Chroma Key with DVCProHD?

[Onematchfire]

Dear All,

I see. That makes sense! So that leads me to wonder: Is it easier/better to chroma key off betacam or digibeta or off DVCProHD to tape?

Any thoughts/experiences?

Best,

OneMatchFire

[Noah Kadner]

In theory Digibeta would be among the easiest because it is the least compressed relatively speaking- when coming off the camera tape. There would be the requirement of capturing to a third-party capture card, requiring a lot of storage space and an expensive DigiBeta deck rental for ingest. If you wanted to stay in the Panasonic/24p/Firewire capture universe you would go with an SDX900 in DVCPRO50.

Of course you are also losing a lot of definition in the bump down from HD to SD so it would depend on your intended delivery format. There’s also the potential of shooting 16mm or 35mm which could be digitally scanned in post to a very high resolution and bit depth- providing a potentially very clean comp.

I have an upcoming project with a lot of greenscreen and I intend to deliver an HD master. I have been exploring the possibility of shooting on an SDX900 and bumping the final output up to HD as opposed to incurring the extra expense and infrastructure required to shoot uncompressed HD. My experience with compositing SDX900 footage on SD projects thus far has been highly positive.

Noah

ps- are you really named OneMatchFire? 🙂

[Dale Mccready]

I’ve done a lot of keys using DVCproHD through firewire and while the uncompressed (or recompressed) versions are better, I’ve had no trouble with Keylight, Ultimate and DVmattePro. As a DP I would now always expose the image brighter than the normal if using Film Rec, and had great results with low noise putting the green and mid point at 50% luminance (as opposed to the spec’d 30.5%). Green easily better than blue.

If your eventual output from your composite is SD then even compressed I’ve found DVCproHD better than DigiBeta.

[Uli Plank]

I’ve found even downscaled HDV very close to Digi-Beta when it’s lit well.

Regards,

Uli

Author of “DVDs gestalten und produzieren”, a book on professional DVD-authoring in German.

[Gavin Stokes]

There seem to be some misconceptions here. First, all these formats are compressed. DVCPro HD, HDCam, whatever.

Furthermore, transferring via FireWire should not add compression. In fact, it avoids it. If you send the material over SDI, you’re decompressing it from its native format (DVCPro or HDCam). What are you doing with it once it gets to the computer via SDI? Saving it in some uncompressed format, hopefully, otherwise you’ve just unnecessarily recompressed it.

Sending DVCPro 50 or 100 (HD) over SDI doesn’t make any sense, because you could have just captured the original data over FireWire; this avoids decompressing each frame, saves a ton of storage space, requires no special hard drive performance, and is still fully editable. Any frame you don’t change can be saved back to its native format with no recompression.

Unfortunately (and typically), Sony is different. They refuse to provide a software implementation of the HDCam codec, so you ALWAYS have to decompress it (unless you’re using Xpri, I think). There’s no way to get the data off the tape in its native format, so SDI is your only choice. Then when you’re done with your project and want to write it back to tape for archiving, it gets recompressed. Dumb de dumb dumb, Sony.

So that’s compression. Another problem is the color sampling. DV50 and 100 are 4:2:2; for every two pixels, you get only one color. Since you’re doing keying based on color, that creates a blocky, low-resolution key. Sony is apparently 3:1:1, and I’m not sure how that plays out spatially; but you still suffer from the same problem: color information that’s of a lower resolution than the details in the picture.

I’ve wondered whether Beta SP or other analog formats might actually produce a better-looking key by “smoothing” the picture and obscuring the blockiness of the color information. Maybe someone here has done some experiments along those lines. One simple technique you can use is to put a blur on the green channel in your compositing software if you’re doing a greenscreen with a chunky color signal. I’m sure these DV-specific keyers use a combo of techniques to interpolate and smooth the jagged edges of the color signal.

[David Battistella]

Stoke,

I agree with you. the footage is all compressed at the capture stage, but depending on how much manipulation you are doing it’s nice to work in an uncompressed format because it will hold up better with mulitple renders. All of this compression is designed to get an HD signal to tape (which is not at all easy to do without compression) and each manufactureer makes their own compromises/choices. Coming from Discreet, you know all of this.

So when shooting green screen you have many options, if you shoot Varicam you know that it is being compressed going to tape so the next logical step could be to keep it as prestine as possible from that point on. The DVCPRO HD coec does not respond as well as the Uncompressed codec when it comes to multigenerational performance, but it makes a ton of sense for aquisision. I am not saying there is any more information when you capture uncompressed, just that you can preserve what you have already given up when you shoot.

I just did some tests with HDV and I will be posting an article shortly on it. It’s not a chroma Key test but we did find some intersting things and you may be right, it might just be better to stay in the native codec and do everything in that codec with a firewire workflow, because it never gets better than it’s original compressed self.

David

[Gavin Stokes]

Hi David.

I was certainly not recommending that anyone return their footage to DV50 or HDCam as it goes through stages of manipulation. I agree, that must all be done in an uncompressed domain. I meant to emphasize that there’s no need to recompress any frame that isn’t being altered.

I’m very interested to see the results of your HDV testing, since my opinion of this format is very low indeed.

I’m also awaiting info from a Panasonic person who was kind enough to respond to my question about why there’s still no DV50 (or DV100) codec for Windows from Panasonic. This is going to be a big issue when HDV’s nemesis, the HVX200, hits the market.

Regards,
Gavin

[David Battistella]

Hi Gavin,

“I was certainly not recommending that anyone return their footage to DV50 or HDCam as it goes through stages of manipulation”

Not sure where I suggested this. it’s all cool man. HDV is not all that promising. It’s odd. it’s not really HD and not really SD and not that great an in between.

David
🙂

[Gary Adcock]

[stokestack] “Sending DVCPro 50 or 100 (HD) over SDI doesn’t make any sense, because you could have just captured the original data over FireWire; this avoids decompressing each frame, saves a ton of storage space, requires no special hard drive performance, and is still fully editable. Any frame you don’t change can be saved back to its native format with no recompression.”

That is only true for content that is cuts only with no additional processing. Graphics and color correction especially do not like to be compressed in any way and software does not accurately compress the content the same way the chip in the cameras do- so how does some one handle the content?
Remember that most of us still deliver on another format so their is no way around the issue other than working in the Highest Quality until you need to lay back to what ever tape format.

it is the difference between what the Camera can do in hardware and the Computer can try to do in software.

From IBC

Gary Adcock
Studio37
HD and Film Consultation
Chicago, IL

[Gavin Stokes]

Hi Gary, fellow Chicagoan.

“That is only true for content that is cuts only with no additional processing”

Yes, that’s why I said “any frame you don’t change.” And regardless of what you’re doing, there’s still no reason not to ingest the footage in its native form over FireWire. You can then manipulate it and save the result uncompressed. There’s no benefit to decompressing 100 percent of the footage before bringing it into the editing or compositing app. As long as the app takes the format you’re supplying, use it directly and only worry about how to store the result.

Regards,
Gavin