Sign in with
Sign up | Sign in
Your question

TG Daily Video: ABS demos quad-SLI triple monitor solution

Last response: in Graphics & Displays
Share
May 4, 2006 10:12:29 PM

Westlake Village (CA) - Only large screens can take advantage of the pixel horsepower of quad-SLI systems today. But those 24" or 30" monitors are everything but cheap and may not be an option for every user who already paid more than $5000 for a computer. An alternative could be ABS Computer Technologies' triple screen solution based on the Matrox TripleHead2Go.

Traditionally TG Publishing hasn't done much video content, although we do have a smattering of clips on our obscure video section - http://www.tomshardware.com/site/videos/index.html

We have purchased a whole bunch of new gear which includes the awesome Panasonic DVX-100B. You can expect to see many more videos in the future... hint hint, next week is E3.

What do you guys think about offering both WMV and H264 versions of videos? How about the general quality and production value?
May 4, 2006 10:37:47 PM

"Page Not Found" on both of them :) 
May 4, 2006 11:01:05 PM

Both links work for me also.. 4:00 PM PST
Related resources
a b U Graphics card
a b C Monitor
May 4, 2006 11:21:11 PM

Quote:

What do you guys think about offering both WMV and H264 versions of videos?


Good options so you can watch them at work if you don't have anything by plain WMP.

Quote:
How about the general quality and production value?




Making more mention of the Matrox part of the equation would seem to have made sense, and there is multi-monitor on SLi solutions, just not a performance benifit from running the 3 monitor seperately compred to having the SLi render a single wider FOV image.

Rough transistions, very basic not much tech detail, almost like a mini infommercial as the 'interview' seems very scripted. But though seeming somewhat scripted, I found it funny when the cameraman zoomed in on the outside of the Tehermaltake case to reveal really little more than the wide shot did, removing the side panel before starting or making it easy to remove mid-interview would've made the descriptions more vivid instead of trying to picture in your mind's eye what he's talking about.

Visually you kept the camera pretty still which is better than most, but the stark title bars were very harsh IMO. Speaking of harsh that intro music was bad, and too loud for too long, quick fade to quiet is a great idea since none of the audience came to listen to the music, also something less strident is good. Of course music is problem due to copyright reasons, etc, but heavy beats is hard for something where you're going to have less than ideal vocal audio following it. You don't want people lowering their volume, and then they can't hear the interview.

Just some quick feedback.
May 4, 2006 11:33:25 PM

"Making more mention of the Matrox part of the equation would seem to have made sense, and there is multi-monitor on SLi solutions, just not a performance benifit from running the 3 monitor seperately compred to having the SLi render a single wider FOV image."

Actually there is a good reason for that. We had originally intended the video to go up as part of the SLI article, but the content lended itself more for the Matrox one. Hence, it's about 1/3 talking about SLI and 2/3 talking about Matrox.

"Rough transistions, very basic not much tech detail, almost like a mini infommercial as the 'interview' seems very scripted. But though seeming somewhat scripted, I found it funny when the cameraman zoomed in on the outside of the Tehermaltake case to reveal really little more than the wide shot did, removing the side panel before starting or making it easy to remove mid-interview would've made the descriptions more vivid instead of trying to picture in your mind's eye what he's talking about."

Good point about removing the side panel. Definitely something to keep in mind for the future. By rough transistion do you mean content wise? The scripted feel probably comes from the participants feeling uneasy in front of the camera. You should have seen our original source video and outtakes!

"Visually you kept the camera pretty still which is better than most, but the stark title bars were very harsh IMO. Speaking of harsh that intro music was bad, and too loud for too long, quick fade to quiet is a great idea since none of the audience came to listen to the music, also something less strident is good. Of course music is problem due to copyright reasons, etc, but heavy beats is hard for something where you're going to have less than ideal vocal audio following it. You don't want people lowering their volume, and then they can't hear the interview. "

We used a tripod and stayed with zooms and slow pans (according to Cinematic panning ratios to reduce strobing). The intro music was definitely chosen for copyright reasons. I think we'll stick with it for the intro and use something else for the background during the main part of the video. There were times when the background music overpowered the vocals. (even after lowering the track to -25 db in Vegas).

Overall we wanted to convey the "coolness" of using a triple-head display in quad-SLI mode. I think the video does that much better than any picture can.

Yes it does seem sorta infomercial-like, but our source video was about 40 minutes long. It's always a balancing act deciding what to keep and what to cut out. I'm not sure if someone wants to download and watch that long of a video.
a b U Graphics card
a b C Monitor
May 5, 2006 12:28:12 AM

Quote:

By rough transistion do you mean content wise?


You could see quick cutaways including a quick 1 second insert during the FartCry segment, als the transitions were quick but similar setting, thus making them more noticeable, but had it been a cut away to a reaction shot then back you can edit it however you want and then people don't notice it as much. The most obvious ones were one moment guy wearing suit jacket, next red shirt. That happened twice and was of course very noticeabe. @ things that help this are like I said reaction shots (usually done after the interview) and also things like close-ups of screens, parts, pre-format product info, etc. All of that makes editing transitions seem seemless because it's no like the people in the scene look like they've teleported 6 inches suddenly. IT's a minor issue but it is something that makes the interview look like a home video instead of even cable access. I'm sure with more time you'll learn more of what works and what looks/sounds strange.

Quote:
The scripted feel probably comes from the participants feeling uneasy in front of the camera. You should have seen our original source video and outtakes!


Yeah and that kinda translates into the video. The thing is to have them be prepared. Because being nervous makes it look like they're memorizing a script or that they are simply uncomfortable talking. Best way to interview is to have a pre interview, present them your main issues you'll be talking about and that way they are comfortable answering and you're comfortable with how the interview is progressing and comfortable with a laid back question session, so that if you miss a step you're comfrtable enough knowing you can re-edit or even just come back to it. Once again experience helps here.

Quote:
We used a tripod and stayed with zooms and slow pans (according to Cinematic panning ratios to reduce strobing).


Yeah which is good, the hardest thing for cameramen to learn is to go slow be steady. Fast movement and especially jitters are terrible and amplify very quickly on screen, but you guys did a good job with that. Remember that the wider the shot the less impact any jitter has.

Quote:
The intro music was definitely chosen for copyright reasons. I think we'll stick with it for the intro and use something else for the background during the main part of the video.


If you stick with the music then I'd suggest 3/4 volume. You want your interview to always be louder than music, especially the intro music.

Quote:
There were times when the background music overpowered the vocals. (even after lowering the track to -25 db in Vegas).


And that will happen, you can't have your level too high for recording the interview due to distortion and noise, audio's tricky, but when the overlay audio is what's giving you problems, then that's completely in your control.

Quote:
Overall we wanted to convey the "coolness" of using a triple-head display in quad-SLI mode. Yes it does seem sorta infomercial-like, but our source video was about 40 minutes long. I'm not sure if someone wants to download and watch that long of a video.


No of course not, but if it's going to be brief, then the intro, 'cool-company' stuff needs to be quick and to the point. We're talking with... of... which is a cutting edge company doing... , hey mr X shows us what cool stuff you're working with....

Then focus on the guts of what you're trying to convey. Think about what you were trying to expose and what you think the viewer wants and needs to know. You skip a description of Quad SLi but the guy you're interviewing takes 5-10 seconds to eplain that the black cable is source and the 3 blue ones are for the display. Once again, when you get a more critical eye you'll notice these things. It's not just about what you cut out it's what you leave. Auto shows are the worst for that cause "everything's cool", but really it's a question of dividing up the value of the time. And for this segment I think you're going to have to think, what are we talking about or investigating that really a) comes across strong in a video format b) NEEDs video to really get the point across that we can't do in our typical static pieces. What would have been nice but maybe undoable depending on what's in that area, but compare a Quad SLi running a time demo in WideFOV triple monitor versus Single card in either single monitor or two tri-monitor setups side by side. The video of 'hey surround gaming is cool' is cool, but would take 10-15 seconds, and also hase been done from the Parhelia era. Showing the SLi green load balancing lines or dynamic task switching in the desktop, etc. all of those would play well to video's strengths more so than just turning a still review into a moving interview.

If Epic, Bethesda, Valve, Id are doing an interview seeing their faces the whole time doesn't add much over a quote by quote interview, but some nice demos of the technology does add alot to an interview, and if time is limited have a transcipt of information for the nitty gritty, and then a small clip of the 'good stuff'.

The more you do and the more feedback people give you the more you'll find what works for you guys. This is just constructive criticism as someone who's been there, done that.
May 5, 2006 1:02:02 AM

Quote:

What do you guys think about offering both WMV and H264 versions of videos? How about the general quality and production value?


I love the quality of the H264! - How was it encoded?

BTW, one does not need a massive rig to play the H264. I have a Sempron 2200 box with a 2 year old NVidia card running Mandrake Linux here - xine barfs, but mplayer plays it absolutely smooth. Also, why is it ZIPed - the 2M of size savings are less than 1%?
May 5, 2006 2:51:06 AM

The videos were edited and encoded on Sony Vegas 6. Version 6.0d has the Sony AVC/H.264 codec and our bitrate settings were 1mbps for the audio and 32khz/22kbit for the audio.

The downside of H.264 is the encoding time. Even on a beefy laptop it's a 4 to 1 ratio of rendering versus running time. 4 minutes to render 1 minute of video. Once we get the video production quality squared away, we will look into buying a hardware H.264 encoder (cards that plug in).

Incidently the WMV file takes about 1/2 the time to render. It is variable-bit rate and 1/4 the total number of pixels per frame.

On our inhouse machines, we've seen Quicktime 7 H.264 playback go up to 30-40% processor utilization on our Dual-core Pentiums. What you see in mplayer is probably a testament to how efficient Linux may do things versus Windows/Quicktime. But that is something we definitely need to look into for a later article.

The reason for zipping the file is not to save space. It's so the players don't stream the video from our server. When a player sees a video file it also tries to download other files from the server (index, captioning, etc etc). This is a tremendous hit.
May 5, 2006 10:00:24 PM

Here is some interesting information. It seems the H264 version was download three times more often than the WMV one. I had originally thought H264 would be downloaded far less because of its larger size.

I won't give you the exact numbers of the total downloads because that sorta sensitive, but it looks pretty good on our end.

This brings up a question.. a few years ago you really wouldn't expect people to download anything above ten megabytes or so (for videos). It seems that magic number has changed.. what would your tolerance level be on a download? 20 MB, 30 MB, or more?

What if we had a video of just booth babes from E3?? What would the limit be for that?
!