Video quality needs improving

[James-WBF]

mjl5007,

This is some great information. I'll take a bit further with the video guys on these points.

So, just to throw this out there, if I wanted to swap out the DataVideo SE500 to something that supports wide screen format, what am I getting myself into? If I do not upgrade the projectors & screens(all 4:3) how would I successfully support that type of setup? I've head of scalers, but not looked deeply into what they would do for me.

Thanks again...

[Gracetech]

Well since you already have HD cameras wouldn't it make more sense to go HD?

If you wanted to pick up a Black Magic ATEM Television Studio and some really long HDMI cables you could definitely take things up a notch. Only issues would be getting the right converters for your connection to other devices post switch. The other issue is loosing usage of a percentage of your projection screens. With HD in 16:9 you will have letterbox on your projectors.

crt

Edit: Come to think of it i'm pretty sure the ATEM allows SD(4:3) footage through HDMI. Your cameras should have that as an option as well. This means that you could have a 4:3 work flow until you get new projectors/screens and then with a flip of a switch go HD(16:9). Just a thought

Yet another though occurs to me. The ATEM series has built in H.264 encoders which makes capture to file a breeze.

[James-WBF]

Hey Chad,

So the ATEM card is software driven to do the video switching. How well does that work compared to what I am using now (the functionality, that is, not the quality)?

Have you used this before? It looks interesting. I've got to get my mind wrapped around it.

I guess I would also need a laptop to run the software.

So the camera can output S-Video or Composite (no HDMI). So I would need something to convert to either HDMI or component (BNC) input. Which is what I THINK we do today, convert from composite to BNC with a basic connector.

On the Media-shout side, I am coming from VGA to S-Video converter. What would I want to do on that side? Leave it the same?

James

[James-WBF]

On another note, We changed the camera to a 4:3, and get bars on the top & bottom now in the DataVideo SE-500. Have not recorded yet to see how this will work for stetching/squeezing.

I've changed the fps to 24. We are de-interlacing as well (think htat was already set).

One other note. I asked if we have encoded FLV before, and told that the quality was terrible. Is there something we need to do differently or, is this just the reality of FLV?

Thanks,

James

[mjl5007]

A couple things...

BNC is a type of connector -- a method of mechanically connecting the end of a wire/cable to something else. It has nothing to do with a type of video signal, or the quality/format of the video contained in said signal. RCA and mini-DIN would be examples of other kinds of connectors.

Composite, S-Video, and component are different types of analog video signals. A composite signal contains both the light/dark (luminance/luma, Y) information as well as the color (chrominance/chroma, C) information, all in one signal. S-Video separates the luma and chroma into two separate signals, which is why it is sometimes referred to as YC or Y/C. The term 'component video' is really an ambiguous term, because any kind of video signal separated into two or more separate pieces ('components'... get it?) is considered a component video signal, so S-Video falls under that category as well. The correct technical term for what is referred to as component video in the common vernacular is 'YPbPr', where Y is the luma as before, and Pb and Pr represent the blue and red portions of the chroma signal, respectively (with minor technical details that aren't important). The green portion of the signal doesn't need to be transmitted because it can be accurately derived from the Y, Pb, and Pr signals.

The type of connector used on the ends of the cables has nothing to do with the signal transmitted over said cables (in general). While consumer equipment uses RCA connectors for composite and 'component' video and a 4-pin mini-DIN connector for S-Video, professional equipment generally uses BNC connectors because they're generally considered to be mechanically superior (e.g. they twist and lock on). Of course, there are simple mechanical adapters available between different types of connectors. Hence your 'basic connector' that 'converts' from composite to BNC isn't really converting anything; the signal is still a composite video signal, you're just using an *adapter* to connect a cable and a piece of equipment that are using two different types of connectors.

(If any of that sounds condescending, that's absolutely not my intent; just trying to clear up misconceptions and convey some knowledge!)

So, as far as the ATEM goes, the only inputs it has are SDI and HDMI, neither of which your camera can output. You would need some piece of hardware that can take in a composite or S-Video (or component, as the HDR-FX1 can output component, too) signal and output an SDI signal, which would then go into the ATEM. Note that there's an analog-to-digital conversion there (SDI is a digital signal), so you're not talking about a cheap piece of passive equipment; there are active electronics involved which you pay for.

Now, as for "chang[ing] the camera to a 4:3"... did you change to a different camera, or just change the settings on the HDR-FX1 you've been using? Either way, if what you got on the SE-500 was a letterboxed (black top and bottom) image and it wasn't distorted (squeezed or stretched), then something still wasn't configured correctly and the camera wasn't outputting a 4:3 image. Without actually having the same camera and being able to test it out myself, I'm 99% certain there should be a way to get the camera to either output a 16:9 image squeezed into a full 4:3 frame (no letter- or pillar-boxing), *or* output an undistorted 4:3 image in a full 4:3 frame.

[mjl5007]

Oops, meant to comment on the FLV question as well...

Originally Posted by James-WBF

One other note. I asked if we have encoded FLV before, and told that the quality was terrible. Is there something we need to do differently or, is this just the reality of FLV?

FLV is just a container format; it doesn't say anything about the format of the actual video (or audio) data, just how the video and audio data are packaged into a single file. The format of the video data itself (also called the compression or "codec" used) and the settings used are what determine the quality. The FLV container format supports several different video compression formats, including Sorenson Spark and VP6 (variants of the H.263 video standard), and more recently, H.264 (since Flash Player 9 Update 3). H.264 is also known as AVC, and is the same basic format used for the video streams on most Blu-ray discs, as well as the new lines of "AVC" and "AVCHD" cameras/camcorders.

What this means is that depending on what actual video format you were using when FLV was originally attempted, it's certainly possible that it looked bad if an older video format was being used (like Spark or VP6), the compression settings (bitrate, etc) were set incorrectly, or a combination of both, or other factors entirely. It's possible to get quite good quality video in an FLV file using an H.264-based video format with sufficient bitrate for a given video resolution.

[James-WBF]

Wow, I must have really gotten hit by the holidays, as for some reason I never responded to the last two postings. In the mean time, we did change a few things from the notes you folks let me

We are now doing f4v for the Flash, and with the H.264 (AVC) format, definitely looks decent. MP4 for our Apple viewing folks out there. and mp3 for audio only listeners.

Here is a sample of a video:
http://www.waterbrook.org/three-gift...r-18th-sermon/

Thanks again folks.

James

[KDW4Him]

Originally Posted by mjl5007

The issue here is that your camera records in 16:9, but you're outputting via composite and through the SE-500, which uses an NTSC signal at 4:3 only.

As I read this thread and all of the ideas this quote caught my attention.

MJL, I use an SE-500 and I thought I tried 16X9 though it successfully before so I tried it again today. We normally shoot 4:3 SD because we send the DVD's out for public access broadcast that is SD 4:3. We run composite vid into the SE-500 and send record out to a DVD recorder.

Today for the test I set one camera to 16X9 and one to 4X3 and recorded a few seconds of just switching between the two cameras. I then took the finalized DVD to a different DVD player hooked to a 16X9 TV and set the screen to display 16X9. On the 16X9 cam I had normal looking people in the shot but on the 4X3 cam I had short fat people displayed on the screen. If I set the screen to display 4X3 I had normal looking people in the 4X3 camera shot but tall skinny people displayed with the 16X9 camera shot.

I went a step further and ripped the DVD to QT for insert into Final Cut. One rip to 4X3 and one to 16X9, all other settings were the same. Same result in FCP as displayed on the TV in my test above, the 4X3 cam is correct in the 4X3 timeline and the 16X9 cam is correct in the 16X9 timeline.

This looks like the SE-500 is transparent to the aspect ratio being sent.

[KDW4Him]

Originally Posted by KDW4Him

As I read this thread and all of the ideas this quote caught my attention.

MJL, I use an SE-500 and I thought I tried 16X9 though it successfully before so I tried it again today. We normally shoot 4:3 SD because we send the DVD's out for public access broadcast that is SD 4:3. We run composite vid into the SE-500 and send record out to a DVD recorder.

Today for the test I set one camera to 16X9 and one to 4X3 and recorded a few seconds of just switching between the two cameras. I then took the finalized DVD to a different DVD player hooked to a 16X9 TV and set the screen to display 16X9. On the 16X9 cam I had normal looking people in the shot but on the 4X3 cam I had short fat people displayed on the screen. If I set the screen to display 4X3 I had normal looking people in the 4X3 camera shot but tall skinny people displayed with the 16X9 camera shot.

I went a step further and ripped the DVD to QT for insert into Final Cut. One rip to 4X3 and one to 16X9, all other settings were the same. Same result in FCP as displayed on the TV in my test above, the 4X3 cam is correct in the 4X3 timeline and the 16X9 cam is correct in the 16X9 timeline.

This looks like the SE-500 is transparent to the aspect ratio being sent.

Bump after the server issue

[mjl5007]

Originally Posted by KDW4Him

This looks like the SE-500 is transparent to the aspect ratio being sent.

To be more technically correct, I should've said that the SE-500 is "NTSC analog video"-only, not 4:3-only.

NTSC analog video doesn't really have the notion of 4:3 vs. 16:9 in the first place; it doesn't even really have a definite "image width", it's up to the display device as to how wide the final displayed image is. Think back to when NTSC was created, when digital televisions didn't exist, it was all CRTs. The way CRTs work is with the electron gun beam (or three beams in the case of color TVs) scanning each horizontal line across the screen with the intensity of the beam(s) varying continuously with the incoming analog video signal for that line, resulting in varying levels of light output from the phospors on the backside of the screen. Note that I said the beam varies *continuously* with the incoming *analog* signal -- there is no concept of a pixel or any discrete unit of width as there would be with a digital signal, only discrete lines which vary continuously along their width/length. Thus, the width of the image is determined strictly by the physical width of the scan lines. The fact that the ratio of width to height of CRT TV screens was 4:3 was not dictated by anything in the technical aspects of the video signal itself -- instead, it just so happened that the *images* carried within the NTSC signal happened to be in a 4:3 ratio, and so would look stretched/squished if displayed in any other way.

Enter digital video: without going too overboard with details, the conversion from an analog NTSC video signal (again, with no defined 'width' inherent in the signal itself) into a digital video signal (with discrete pixels) requires that the analog-to-digital conversion hardware sample the analog signal at regular intervals along the length of each scan line; how often this sampling occurs dictates how many total samples will be taken across the entire line, and thus how many pixels wide the resulting digital image will be.

Omitting a lot of steps in between, this eventually leads into how a DVD player can display widescreen video over an NTSC analog video output (composite, S-Video, SD component). The DVD player generates an NTSC analog signal, into which a widescreen *image* has been fit, but it is entirely up to the TV as to how to display that NTSC video signal (i.e. in 4:3 pillar-boxed mode or in 'wide' mode which takes up the entire screen). Thus, if you play a widescreen DVD from a DVD player hooked up to a widescreen TV via an NTSC analog output* and set the TV to "normal" mode, you'll see the full widescreen image squished into 4:3 pillar-boxed frame (tall/skinny people); but if you switch the TV to 'wide' mode, the TV stretches out the signal to take up the full width of the screen and suddenly everything looks correct. Similarly, if you play a 4:3 ('full screen') DVD, you'll get a normal looking picture if the TV is set in 'normal' mode, but short/fat people if you set the TV to 'wide' mode. One could make a fairly decent argument that this business of squeezing widescreen content into an NTSC analog signal and then stretching it back out at the display is nothing but a big hack.

* - this requires that the TV Output Type setting of the DVD player be set to wide/16:9 instead of full/4:3. Ask if you'd like more information on the reasoning behind this.

So, how does this affect the SE-500 and, eventually, James-WBF's situation? The SE-500 just processes NTSC analog video from the inputs to the outputs, it has no idea any of the properties of the video image contained within the signal like whether it is 4:3 or 16:9. How that signal is interpreted by downstream hardware, however (like a video monitor, projector, an analog-to-FireWire converter, a video capture card in a computer, a DVD recorder, etc) makes all the difference in how the analog signal eventually gets displayed (or converted to a digital format).

[KDW4Him]

SE-500 and NTSC analog video, makes sense.

[KDW4Him]

Originally Posted by CharltonChars

How can we improve the video quality of any video, is there any software for this?

I suspect you are speaking of the quality of the video signal as opposed to the technique used to record. What equipment are you using? Is there a link to a sample video?