[David Slater]

If there was a way to down res the pure raw image off the DSLR chip for the HDMI output without the embedded red dot and graphics, I believe the best resolution you could push through the HDMI output (1.5 GB physical limitation) is full raster 1920 x 1080 HD at 4:2:2 (10 bit color?). This would still be excellent quality and beat the native recording codecs of almost every video camera under $50,000 (except maybe the Pansonic 300 which records AVC-Intra at 4:2:2 with 10-bit color, but has only 1/3rd inch chips and is noisy).

Recording stills at 24 fps to Memory Cards would be a way to get even higher resolution, such as 2k, 4k, and maybe even full 18k raw images, but according to Robbie Carman, this is not physically possible with these cameras and the limited speed with which they can transfer data to the Memory Cards (they max out at 8-10 fps for about 10 seconds). This would seem to be an interesting area for future development.

[David Slater]

Thank you for your excellent post. That was very helpful information about the AVCHD codec and that it is acceptable for broadcast. Still, I hope that Syndicate or Magic Lantern is able to hack the firmware to allow the full raster HDMI signal for encoding to a higher quality codec by Convergent Design Nanoflash or AJA Ki Pro.

It is most unfortunate that these large camera companies purposely hold back on what they can offer to budget-minded indepedent filmmakers so that they do not cannibalize sales of their broadcast quality cameras. I thought that Canon was immune from this pressure because it does not sell expensive broadcast quality cameras and therefore should not have to hold back on what it can offer consumers in lower cost categories. But now I see that even Canon may be subject to this pressure, as it does not want to alienate Sony or Panasonic broadcast divisions by bringing out a low cost camera that could compete with their expensive broadcast cameras (If they did, Sony and Panasonic would stop buying Canon lenses for their expensive cameras). It has gotten a little too incestuous and cozy among these camera manufacturers.

The proof will be in the new Canon HD video camera that they will be introducing at NAB. We already knows from their press releases that it will records to a new superior MPEG-2 50 MByte 4:2:2 Codec. This is much to be desired in a low cost HD camera. But will it, as every independent, budget-conscious filmmaker desires, contain a full-sized sensor like those used in their HDSLR cameras? I have seen speculation which says probably not, which will be a huge disappointment. And so we are left with Red’s Scarlet, and its perpetually receding release date. At least Red knows what the consumers want and is responding to their needs. I just hope that Red is not also concerned that the Scarlet will cannibalize sales of its own higher-end camera, the Red One. In which case Red has become just like all the others. Is it so hard for these camera manufacturers to understand what their market craves and produce an under-$10,000 HD video camera with a full-sized DSLR sensor, encoding to a superior 4:2:2 codec, and interchangeable lenses? We shall see…

[David Slater]

Thank you for your prompt response. I’ve read a lot of your posts at Creative Cow and find them very helpful and well informed. Ironically, I attended an Apple Final Cut seminar at B&H in NYC, and the guy mentioned you as an expert on codecs, transcoding, Kona cards, etc.

On the 24 fps v. 30 fps issue, thank you for pointing out that 24 fps is probably better for web delivery and DVD since it allows for either better compression or longer run times. Still, it’s a hard decision when the most important masters will be deliverables for television broadcast. If only there were a clear winner…

As for the issue of 1080 versus 720, I didn’t mean to revive old debates on the different formats. I was just curious which would be better from a workflow format for the HDX 900. I just read in the HDX 900 camera manual that the chip only has 1280(h) x 720(v) effective pixels. I think this means that I don’t gain anything by recording to 1080 in the camera, since this is just blowing up or interpolating pixels to get the larger recorded image. So, I’m thinking I’m better off saving the memory and staying in a 720p workflow. I can always output to 1080i if it is needed for a master.

I’ve got to get my head around the Kono cards, their software, and how they work with HD-SDI. It seems that is the better way to go than Firewire, especially if, as you say, it permits removal of duplicate frames on ingest, which Firewire does not allow. I’ve got to study that a bit more. I guess that outweights the issue of transcoding the pure DVCPro HD Data to HD-SDI, and then transcoding back to DVCPro or Apple Pro Res on ingest. On the other hand, you are saying that if I want use 24pA (advanced pulldown), I am better off going through Firewire, because that removes the 24pA better than HD-SDI/Kona Card. It all seems kind of confusing.

You also seem to really prefer ingesting the DVCPro HD into ProRes rather than keeping it in the native format. It seems like a lot of extra memory to use up for the more robust and memory-hungry ProRes codec. You seem to be saying that it is easier to work with ProRes than native DVCPro HD on the Final Cut timeline.

Sometimes, I wish there was just one magical end-to-end solution that works for most of our productions. Maybe just a pipedream…

Thanks again.