HD or DV???

[Darby Edelen]

[randall murphy] “Would it be a bad call to go to an HD camera if I’m going to be shrinking the image?”

Working with HD footage will be more cumbersome, but the added resolution can improve over all picture quality even after it has been scaled down. If you are considering using an HDV camera, then I would say it’s probably not worth the headache of transcoding from HDV (MPEG-2), working in HD and then down converting. I would guess that it would increase your time spent on the project by at least 10% and possibly more.

SD on the other hand is faster to work with, but results in inferior picture quality (note that HD can look bad too, it has as much to do with how something is shot as what it’s shot on).

[randall murphy] “Would it also create bigger files that would stream or load slower?”

That depends only on the bit rate of your finished file… the source has no impact what-so-ever on it.

Darby Edelen
Lead Designer
Left Coast Digital
Santa Cruz, CA

[Randall Murphy]

Thanks Darby.

Since I’ll probably be compressing it big time, would I still be better off with HD?

Have you seen the video on apple.com on the iphone page. The guy in the black with the black background? It’s amazing the quality and it streams really fast. Any idea how they did it? SD or HD? Compression? Streaming or nonstreaming?

[Tl Westgate]

I think the real question here is, how do you want the final video presented on the web? 4:3 or 16:9? Nowadays, I would go 16:9 if at all possible.

— TL

[Lars Bunch]

Hi,

The Apple video was almost certainly shot in HD, but that doesn’t really matter here.

I’m not sure if Apple is providing the video in different codecs, but mostly likely the one you saw was in the h.264 codec which can be highly compressed and still look pretty good.

But I suspect the reason why it streams so fast is the simple fact that there isn’t a lot of data in the image. If you have a guy dressed in black on a black screen and all you can see is his face plus and edge to define the outline of his body, then there is a lot less information that has to be stored. Everything that is black can be compressed to a very small size. There may have been a conscious choice by Apple and their advertising agency to do this specifically so the videos would be fast to download.

You can see the same sort of thing by creating two Jpeg images, one with a lot of detail; trees, faces, whatever and the other will very little detail; a simple object in a black or white field. Make the pixel dimensions the same and store them with the same quality settings. Most likely the simple image will be a good deal smaller than the complex image.

Many codecs (particularly Mpeg2 and its variants) look at how much change there is between one frame and the next and only store the data needed to construct frame 2 from frame 1. The less movement or change between frames, the less data that needs to be stored. A simple image, without a lot of movement can be compressed quite a lot. So if you are creating a video podcast and you want it to download quickly, you may want to favor a static background rather than something with a lot of movement.

Hope this helps,

Lars

[Randall Murphy]

Thanks Lars, that’s the answer I was looking for. I do get really good results with my DVX-100B and I’d like to for production time and ease sake, keep the equipment I’ve got now. So, during rendering, if nothing changes, i.e. static black background, it helps to shrink the output file size. That’s great. What did you mean, “an edge to define his body” would that be a mask?

The h.264 codec, I hear works on the latest flash player, how do you make the render so it uses this codec?

[Lars Bunch]

Hi,

I don’t really know enough about flash to help you on that issue. Hopefully someone else has a good answer.

When I referred to and edge, I was referring to a style of lighting, basically a rim light, to create a highlight on the person’s shoulders and arms so that you don’t end up with a head floating in space. In the case of the iPhone video, it appears that the guy’s shirt is a dark gray and they aren’t using a rim light to separate him from the background. Instead the key light brings the tone of the shirt up a bit so we can see his head is attached to a body.

Lars

[Randall Murphy]

Oh, okay. I will be shooting against greenscreen with three point lighting with backlight to pull the talent off the background. Then replace the background with absolute black…nice even pixels 🙂 That should cut down on the render size…now just to figure out how they streamed/loaded the video and i’m set. thanks for your help

[Darby Edelen]

[randall murphy] “now just to figure out how they streamed/loaded the video and i’m set.”

I was told by an Apple employee that they use Compressor to compress all of their videos for the web… I’m not sure if I was being lied to though =)

The Fast Start or Fast Start Compressed Header modes of streaming work pretty well. I believe Hinted Streaming only applies if you’re using a QT Streaming Server, which I’m pretty sure Apple does, but you probably won’t.

Darby Edelen
Lead Designer
Left Coast Digital
Santa Cruz, CA