured.  However, Google Video
allows you to link back to your own web page for each clip, so I can have
that link go to an LSID resolver, or some other web interface where more
structured metadata can be served.  Another feature I like is that you can
lay as many subtitle/caption files as you want.  For example, if you go to
the link above, in the lower right corner you'll see a little "CC" icon.
Click on the drop-down button to the right of the "CC" button, and you can
choose from any number of subtitle tracks.  In the example above, there are
two different tracks: "Audio Dialog" transcribes the spoken words you hear
on the clip's soundtrack, and "Species List", which names the species as
they appear in the clip.  Rob Whitton is developing his software to
automatically generate the text for the metadata and multiple CC tracks, so
that we can (eventually) automate the upload process.

The main problem -- which I think will be true of any of these video hosting
services -- is the limited resoloution of the clips as they are streamed.
For example, here is a frame from the original HDV clip in the above link:

http://www2.bishopmuseum.org/testvideo/Frame01.jpg

Here is the same frame at the resolution that the video is rendered on
Google Video:

http://www2.bishopmuseum.org/testvideo/Frame02.jpg

Obviously, the full-resolution video contains a LOT more information.  The
problem is that an MPG (i.e., compressed) copy of the full-resolution HDV
clip is 172MB, whereas the compressed version that Google streams is 3.5MB.
The problem is not with Google Video -- it's with the internet.  Most people
will not have access to the badwidth necessary to stream video at the full
HDV resolution.

So...what we'd like is a service that will allow people to view the clips at
a resolution that is reasonable to stream over the internet (Google Video,
YouTube, etc.), but then have the option of downloading the full-resolution
file (in this case, 178MB) if they want to see it on their own computer, and
are willing to wait for the full download.  Obviously, we'll have to somehow
regulate the downloading so that we don't choke our bandwidth -- but we want
to allow people to have access to the full-resolution imagery.

My hope is that Google (or whoever) itself would offer the service of
streaming content at an appropriate resolution, but then allowing people to
download the full resolution clip as a file, if they want (i.e., using
Google's or whoever's bandwidth, and not ours).  But for the time being, we
mostly see Google as a way to: 1) Manage streaming of video content at low
resolution, and 2) increase visibility (through Google searches_ of the
content we do have.

Of course, the latter depends heavily on how well the metadata are fleshed
out and structured -- which brings me back to Éamonn's post. Like him, I am
very-much looking forward to conversations at the upcoming meeting in
Bratislava.

Meanwhile, I guess the main point of this message is to ask whether others
know of analagous projects, and how they have dealt with issues of
bandwidth, bulk uploading to video hosting services, and metadata structure
and content.

Aloha,
Rich

Richard L. Pyle, PhD
Database Coordinator for Natural Sciences
  and Associate Zoologist in Ichthyology
Department of Natural Sciences, Bishop Museum
1525 Bernice St., Honolulu, HI 96817
Ph: (808)848-4115, Fax: (808)847-8252
email: deepreef@bishopmuseum.org
http://hbs.bishopmuseum.org/staff/pylerichard.html




> -----Original Message-----
> From: tdwg-bounces@lists.tdwg.org
> [mailto:tdwg-bounces@lists.tdwg.org] On Behalf Of Eamonn O Tuama
> Sent: Tuesday, August 28, 2007 5:20 AM
> To: 'Timothy M. Jones'; tdwg@lists.tdwg.org
> Subject: RE: [tdwg] Species pages and video
>
> Dear Timothy,
>
> I think the use of video is valuable even if there is not
> much motion involved - combining a series of stills with
> voice over can be very effective - and the many video hosting
> services makes it relatively easy to get online. However,
> unlike text which can be mined for information, video (and
> images) require good metadata to describe what the content is
> about - to aid in searches, etc.
>
> Your species pages with their general facts and interactive
> taxonomic keys span the task areas covered by SDD (Structure
> of Descriptive Data) and SPM (Species Profile Model) TDWG
> interest groups. I look forward to fruitful discussions
> between the two at the forthcoming meeting in Bratislava that
> will lead to standardised ways of marking up your species
> content so that it is more easily discoverable, accessible
> and re-usable (assuming permissions
> granted) across what GBIF has labelled "The Universal
> Biodiversity Data Bus".
>
> Best regards,
>
> Éamonn
>
> -----Original Message-----
> From: tdwg-bounces@lists.tdwg.org
> [mailto:tdwg-bounces@lists.tdwg.org] On Behalf Of Timothy M. Jones
> Sent: 10 August 2007 16:48
> To: tdwg@lists.tdwg.org
> Subject: [tdwg] Species pages and video
>
> Hello,
>
> I will not be attending the meeting this fall  but  thought
> that this may be of interest to those interested in species
> pages models.
> I am working on species pages that include the use of video.
> The videos were only added a month ago and are a bit
> rudimentary (with budget-conscious equipment) but the
> potential now seems truly limitless.
>
> Examples -
> http://utc.usu.edu/factsheets/CarexFSF/new/carex_eburnea_species.htm
>
> http://utc.usu.edu/factsheets/CarexFSF/new/carex_nebrascensis_
> species.htm
>
> http://utc.usu.edu/factsheets/CarexFSF/new/carex_mitchelliana_
> species.htm
>
> Comments appreciated,
>  Timothy M. Jones
> http://utc.usu.edu/keys/Carex/Carex.html
> _______________________________________________
> tdwg mailing list
> tdwg@lists.tdwg.org
> http://lists.tdwg.org/mailman/listinfo/tdwg
>
>
> _______________________________________________
> tdwg mailing list
> tdwg@lists.tdwg.org
> http://lists.tdwg.org/mailman/listinfo/tdwg


_______________________________________________
tdwg mailing list
tdwg@lists.tdwg.org
http://lists.tdwg.org/mailman/listinfo/tdwg



--
Robert A. Morris
Professor of Computer Science
UMASS-Boston
ram@cs.umb.edu
http://bdei.cs.umb.edu/
http://www.cs.umb.edu/~ram
http://www.cs.umb.edu/~ram/calendar.html
phone (+1)617 287 6466